1 CA/CAS 2/2025 (“CAS 2”) and CA/CAS 4/2025 (“CAS 4”) are two related appeals brought by various investors who described themselves as “lunatics” after the Luna token (“LUNA”) which collapsed in May 2022. Before its collapse, LUNA was one of the largest cryptocurrencies on the market. The investors had purchased TerraUSD (“UST”), an “algorithmic stablecoin” and sister cryptocurrency of LUNA. One of the main attractions of UST was its alleged claim “that the UST tokens were stable by design, as they were pegged to a stable fiat currency” which was allegedly supported by a protocol that “could maintain the price stability of UST regardless of market size, volatility or demand through an algorithm that enabled an arbitrage process to happen with UST’s sister token, LUNA, guaranteeing that UST would always return to the 1 USD peg”.

2 Just prior to the trial of the representative action before an international judge (“Judge”) of the Singapore International Commercial Court (“SICC”), the defendants below, now the respondents, conceded that various representations including those highlighted above (“Conceded Representations”) were false (see [15] below). Thereafter, the main issues at the trial largely focussed on issues of reliance and causation.

3 The essence of any claim for fraudulent misrepresentation is premised on the representee’s reliance on the truth of a certain representation which was subsequently discovered to be false. Implicit in every such claim is the discovery of the falsity of the representation by the representee at some point in time leading to the commencement of legal proceedings to recover the reliance damages. It follows that a key factual issue in such a claim would entail the determination of the time when reliance would have ended as that would have a crucial bearing in the assessment and quantification of the reliance damages.

4 It was quite clear that the issue as to when reliance on the Conceded Representations had ended troubled the Judge who heard the representative action. This issue was not squarely addressed by the parties. The Judge raised the issue for the parties’ consideration and invited them to submit on the cut-off date when reliance on the Conceded Representations would have ended. Sometime thereafter, the defendants applied to amend their defence to plead among other things a bare assertion that the claimants had failed to mitigate their damages. On the available evidence, the Judge eventually adopted the cut-off time of 12 May 2022 (“Cut-Off Date”) at 12.01am UTC (“Cut-Off Time”), which the appellants challenge on appeal.

5 Unfortunately, the juxtaposition of the Judge’s discussion of the Cut-Off Time with the amendment application which was allowed by the Judge appears to have inspired the appellants to raise various arguments which are centred on the Judge’s decision to allow the mitigation amendment. Apart from making various findings which are common to and binding on all the represented appellants, the Judge, in assessing the claims of the ten appellants, dismissed three of the claims on the basis that there was no reliance on any of the Conceded Representations. While the Judge allowed the remaining seven claims, he limited the assessment of the damages to the Cut-Off Time because by then, reliance on the Conceded Representations would have ended.

6 It was in this context that one of the principal arguments raised by the appellants in these appeals is grounded on the Judge’s alleged error in allowing the mitigation amendment. The appellants claimed that they were denied the opportunity to adduce further evidence because the issue of mitigation was only raised at a very late stage of the proceedings below. The premise of this argument, which is also relied on in aid of the appellant’s application in CA/SUM 29/2025 (“SUM 29”) to adduce further evidence, assumes that the determination of the Cut-Off Time arose directly from the mitigation amendment. As we explained at the hearing in the course of dismissing SUM 29, this is incorrect.

7 As this court recently held in POP Holdings Pte Ltd v Teo Ban Lim [2025] 2 SLR 90 (“POP Holdings”) at [64], until the time when the deceit or misrepresentation is discovered, there is no question of the claimant mitigating its loss. However, once the misrepresentation is discovered, which is tantamount to the time when the reliance on the misrepresentation would have ended and the claimant is able to sell the property that was acquired under the influence of the misrepresentation, it would be deemed to have done so at that time to mitigate his loss. In this judgment, we will take the opportunity to explain the interplay between the determination of the cut-off time and the duty to mitigate.

8 The facts of this case are set out at [4]–[38] of Beltran, Julian Moreno v Terraform Labs Pte Ltd [2025] SGHC(I) 17 (“Judgment”). We will summarise the facts pertinent to this appeal.

9 The appellants are the claimants in SIC/OA 3/2024 (“OA 3”). That suit involves 366 claimants who sued the three defendants for misrepresentation (fraudulent, negligent, and innocent), as well as for breach of unilateral contract, inducing breach of contract, and unlawful means conspiracy. OA 3 was commenced as a representative action. The trial was bifurcated, with the first tranche dealing with issues common to all claimants, and also disposing of the individual claims of the ten representative claimants in full. The subsequent tranche(s) would address the claims of the remaining claimants, ie, the represented claimants.

10 The ten representative claimants are now the representative appellants, with the exception of the first representative appellant. The first representative appellant was initially the first representative claimant, Julian Moreno Beltran (“Mr Beltran”). However, prior to the hearing of the appeal, the appellants filed applications for Mr Jonathan Kupetz to replace Mr Beltran as the first representative appellant in the relevant appeals. These applications were eventually uncontested and we allowed the same prior to hearing the main appeals.

11 The respondents are the defendants in OA 3. The first defendant, Terraform Labs Pte Ltd (“Terraform”), is a Singapore-incorporated company that developed UST and LUNA. The second defendant, Kwon Do Hyeong (“Mr Kwon”), was the co-founder, majority shareholder, director and CEO of Terraform. The third defendant, Luna Foundation Guard Ltd (“LFG”), is a Singapore-incorporated company of which Mr Kwon was a member and guarantor.

12  UST and LUNA were part of the “Terra Ecosystem”. It was intended that LUNA would help UST maintain its peg to the US dollar (“US$”) by absorbing any volatility in UST’s market value via an arbitrage mechanism. This mechanism was supposed to allow users to always trade US$1 worth of LUNA for one UST, and vice versa. This would incentivise arbitrageurs to “burn” LUNA and “mint” UST when the price of UST exceeded US$1, and to “burn” UST and “mint” LUNA when the price of UST fell below US$1, thereby maintaining the price of UST at US$1 through the operation of market forces of supply and demand.

13 Another aspect of the Terra Ecosystem was the “Anchor Protocol” (“Anchor”), a lending and borrowing platform where users could deposit or “stake” their UST in exchange for promised returns on an annualised yield basis, analogous to placing deposits in a bank to earn interest.

14 The appellants alleged that they had been induced by seven representations to purchase UST, and ultimately suffered losses when the price of UST collapsed. The pleaded representations (“Representations”) were as follows:

(a) “First Representation”: “that the UST tokens were stable by design, as they were pegged to a stable fiat currency”.

(b) “Second Representation”: “that the Terra protocol and its underlying token mechanics would be able to maintain the price stability of UST regardless of market size, volatility or demand through an algorithm that would allow an arbitrage process to happen with UST’s sister token, LUNA, guaranteeing that UST would always return to the 1 USD peg”.

(c) “Third Representation”: “that UST holders would always be able to exchange 1 UST for 1 USD’s worth of LUNA on the Terra protocol”.

The First to Third Representations were alleged to have been made in a white paper authored by Mr Kwon and others, and/or published on Terraform’s website.

(d) “Fourth Representation”: “that Anchor was a principal-guaranteed stablecoin savings product where UST holders could enjoy the stability of holding a stablecoin while earning passive income”.

(e) “Fifth Representation”: “that purchasers of UST would earn up to 20% Annualized Percentage Yield on Anchor if they staked their UST on the protocol”.

The Fourth and Fifth Representations were alleged to have been made in another white paper published on Anchor’s website, and a page on Anchor’s website, respectively.

(f) “Sixth Representation”: “that [LFG] had successfully raised USD 1 billion through the sale of LUNA tokens as a reserve fund to maintain the UST peg to the US Dollar”.

The Sixth Representation was alleged to have been made in a press release by LFG.

(g) “Seventh Representation”: “that [Terraform], and [LFG] were deploying more capital and had a plan to rescue the 1 USD peg that would be employed soon, that UST holders should not panic, and that UST and the Terra Ecosystem would ‘return to form’” [emphasis in original].

The Seventh Representation was alleged to have been made in a series of tweets by Mr Kwon on his official Twitter account from 10 to 11 May 2022.

15 Before the trial, the defendants conceded that all the representations, except the Fifth and the Seventh Representations (“Disputed Representations”), were misrepresentations of fact which had been fraudulently made. For the avoidance of doubt, these are the First to Fourth and the Sixth Representations, and we will refer to them as the Conceded Representations (see [2] above). The defendants did not, however, admit that the appellants were induced by the Conceded Representations or suffered any loss in reliance on them.

16 After the first tranche of the trial, the Judge found that the Disputed Representations did not constitute actionable misrepresentations (Judgment at [79] and [85]). The Judge also dismissed the appellants’ claims for breach of unilateral contract, inducement of breach of contract, and unlawful means conspiracy (Judgment at [238]–[239]). There is no appeal against these findings.

17 The Judge dismissed the claims of three of the representative appellants in their entirety, finding that reliance and causation of loss were not made out for those appellants (Judgment at [122], [136] and [157]–[158]). The Judge found that the remaining seven of the representative appellants had relied on the Conceded Representations and suffered loss, but only awarded them part of the damages they claimed. In summarising the general approach which the court undertook for the assessment of the damages, the Judge explained his reasoning for adopting the Cut-Off Date and Cut-Off Time at [104]–[114] of the Judgment, which we set out in some detail:

18 It is this aspect of the Judge’s decision that the appellants principally take issue with in this appeal.

19 After the conclusion of the trial, the Judge ordered the parties to bear their own costs. The Judge reasoned that neither the appellants nor the defendants were successful overall, considering that the appellants failed in the majority of their claims, that the representative appellants only recovered 5.58% of the amount claimed by those appellants, that three of the representative appellants failed altogether, and that the first defendant succeeded in part in its counterclaim by obtaining two declarations related to the unsustainability of the representative appellants’ claims for breach of unilateral contract.

20 In CAS 2, the appellants appeal against what they characterise as the Judge’s findings on mitigation as set out at [104] and [109]–[114] of the Judgment, as well as the award of damages for four of the representative appellants. They raise the following grounds of appeal:

(a) First, the respondents had not pleaded and particularised the appellants’ failure to mitigate prior to the trial, thereby prejudicing the appellants as they could not adduce evidence to address the issue.

(b) Second, the Judge erred in applying the MitigCut-Off Time to all the appellants, including the represented appellants. The Judge failed to consider when each individual appellant actually discovered the respondents’ fraud, and whether their actions thereafter were reasonable given their subjective circumstances. The Judge also erred in that a “reasonable investor” would not have discovered the respondents’ fraud by the Cut-Off Time, and even if they did, it was not unreasonable for them to have held UST past the Cut-Off Time.

(c) Third, the Judge erred in using the cut-off price of US$0.8011 per UST token (“Cut-Off Price”) because, among other things, there would have been insufficient liquidity to absorb a sale at that price and some appellants would not have been able to sell their UST at all due to platform restrictions.

(d) Fourth, the Judge erred in his quantification of damages for four of the representative appellants, namely Phillip Robert Epstein (“Mr Epstein”), Arun Kumar Sivaramavimalamoorthy (“Mr Arun”), Tan Jian Loon (“Mr Tan”), and Daniel Davis (“Mr Davis”).

21 In support of their appeal, the appellants also filed SUM 29, an application to admit the following pieces of further evidence:

(a) witness statements of Mr Davis, Mr Tan and Khawja Quader (“Mr Quader”, one of the represented appellants) showing that it was impossible for them to have sold their UST tokens by the Cut-Off Time (collectively, “Locked UST Evidence”); and

(b) a witness statement of Mr Epstein containing:

(i) Mr Epstein’s analysis showing that the appellants would not have been able to sell their UST tokens on the Cut-Off Time for the price of US$0.8011, due to insufficient market liquidity (“Liquidity Analysis”);

(ii) evidence of a deleted link in a tweet posted by the second respondent on 11 May 2022 (“Tweet Link Evidence”); and

(iii) evidence of the second respondent’s admissions in his plea of guilt in the US criminal proceedings (“New PG Evidence”).

22 In CAS 4, the appellants appeal against the Judge’s decision on costs, arguing that they were the successful party in the first tranche of the trial below, and that their incurred costs and disbursements were reasonable and proportionate.

23 In response, the respondents argue that:

(a) First, “mitigation” was a term of convenience reflecting the appellants’ own case that they would have sold UST when they stopped relying on the Representations. The Judge had determined the Cut-Off Date based on principles of reliance, causation and quantification of damages. The appellants had notice of and did address these issues. In any event, mitigation was sufficiently pleaded in the context of the case.

(b) Second, the Judge was justified in deciding the Cut-Off Time on a representative basis for all appellants, representative or represented. The Judge was entitled to find, on the objective evidence before him, that none of the appellants could possibly have continued to rely on the Representations in holding UST after the Cut-Off Date and Cut-Off Time.

(c) Third, the Judge’s hypothetical sale price of US$0.8011 was selected on a principled basis and correct.

(d) Fourth, the Judge’s assessment of the claims in respect of the four specified representative appellants was correct.

24 The respondents oppose SUM 29 on the basis that the appellants have not fulfilled the requirements to adduce further evidence on appeal. In particular, all of the further evidence save for the New PG Evidence was available during the trial, and the New PG Evidence is not material. The respondents also challenge the materiality and credibility of several pieces of the further evidence sought to be adduced.

25 Finally, the respondents defend the Judge’s costs order, arguing that there was no successful party in OA 3, and that the costs sought by the appellants in the trial below were disproportionate and/or unreasonably incurred.

26 Accordingly, the following broad issues arose for our determination:

(a) SUM 29: Should the appellants’ application to adduce further evidence be allowed, whether in respect of some or all of the further evidence sought to be adduced?

(b) Did the respondents sufficiently plead or give notice of their case, specifically in relation to what the appellants characterise as the respondents’ argument that the appellants had failed to mitigate their losses?

(c) Did the Judge err by failing to consider the individual circumstances of each appellant in determining the cut-off date and time to be applied in the assessment of damages?

(d) Did the Judge err in adopting US$0.8011 as the Cut-Off Price?

(e) Did the Judge err in assessing the quantum of damages to be awarded to Mr Epstein, Mr Arun, Mr Tan, or Mr Davis?

(f) Did the Judge err in ordering each of the parties to bear their own costs of the trial?

27 At the hearing, we dismissed SUM 29 and reserved judgment in relation to the remaining issues. We now provide our detailed reasons for dismissing SUM 29 and our decision in respect of the other issues.

28 Before doing so, however, we find it apposite to discuss the principles relevant to the assessment of damages for the tort of deceit. This is relevant to both SUM 29 and the main appeals, in which the arguments evince some confusion, particularly relating to the interplay between the duty to mitigate and the determination of a cut-off date when reliance on the misrepresentation ends.

29 The appellants have brought a claim for damages, which is grounded in their reliance on the Conceded Representations to purchase UST.

30 In deciding this claim, the overarching principle is to put the claimant, as far as money can do it, in the same position that it would have been in had the wrong not been committed (Livingstone v The Rawyards Coal Co (1880) 5 App Cas 25 (“Livingston v Rawyards”) at 39). In the case of tortious misrepresentation, that means putting the victim in the position that they would have been in if the misrepresentation had not been made, as opposed to putting them in the position they would have been in if the representation had been true (Wishing Star Ltd v Jurong Town Corp [2008] 2 SLR(R) 909 (“Wishing Star”) at [28], citing Edwin Peel, Treitel on The Law of Contract (Sweet & Maxwell, 12th Ed, 2007) at para 20-018). In the case of fraudulent misrepresentation (ie, deceit), the claimant can recover damages for all loss that flowed directly from their entry into the transaction in reliance on the misrepresentation, regardless of whether or not such loss was foreseeable, and including all consequential loss as well (Wishing Star at [21]; Doyle v Olby (Ironmongers) Ltd [1969] 2 QB 158 at 167B–D).

31 To calculate the quantum of damages, the court must therefore determine the position that the claimant is in as a result of the misrepresentation (“breach position”), and compare that with the position the claimant would have been in if not for the misrepresentation (“non-breach position”). The difference between these two positions constitutes the claimant’s loss (POP Holdings at [28]). Where the claimant was induced by the misrepresentation to purchase an asset, as occurred in this case, it is often appropriate to allow the claimant to recover the full price paid for the asset, less the value of any benefits they have received as a result of the transaction (Smith New Court Securities Ltd v Citibank NA [1997] AC 254 (“Smith New Court”) at 267A–B per Lord Browne-Wilkinson and 284A per Lord Steyn). The assumption underlying this approach is that the claimant would not have entered into the transaction at all if not for the misrepresentation (McGregor on Damages (James Edelman, Jason Varuhas & Andrew Higgins eds) (Sweet & Maxwell, 22nd Ed, 2024) (“McGregor”) at para 50-010, cited with approval in POP Holdings at [38]). It is therefore appropriate to return the claimant to the position they would have originally been in, as if the transaction had never occurred.

32 The appellants’ case is that they would not have bought (and/or held) UST if not for the respondents’ misrepresentations. It is therefore appropriate to assess damages by taking the difference between the price at which the appellants purchased the UST and the value of the benefits they received as a result of the purchase. In adopting this method, it is necessary to fix a date at which to calculate the value of the benefits received (POP Holdings at [44]; Smith New Court at 284A). The Judge correctly identified this as an issue and invited the parties to address him on the appropriate cut-off date for the purpose of assessing the damages. This cut-off date can be influenced by various circumstances. Lord Browne-Wilkinson in Smith New Court at 267C identified two factors in particular:

33 In our view, the starting point in determining the cut-off date is the date when the misrepresentation ceases to be operative. It is implicit in every action for deceit that the misrepresentation must have been discovered by the representee, thus enabling the commencement of legal proceedings. The question is when the fraud was discovered. For as long as the claimant continues to labour under the misimpression created by the defendant’s deceit, the claimant is unable to extricate himself from the consequences of the defendant’s wrong, and the causal influence of the defendant’s wrongful act on the claimant’s retention of the asset persists (POP Holdings at [59]). Conversely, once the fraud is discovered, if the claimant exercises a free choice to hold on to the acquired asset, then the chain of causation is broken and the defendant is not liable for any further losses. Such losses would have been caused by the claimant’s informed decision to hold on to the property and not the misrepresentation, which ceases to be operative. Of course, it may be that after discovering the fraud, the claimant remains locked into the property due to circumstances outside his control. The claimant then cannot be said to have made a free and informed decision to hold on to the property. Any additional losses incurred by the claimant due to, for instance, a further fall in the property’s value, may still be causally attributable to the defendant. But if the claimant seeks to recover damages for such losses, it is incumbent on them to plead and prove the relevant cut-off date that they seek to rely on.

34 The foregoing analysis concerns causation of loss, which the claimant must prove in order to make out its claim for damages in the amount it seeks. It can be confusing to speak of any further losses past the cut-off time as arising from the claimant’s failure to mitigate, even if these concepts are related. As Lord Steyn explained in Smith New Court at 284G:

35 In Smith New Court, the plaintiff, Smith New Court Securities Ltd (“Smith”), bought shares in a company with a view to holding them as a market-making risk and selling them at a later date, relying on certain fraudulent misrepresentations but for which it would not have acquired the shares at all (at 260F–G). The value of the shares dropped upon the subsequent discovery of an unrelated fraud that had been perpetrated on the company, and the plaintiff gradually sold off all the shares at a loss (at 261A–C). In assessing the damages due to the plaintiff, Lord Steyn considered that the causative influence of the fraud persisted from the time of the initial purchase until the eventual sales of the shares, because Smith was locked into the transaction, having been induced to buy the shares as a market-making risk (at 285E–H). This analysis was without reference to the principle of mitigation; indeed Lord Steyn stated that there was “no issue” in relation to mitigation in that case (at 285D). Similarly, Lord Browne-Wilkinson found that the total loss incurred by Smith was caused by the defendant’s fraud, and not from any decision by Smith to retain the shares, because Smith had paid for the acquisition as a market-making risk, and was locked into the shares, having bought them for a purpose and at a price which precluded them from sensibly disposing of the shares (at 267G–268B). It was not alleged or found that Smith had acted unreasonably in retaining the shares for as long as they did or realising them in the manner in which they did (at 268B). Smith New Court illustrates that the court must consider whether the claimed loss is caused by the deceit, regardless of whether it is alleged that there was a failure to mitigate.

36 This distinction between causation and mitigation is also reflected in more recent cases applying the principles in Smith New Court.

(a) In Gestmin SGPS SA v Credit Suisse (UK) Ltd [2013] EWHC 3560 (Comm), the claimant purchased shares in a company on the advice of the second defendant, and suffered losses. In respect of the quantum of damages, the main point in dispute was the date at which the claimant’s loss should be assessed (at [178]). Although the claim was in negligent misrepresentation, it was common ground that the principles relevant to fraudulent misrepresentation, as stated in Smith New Court, were applicable (at [180]). The High Court of England and Wales found that the claimant could not have at any time sold its very large shareholding in the company other than with difficulty and at a significant loss of value. It would therefore not be appropriate to assess the claimant’s loss at any historic date, and loss was assessed at the time of the trial instead (at [190]). There was no reference to mitigation in the exercise of quantifying the damages.

(b) In Mather v Rattan [2025] EWCA Civ 1596, the claimants bought shares in a company in reliance on fraudulent misrepresentations made by the defendant that a personal friend whom the claimants trusted had been appointed as a director of the company. Applying Lord Browne-Wilkinson’s principles in Smith New Court at 267, the Court of Appeal of England and Wales held that the misrepresentation continued to operate, and the claimants were effectively locked into their investment in the company until they were put on alert some years after the purchase of the shares (at [43]). This was a straightforward application of the law of causation, and no issue of mitigation arose.

37 We summarise our discussion above. In claims for damages, a claimant must first establish a causal link between the breach and the loss in question. After the causal link is established, any avoidable loss arising from a claimant’s unreasonable inaction is not recoverable as such a loss is not due to the defendant’s breach. The burden of proving that a claimant failed to mitigate is on the defendant.

38 A claimant’s duty to mitigate starts when the claimant becomes aware of the loss or, in the context of Smith New Court, the fraud. This “timing” of the application of the mitigation principle is what Lord Steyn’s meant when he referred to an “overlap” between the issues of causation and mitigation, or as this court put it recently in POP Holdings, a “linkage” between them. As stated in POP Holdings, the link is simply this: the duty to mitigate by selling the property generally only arises after the deceit is discovered; thereafter the reasonable standard in the form of a claimant taking reasonable steps to mitigate comes into play. In Smith New Court, the court found that the claimants acted reasonably in their handling and eventual sale of the shares (Smith New Court at 268). Absent any allegation of a failure to mitigate, the inquiry into what reasonable steps ought to have been taken does not arise. This inquiry should not be conflated with the element of causation, which must in all cases be proved by the claimant. The upshot is this: a claimant cannot blame the defendant for failing to plead mitigation, in order to escape the consequences of their own failure to adequately prove causation of their loss.

39 Bearing the above principles in mind, we turn to assess the issues that have arisen in this appeal.

40 Before addressing the substantive issues of the appeal, we should first explain why we dismissed the appellants’ application to adduce further evidence in SUM 29. Given the importance that the appellants had placed on SUM 29, we consider it necessary to explain in some detail why the application was dismissed. Besides, the reasons for our dismissal of SUM 29 have a material bearing on the substantive issues in the appeal, in particular the question as to the sufficiency of the respondents’ pleaded defence.

41 The application is governed by O 21 r 23(7) of the Singapore International Commercial Court Rules 2021 (“SICC Rules 2021”), which provides as follows:

42 The three circumstances outlined at O 21 rr 23(7)(a)–23(7)(c) mirror the traditional common law requirements for adducing fresh evidence on appeal or in a new trial, as articulated in Ladd v Marshall [1954] 1 WLR 1489 (“Ladd v Marshall”) at 1491. We will refer to these three requirements as “non-availability”, “materiality” and “credibility” respectively.

43 As explained in AnAn Group (Singapore) Pte Ltd v VTB Bank (Public Joint Stock Co) [2019] 2 SLR 341 (“AnAn”) at [56]–[58], a two-step analysis ought to be considered by this court in dealing with an application to adduce fresh evidence on appeal:

(a) First, the court should consider the nature of the proceedings below and evaluate the extent to which it bears the characteristics of a full trial, or whether it more closely resembles an interlocutory appeal. In appeals against a judgment after trial or a hearing bearing the characteristics of a trial, the interests of finality assume heightened importance, and the court should apply the Ladd v Marshall requirements with its full rigour, subject to the second stage of the analysis (AnAn at [57]).

(b) Second, the court should determine if there are any other reasons for which the Ladd v Marshall requirements should be relaxed in the interests of justice. This broadly includes three (non-exhaustive) categories (AnAn at [58]):

(i) where the new evidence reveals a fraud that has been perpetrated on the trial court;

(ii) where the applicant was prevented from adducing the fresh evidence during the hearing below in circumstances which amount to a denial of natural justice; and

(iii) where the subject matter of the dispute engenders interests of particular importance whether to the litigant or to society at large.

The appellants rely on the second category to justify adducing most of the further evidence contained in their application. We now explain why we dismissed SUM 29 in relation to each piece of further evidence.

44 We first address the Locked UST Evidence and the Liquidity Analysis. We were satisfied that the Locked UST Evidence and the Liquidity Analysis could be rejected on the “non-availability” ground alone.

45 The Locked UST Evidence comprised:

(a) Mr Tan’s evidence that he deposited all the UST he owned into the Gemini Earn platform, and had sought to withdraw these holdings on 11 May 2022, but most of his UST holdings were only released to him on 14 May 2022. Therefore, he could not have sold his UST holdings by the Cut-Off Time.

(b) Mr Davis’ evidence that he had deposited all the UST he owned into a “Locked Product” on the Binance Earn platform, that required him to hold the UST tokens in that produce for a fixed period of 30 days. He had deposited his UST into the “Locked Product” on 12 April 2022, and it was only returned to him on 12 May 2022 at 10.00am UTC. Therefore, he could not have sold his UST holdings by the Cut-Off Time.

(c) Mr Quader’s evidence that he had transferred all the UST he had purchased to his Terra wallet, for the purpose of staking the purchased UST on the Anchor protocol. He was unable to withdraw his UST to KuCoin, a centralised exchange that he used as his only means to buy and sell UST, before the Cut-Off Time due to technical difficulties. Therefore, he could not have sold his UST holdings by the Cut-Off Time.

46 The Liquidity Analysis comprised Mr Epstein’s analysis of publicly available data to determine whether it would have been possible for the majority of the appellants to have sold their UST at the Cut-Off Time for US$0.8011 on centralised and decentralised exchanges. After conducting some simulations, Mr Epstein concluded that the appellants would only have been able to sell their UST at various average sell prices on different centralised or decentralised exchanges, virtually all of which were below US$0.8011.

47 The appellants did not dispute that the Locked UST Evidence and the Liquidity Analysis were in existence during the trial. However, they submitted that they were denied an opportunity to place these pieces of evidence before the trial court because the respondents only raised the issue of mitigation at a very late stage in the proceedings, after all the representative appellants had given evidence and while parties were preparing for their oral closing submissions. The appellants’ argument was therefore premised on their understanding that the further evidence was necessary to respond to the issue of mitigation.

48 The respondents submitted that it was the appellants’ own case that they bought and held UST in reliance on the Representations. Therefore, the appellants themselves put in issue the question of when this reliance ended. The respondents treated the issue of mitigation as merely the “flipside”, “synonymous with”, or a re-characterisation of the issue of when reliance on the Representations had ended.

49 In essence, the appellants relied on the second category of cases outlined at [43(b)] above. In such cases, the lack of strict compliance with the “non-availability” ground does not preclude the admission of fresh evidence, because the applicant was denied a fair opportunity by the trial judge to put forth relevant facts before the court, and their failure to do so was no fault of their own (AnAn at [41]). In such cases, the admission of further evidence may also be allowed on the basis that the requirement of non-availability is in fact fulfilled, based on a purposive interpretation of that requirement. This is because in cases where the party was denied a fair opportunity to advance certain evidence at the hearing, that evidence could not have been adduced at the hearing with reasonable diligence, even if it could have been obtained (AnAn at [43]–[44]). Whichever analysis is adopted, the key question remainsed whether the appellants had a fair opportunity to put forward the Locked UST Evidence and the Liquidity Analysis in the proceedings below, but failed to do so.

50 In our view, the appellants did have a fair opportunity to adduce the Locked UST Evidence and the Liquidity Analysis. However, they did not do so.

51 To begin with, it was for the appellants to prove, as part of their case, the relevant cut-off time at which the value of their UST holdings was to be valued. As explained above at [32], this followed from the method of assessing the damages adopted by the appellants, which involved taking the difference in value between the purchase price of the UST and the price at which the UST was being held or sold. This was true regardless of whether the respondents had pleaded that the appellants had failed to mitigate their losses. The appellants therefore could not claim to have been unaware that the question of the appropriate cut-off date was a live issue at the trial.

52 Moreover, the issue of the cut-off date was expressly brought to the parties’ attention, and the parties were invited to make submissions on the same, even before the respondents raised the issue of mitigation in their pleadings. On the fifth day of trial, ie, 20 May 2025, the Judge invited the parties to address the court on their positions on the quantum of damages in respect of each appellant, and the underlying principles that should guide the court, including the time period of UST transactions that the court should consider in calculating the damages. The parties agreed to do so:

53 The next day, the Judge made clear that he had in mind the specific question of when the representations were no longer operative.

54 Subsequently, on 28 May 2025, the respondents filed applications to amend their defences to include, inter alia, a bare pleading that the appellants had failed to mitigate their losses (“Mitigation Amendment”). At the hearing on 29 May 2025, the Judge agreed with the appellants that it was too late for the respondents to raise the issue of mitigation, other than to characterise the evidence already before the court, in particular the cut-off point at which the appellants ceased to rely on the representations and were simply trading:

55 In accordance with the Judge’s directions, each party submitted different permutations of the dates and times when reliance on the Representations would have ceased. In particular, the first and third respondents, in their Oral Closing Slides dated 29 May 2025, advanced a permutation (“Second Permutation (with mitigation)”) that stipulated 12 May 2022, 12.01am UTC as the cut-off time by which the appellants should have sold their UST. This was derived by starting with 11 May 2022, 10.10am UTC, when the second respondent published a series of tweets stating, inter alia, that UST would be collateralised (“11 May 2022 tweets”), thereby indicating that the earlier representations no longer held true. A 14-hour window was then accorded to the appellants to read the tweets and sell their UST. The first and third respondents also proposed a cut-off price of US$0.60485, derived by taking the average of the opening and closing price of UST on 12 May 2022 (UTC). The Judge ultimately adopted the cut-off time proposed in the Second Permutation (with mitigation) (Judgment at [109]), although the Judge fixed the cut-off price at US$0.8011 instead, being the opening price of UST on 12 May 2022 (Judgment at [114]). In their Further Slides on Damages dated 30 May 2025, the first and third respondents furnished computations of damages for each of the ten representative appellants based on the permutations they provided, including the Second Permutation (with mitigation). In other words, it was clear that the Cut-Off Time applied to all representative appellants. We return to this point below (at [60]). The second respondent, in its Aide Memoire dated 29 May 2025, also advanced a similar permutation in which the making of the Seventh Representation on 11 May 2022 rendered any continuing misrepresentation inoperative.

56 The appellants were given the opportunity to, and did, respond to the permutations advanced by the respondents, in the appellants’ submissions on the first and third respondents’ Further Slides on Damages, dated 5 June 2025. By this time at the very latest, the appellants would have been aware not just of the significance of the cut-off time and cut-off price generally, but also of the specific permutation of the Cut-Off Date and Cut-Off Time of 12 May 2022, 12.01am UTC and the cut-off price of US$0.60485 that was proposed by the first and third respondents. It would then have been open to the appellants to adduce evidence to show that it was not possible for certain appellants to have sold their UST holdings by that time, or to adduce evidence to demonstrate that even if the appellants had attempted to do so, they could only have sold at some price lower than that proposed by the respondents.

57 Indeed, in their responsive submissions, the appellants did deal with the Second Permutation (with mitigation). They argued that the respondents had failed to plead mitigation as a positive defence and failed to cross-examine the appellants’ witnesses on their alleged failure to mitigate, and that the 11 May 2022 tweets contained assurances that UST would repeg, arguments that the appellants are likewise raising on appeal. In addition to those points, the appellants could also have argued that some of them would have been unable to sell their UST by 12 May 2022, 12.01am UTC, or that they could only have sold at significantly depressed prices. They could have sought the leave of the Judge to re-examine the witnesses or to adduce further evidence to substantiate this point, but they chose not to do so, perhaps preferring to rely on the submissions they had already made. They must therefore bear the consequences of that decision.

58 Finally, we address a few other points raised by the appellants that, in the final analysis, we found to be red herrings that distracted from the key question of whether the appellants had a fair opportunity to adduce the Locked UST Evidence and the Liquidity Analysis at the trial below.

59 First, the appellants submitted that the Judge had stated that he would only consider the issue of “mitigation” to decide whether the losses and gains from the appellants’ post-depeg “swing trades” – ie, buying and selling – should be relevant to the damages assessment. They argued that they could not have contemplated that the Judge would make a finding that all appellants who simply retained their existing UST after the Cut-Off Time did so for speculative purposes. The appellants based this on the Judge’s comments that Mr Arun’s case was one where the claimant might have ceased to rely on the representations at some point, after which he was simply trading (see [54] above). The appellants also claimed that this was reflected in the Judgment at [61]:

60 We did not accept that the Judge had confined himself to consider the issue of the cut-off date, which was framed in terms of “mitigation”, solely to the issue of swing trades. We have explained (at [32] above) that it was necessary for the appellants to identify a date at which to assess the value of the benefits they received, for the purpose of calculating the net value of the losses they suffered. There was no principled reason why the cut-off time would only be relevant to cases where swing trading occurred after the cut-off time, and not to all cases, including those where the appellant simply bought and held UST. Indeed, the Judge’s comments during the trial on 20 and 21 May 2022 also made it clear that he was concerned with the question of when the representations ceased to be operative in assessing damages for each individual appellant, and not just appellants such as Mr Arun who engaged in post-depeg swing trades (see [52]–[53] above). Finally, the respondents applied their proposed cut-off time of 12 May 2022, 12.01am UTC to the computation of damages for all representative appellants, not just specific claimants such as Mr Arun (see [55] above). Thus, when the appellants received the first and third respondents’ slides on 30 May 2025, which contained the calculations, it must have been clear by then that the issue of the appropriate cut-off time was relevant to all appellants.

61 Second, the appellants submitted that the respondents failed to put to the appellants’ witnesses that they should have sold their UST by the Cut-Off Time, in breach of the rule in Browne v Dunn (1893) 6 R 67 (“Browne v Dunn”). Consequently, they claimed that they were deprived of the opportunity to explain why it was not possible for them to have sold their UST by the Cut-Off Time.

62 In our view, the appellants have misunderstood and misapplied the rule in Browne v Dunn. The rule in Browne v Dunn is essentially a rule of fairness. Its effect is that where a submission is going to be made about a witness or the evidence given by the witness which is of such a nature and of such importance that it ought fairly to have been put to the witness to give him the opportunity to meet that submission, to counter it, or to explain himself, then if it has not been so put, the party concerned will not be allowed to make that submission (Hong Leong Singapore Finance Ltd v United Overseas Bank Ltd [2007] 1 SLR(R) 292 at [42]).

63 Here, the respondents’ case was that should the court find that there was reliance at all, the appellants’ reliance on the representations ceased by the Cut-Off Time (amongst other possible cut-off times). This was put to the appellants in cross-examination. The appellants do not dispute this. If the appellants wished to submit that it was not possible for them to sell their UST by the cut-off time proposed by the respondents because of restrictions on the exchanges they were using, or that they could not have sold at the proposed cut-off price at that time due to a lack of liquidity in the market, they could have placed the relevant evidence before the court, via their own testimony or otherwise. If the respondents then wished to challenge the veracity of the appellants’ testimony, it would then have been the respondents’ duty to put to the appellants that their evidence should not be believed. The rule in Browne v Dunn does not require parties to pre-emptively rebut arguments that the opposing party might make. The appellants cannot invoke that rule to, in effect, shift the responsibility for their failure to adduce the Locked UST Evidence and the Liquidity Analysis onto the respondents.

64 Third, the appellants placed significant reliance on the respondents’ Mitigation Amendment, which they complained was allowed by the Judge just two days before the end of trial and after all the representative appellants had given evidence. They argued that the amendment should never have been allowed.

65 In our view, the appellants’ focus on the Mitigation Amendment was misplaced. In our judgment, nothing really turned on the decision by the Judge to allow the Mitigation Amendment. Without it, the Judge could still have invited – and did invite – the parties to make submissions on the appropriate cut-off time, because of its inherent relevance to any claim for damages in the context of purchases made in reliance on fraudulent misrepresentation. This inherent relevance was also borne out by the fact that the Plan Administrator in the US bankruptcy proceedings also needed to identify a cut-off date to assess the losses of the investors. Ultimately, the Judge’s decision on the damages was based on his application of the principles of reliance, causation, and quantification of damages to the circumstances of each representative appellant (Judgment at [115]).

66 In view of the above analysis, we rejected the application to adduce the Locked UST Evidence and the Liquidity Analysis for the appeal.

67 The Tweet Link Evidence comprised a string of (since deleted) tweets explaining UST’s peg stabilisation mechanism, that the second respondent’s 11 May 2022 tweets contained a hyperlink to. The appellants sought to adduce the Tweet Link Evidence as context in the interpretation of the second respondent’s 11 May 2022 tweets.

68 The Tweet Link Evidence was similarly rejected on the “non-availability” ground.

69 The appellants submitted that the Tweet Link Evidence could not have been obtained with reasonable diligence at the trial because the respondents relied on and cross-examined the appellants on other 11 May 2022 tweets but not the tweet that contained the link to the Tweet Link Evidence. The appellants claimed that it was only in the Judgment that the Judge decided that a reasonable person reading the entirety of the 11 May 2022 tweets would have appreciated that UST was not stable.

70 In our view, this submission was unmeritorious. To begin with, the fact that the linked tweets were deleted was irrelevant. The appellants acknowledged that the Tweet Link Evidence was in existence during the trial – it was retrieved from the Wayback Machine website for the purpose of SUM 29, and presumably could also have been retrieved from the Wayback Machine website if the appellants had wanted to adduce it at the trial. Neither could the appellants be said to have been deprived of the opportunity to adduce this evidence at the trial. As the respondents submitted, the meaning and context of the 11 May 2022 tweets were vigorously contested during the trial. The fact that the respondents relied on other 11 May 2022 tweets but not the linked tweets did not assist the appellants. If the appellants wished to argue that the linked tweets were material to understanding the context of the 11 May 2022 tweets, they could and should have adduced evidence of the linked tweets at the trial itself.

71 In any event, we also found that the Tweet Link Evidence was immaterial, and would not have had an important influence on the outcome of the case.

72 The appellants submitted that the Tweet Link Evidence was material because it showed the second respondent adopting and endorsing the statements made by the author of the linked tweets, including the fact that he had expected UST to re-peg.

73 In response, the respondents first stated that the author of the linked tweets did not assert (or did not unequivocally assert) that UST would re-peg; and second, that the second respondent did not endorse any predictions of a re-peg in the linked tweets, even if these existed.

74 We agreed with the respondents. The author of the linked tweets presented a model that predicted a “re-peg date of 1.79 days”. However, the author presented this as “a foundation” that was “only a small part of the story”, stating that “[s]entiment, meme, market feeling will all come into play”, and that the final outcome was “[t]ruly anyone’s guess”. The author also included certain caveats in his tweets, such as the following:

Therefore, the appellants were not correct in claiming that the deleted tweet was clear and unequivocal that the author had calculated that the peg would be restored in 1.79 days.

75 Even if the linked tweets contained a clear prediction to that effect, the second respondent did not clearly endorse such a prediction by his reference to the tweets. Rather, the second respondent referred to the linked tweets as a “good overview” for readers who “don’t understand how Terra’s peg stabilization mechanism works”.

76 Because of the above caveats and qualifications attaching to the linked tweets, we did not think that they would have an important influence on the outcome of the case, even if admitted. We therefore declined to admit the Tweet Link Evidence for the appeal.

77 Finally, we turn to the New PG Evidence. The appellants sought to rely on the following admissions that the second respondent was said to have made in the US criminal proceedings:

(a) Between 2018 and 2022, the second respondent had agreed with others to engage in a scheme to defraud purchasers of cryptocurrencies issued by the first respondent;

(b) Following the de-pegging of UST in May 2021, the second respondent had made false and misleading statements about why UST had regained its peg, and had concealed the fact that a trading firm had been involved in restoring the peg; and

(c) As late as November 2022, the second respondent had sought to obscure and hide his fraudulent scheme from investors by distributing a misleading audit report and falsely tweeting that he had used all of the funds from the third respondent to defend the UST peg.

78 It was not disputed that the New PG Evidence was not previously available, and that it was credible. The key question was whether this evidence was relevant to the appeal in CAS 2.

79 The appellants submitted that these admissions revealed that the second respondent was taking active steps to conceal his fraud from investors up until November 2022, such that it was not reasonable to expect the appellants to have discovered the respondents’ fraud based on the second respondent’s 11 May 2022 tweets.

80 The respondents submitted that the New PG Evidence was immaterial. The Judge had found that the 11 May 2022 tweets could not be understood to mean that UST’s peg was sustainable or could be restored. On the contrary, the Judge had found that a reasonable reader of those tweets would have appreciated that UST’s value was not stable and that its peg to the US$ could no longer be guaranteed, and would thus have ceased to rely on the Representations. These findings would not have been impacted by the New PG Evidence.

81 In our view, the appellants had not shown how the New PG Evidence would have had an important influence on the result of the case. We address each of the second respondent’s admissions in turn.

82 The second respondent’s admission that he engaged in a fraudulent scheme from 2018 to 2022 was too general, and did not address the issue of whether the 11 May 2022 tweets specifically would have conveyed to the appellants that the UST peg could not and would not be restored.

83 As for the second respondent’s admission that he made false and misleading statements as to why UST regained its peg in May 2021, this fact if true might have induced the appellants to make purchases of UST. The second respondent also admitted this in the US criminal proceedings. However, it was unclear how this would affect the Judge’s finding that subsequently, upon reading the 11 May 2022 tweets, and armed with the new information it contained, the appellants would have realised that UST was not stable and might not re-peg.

84 Finally, the second respondent’s admission that in November 2022, he had distributed a misleading audit report and made false tweets in connection with that report, concerned events too late to be material. This fact, even if true, did not affect the Judge’s finding that on or around 11 May 2022, the appellants should have realised that the UST peg would not be restored.

85 We therefore also declined to admit the New PG Evidence for the appeal.

86 For the above reasons, we dismissed SUM 29 with costs. We turn now to consider the issues in the substantive appeals.

87 The appellants essentially repeat the arguments in support of SUM 29. In short, the appellants submit that the respondents failed to adequately plead and particularise the appellants’ failure to mitigate, thereby denying the appellants the opportunity to adduce evidence to rebut that allegation.

88 Specifically, the appellants submit that they were deprived of the opportunity to lead evidence on the following arguments:

89 The appellants submit that the Judge should have neither allowed the respondents to amend their defences to raise the issue of mitigation, nor permitted any argument on mitigation.

90 The appellants also submit that the respondents never put to the representative appellants that they should have sold their UST to mitigate their losses, in breach of the rule in Browne v Dunn. As a result, the representative appellants were not given the opportunity to rebut the claim that they had failed to mitigate their loss, or to explain their conduct.

91 The respondents submit that the Judge’s reference to “mitigation” was but a term of convenience to characterise the inquiry into when the appellants’ reliance on the Representations started and ended. This was really an inquiry into causation, not mitigation. It was part of the appellants’ own pleaded case. The representative appellants were thoroughly cross-examined as to when their reliance on the Representations ended.

92 The first and third respondents also submit that even if the principle of mitigation is considered, it was adequately pleaded because the only way in which loss could have been mitigated in this case was to sell UST. And if mitigation was adequately pleaded, the appellants cannot complain that the Mitigation Amendment was late, because the respondents were entitled to raise it as a consequential amendment to the appellants’ own changes in their case regarding the quantum of damages.

93 From our discussion of the principles applicable to the assessment of damages for deceit claims (at [30]–[38] above), and our reasons for dismissing SUM 29 (at [50]–[66] above), it is clear that the Judge’s decision was not premised on the appellants’ failure to mitigate their loss. That being the case, whether the respondents had sufficiently pleaded that the appellants failed to mitigate their loss ultimately has no bearing on the merits of the appeal.

94 We accept the appellants’ general contention that the burden of proving that loss has not been mitigated lies on the party in breach (Jia Min Building Construction Pte Ltd v Ann Lee Pte Ltd [2004] 3 SLR(R) 288 (“Jia Min”) at [71], citing Geest plc v Lansiquot [2002] 1 WLR 3111 at [16]). However, that is irrelevant in this case. Regardless of whether the issue of mitigation arises, the claimant must discharge its burden of proving the actual amount of damage they claim to have suffered (Jia Min at [72]). It is a general requirement in every claim for damages that the claimant must prove its loss (Wishing Star at [21]). As we have explained above at [32], for the purposes of calculating the damages due to the appellants, it is necessary to select a cut-off date on which to value the benefits received by them. It was therefore incumbent on the appellants to identify the relevant cut-off date they were relying on for the purposes of assessing their damages. This was true regardless of whether the respondents alleged any failure to mitigate on the appellants’ part.

95 It is, of course, open to a defendant to propose alternative cut-off dates by which time it claims reliance on the misrepresentations must have ended, in addition or in the alternative to arguing that there was no reliance at all. This is precisely what the respondents did in this case. This does not mean that the defendant must disprove every possible obstacle to the claimant selling their assets. It would not be feasible or practical for the defendant to do so. Rather, the defendant can lead evidence to show that the misrepresentation ceased to be operative at an earlier date than that claimed by the claimant. The respondents in this case did so by relying on the 11 May 2022 tweets.

96 If the claimant takes the view that the cut-off date must, for some reason, be later than what the defendant suggests, it falls on them to identify that reason and adduce the necessary evidence of the facts they wish the court to believe (Evidence Act 1893 (2020 Rev Ed) (“EA”) s 105). This is particularly so where the claimant relies on facts that are especially within their knowledge, and which are therefore their burden to prove (EA s 108). In this case, the representative appellants could and should have raised before the Judge their argument that they could not have sold their UST by the Cut-Off Time of 12 May 2022, 12.01am UTC, or at least not at the prevailing market prices, once the respondents had proposed the Second Permutation (with mitigation). The representative appellants also could and should have sought to adduce the Locked UST Evidence and the Liquidity Analysis at the trial below to substantiate that argument. It was their decision not to do so, and they cannot complain on appeal that the respondents had failed to sufficiently plead or give notice of their case. It is too late for them to do so given the court’s dismissal of SUM 29. To be clear, for reasons we explain below at [112]–[126], it remains open to the represented appellants to adduce the necessary evidence at the next tranche of the trial to show that, based on their individual circumstances, the cut-off time applicable to them should differ from the Cut-Off Time determined by the Judge.

97 The appellants submit that the Judge erred in deciding the Cut-Off Date and Cut-Off Time on a representative instead of individual basis and in failing to consider each appellant’s individual circumstances including:

98  The appellants submit that the Fraud Discovery Issue turns on whether each representee had actual knowledge of the fraud. In so far as the issue is framed as one of reliance, the appellants similarly submit that for fraudulent misrepresentation, the test is actual – not reasonable – reliance. The appellants submit that the Judge wrongly assumed that all appellants had read the 11 May 2022 tweets within the 14-hour grace period, and that all appellants would have discovered the fraud through those tweets. The appellants highlight that as pleaded in their statement of claim, 12 of the represented appellants only read the 11 May 2022 tweets after the Cut-Off Time.

99 In relation to the Reasonable Steps Issue, the appellants submit that the standard of reasonableness to be applied when determining if a claimant had mitigated its loss is not purely objective, but takes into account the subjective circumstances of the aggrieved claimant. The appellants submit that (a) at least two representative appellants were “locked into” UST and it is therefore unjust to apply the Cut-Off Time to them; and (b) the second respondent promised that there would be a recovery plan to restore the US$ peg, when there was none, and thus it was reasonable for the appellants not to have sold UST by the Cut-Off Time.

100 The respondents submit that while the issues of reliance and/or mitigation are typically individualised inquiries, the circumstances of the case allowed the Judge to rule that the Cut-Off Time should be applied on a representative basis, because it involved the same legal and factual inquiry. The first and third respondents submit that if the issue is looked at through the lens of mitigation, it was even more appropriate for the Judge to have decided it on a representative basis as the requisite standard of reasonableness in the context of mitigation is an objective one, taking into account subjective circumstances where relevant.

101 The first and third respondents characterise the Judge’s approach as a nuanced one. The question of initial reliance should be determined individually, as it involved disjointed statements made in different sources, over a span of years, to the world at large, and not to each appellant specifically. As to when reliance ended, however, the respondents submit that the Judge rightly found that based on the objective evidence – comprising UST’s failure to re-peg from 7 to 11 May 2022, and the 11 May 2022 tweets – it could not be believed that any appellant subjectively continued to rely on the Conceded Representations past the Cut-Off Time.

102 The first and third respondents also submit that it is too late for the appellants to plead and adduce evidence that they had held UST for some reason other than the Representations, and that they should have done this at the trial.

103 Order 10 r 19 of the SICC Rules 2021 provides that:

104 Order 10 r 19 of the SICC Rules 2021 is in pari materia with O 4 r 6 of the Rules of Court 2021 (“ROC 2021”), save for certain changes that adapt the provisions to the SICC, which are not relevant for present purposes.

105 In Koh Chong Chiah v Treasure Resort Pte Ltd [2013] 4 SLR 1204 (“Treasure Resort”), this court discussed the history of and principles applicable to representative actions under what was previously O 15 r 12(1) of the Rules of Court (Cap 322, R 5, 2006 Rev Ed). That provision contained certain differences from the current provisions that are also not germane to the present appeal.

106 In determining whether or not an issue can appropriately be decided on a representative basis, all parties in this appeal rightly accepted that the applicable principles are as outlined in Treasure Resort. Whether an issue is common between claimants in a representative action is largely fact-dependent and will have to be determined on a case-by-case basis. However, as a general guideline, where the legal and factual inquiry required for the determination of an issue in a claim in a representative action is also relevant to the determination of the same issue in other claims in the representative action, it is highly probable that the issue is common to all the claimants (Treasure Resort at [59]).

107 In Treasure Resort, the defendant argued that the issue of reliance on certain alleged misrepresentations was fact-sensitive and therefore had to be individually proved by each claimant. However, this court held that prima facie, nothing on the facts suggested that the legal and factual inquiry in relation to the question of reliance would differ materially as between the claimants. This was because all the claimants were proceeding on the basis that their reliance on the representations was evidenced by the same act of their paying certain monthly subscription fees to the defendant. This court therefore found that the reliance issue was common between all the claimants (at [119]).

108 The first and third respondents rely on Jameson v Professional Investment Services Pty Ltd (2009) 253 ALR 515 (“Jameson”) to support their submission that an issue which may typically require individualised evidence can still be determined on a representative basis, if the court is satisfied that it is fair and logical to do so. Jameson involved representative proceedings brought on behalf of a group of investors that had acquired promissory notes in a company pursuant to recommendations made by representatives of the defendant. The company later went into liquidation. The Supreme Court of New South Wales held that the proceedings should proceed on a representative basis because there were substantial common issues. One such issue concerned the content of a product disclosure statement, which was required by statute to contain certain information about the risks associated with holding the financial product in question. These were “objectively stated requirements” which “[did] not depend on the particular circumstances of each plaintiff” and clearly raised a “substantial common question of fact” (at [82]). The court observed (at [85]):

109 The analysis in Jameson is instructive. The determination of issues such as causation and damages would generally depend on individualised evidence. However, certain factual findings may well be common to all claimants, such as the contents of the product disclosure statement in Jameson, or the contents of the 11 May 2022 tweets in the present case. These would be suitable for determination on a representative basis. Once decided, these issues may significantly narrow the dispute on reliance and causation. But to the extent that the effect of the statements differs for each claimant, an individualised inquiry remains necessary. That inquiry would, in the present case, encompass issues such as when each claimant read the statements, and what actions they took or could reasonably have taken as a result.

110 The UK Supreme Court has in the more recent case of Lloyd v Google LLC [2022] AC 1217 at [80]–[81] expressed a similar view:

111 In sum, claims for damages will by their nature often require an individualised assessment. But in so far as there are common issues of law or fact pertinent to the assessment of damages, these may be decided first on a representative basis, leaving the individualised assessments to a second stage, resulting in the bifurcated approach that the Judge in this case appropriately adopted.

112 In our view, the cut-off date and time may not be entirely determinable on a representative basis in the present case. However, the Judgment appears to leave open the possibility that the cut-off date and time for individual appellants may differ based on their individual circumstances, and to that extent, the Judge did not err. We explain.

113 The appropriate cut-off date for the valuation of property purchased by a representee may, in the appropriate case, depend on when the misrepresentation ceases to be relied upon by the claimant, or whether the claimant was locked into the property (see [32] above). These are issues that will turn on the individual circumstances of each claimant. Reliance on a representation is only established if the statement is not merely intended to induce but does in fact act upon the will of the representee such that it influences or leads the representee to change their behaviour (see Wee Chiaw Sek Anna v Ng Li-Ann Genevieve [2013] 3 SLR 801 (“Anna Wee”) at [43]–[44]). The court is concerned with whether the representee actually relied on the statement, not whether it was reasonable for them to have relied on it (John Cartwright, Misrepresentation, Mistake and Non-Disclosure (Sweet & Maxwell, 5th Ed, 2019) at para 3-53; Panatron Pte Ltd v Lee Cheow Lee [2001] 2 SLR(R) 435 (“Panatron”) at [24]). The question of when reliance ends similarly involves a subjective assessment of when the representee became aware of the fraud (see Smith New Court at 266G–H). Of course, based on the factual matrix in a particular case, the question of reliance may turn on a similar legal and factual inquiry among multiple claimants, as was the case in Treasure Resort (see [107] above).

114 As for the question of whether the claimant was locked into the property, this turns on whether they were unable to sell despite reasonable efforts to do so (see [38] above). As with the duty of mitigation in general, the standard of reasonableness to be considered is an objective one, but takes into account the subjective circumstances of the aggrieved party (The “Asia Star” [2010] 2 SLR 1154 (“The “Asia Star””) at [31]). Whether a claimant has acted reasonably in the circumstances is ultimately a question of fact, not a question of law (McGregor at para 10-016).

115 In the present case, we accept that the question of the appropriate cut-off time for the purpose of assessing damages may depend on some individualised evidence. In particular, we agree with the appellants that the appropriate cut-off time could be affected by when each appellant read the 11 May 2022 tweets, and whether it was possible for each appellant to have sold their UST holdings by the Cut-Off Time. It is logical that, if one of the appellants had not read the 11 May 2022 tweets by the Cut-Off Time, they would not have discovered the fraud and thus could not have acted on those tweets to sell by that time. Likewise, if an appellant had taken all reasonable steps to sell upon reading the 11 May 2022 tweets, but was unable to do so, they would have been locked into the property and any further losses would continue to be attributed to the respondents, not the appellant (see [33] above).

116 In so far as there were issues for which the legal and factual inquiry were similar for all appellants, these would have been suitable for determination on a representative basis. Thus, in our judgment, the Judge was entitled to form a view as to how a reasonable reader would have understood the 11 May 2022 tweets (Judgment at [109]). This is clearly relevant to whether each of the appellants became aware of the fraud, once they read the tweets. Prima facie, the meaning conveyed by a representation is the natural and ordinary meaning which would be conveyed to a normal person (K R Handley, Spencer Bower & Handley: Actionable Misrepresentation (LexisNexis, 5th Ed, 2014) (“Spencer Bower”) at para 4.13). If a representation is ambiguous and capable of being interpreted in more than one sense, it is for the claimant to plead and prove the particular meaning they are relying on (Spencer Bower at para 4.14; Anna Wee at [48]). It is of course not sufficient for the representee to merely assert that they understood the words used by the representor in a particular way, and whether or not the court finds that the representee did in fact understand the words in that particular way would depend on all the evidence available (Anna Wee at [49]). These are the principles relevant to the question of whether a representation induced a representee to act in reliance on it (William Smith v David Chadwick (1884) 9 App Cas 187 at 199–200; Spencer Bower at para 4.12, citing Krakowski v Eurolynx Properties Ltd (1995) 183 CLR 563 at 576–577) In our view, these principles are equally applicable to the question of whether the 11 May 2022 tweets would have conveyed to the appellants that the Conceded Representations were false, thereby ending their reliance on them.

117 Likewise, the Judge was entitled to form a view as to the cut-off time when a reasonable investor acting prudently to mitigate his loss would have liquidated their UST holdings, on the basis of the available evidence that was relevant to all the appellants’ claims (Judgment at [111]). This would be the applicable cut-off date in the absence of any countervailing evidence, such as evidence that an individual appellant could not have sold by the Cut-Off Time because they were locked into their UST holdings.

118 For completeness, in so far as it might be argued that the appellants should have read the 11 May 2022 tweets and then sold their UST holdings by the Cut-Off Time, even if they did not in fact read the 11 May 2022 tweets in time, we reject this argument. It is well-established that a claimant can establish reliance if they were in fact induced to act by a fraudulent misrepresentation, even if they had acted incautiously and failed to exercise reasonable diligence or take prudent steps to verify the truth of the representation (Panatron at [24]). In our view, the appellants would still have been relying on the Conceded Representations, even if they had failed to take prudent steps to be informed of the 11 May 2022 tweets, so as to be in a position to act on those tweets.

119 It follows from the above analysis that, while the Judge may have made certain findings relevant to all appellants, the question of the appropriate cut-off date has not been finally decided on a representative basis, and it remains open to each represented appellant to adduce individualised evidence pertaining to their specific circumstances. In our view, however, this does not mean that the Judge erred, as the appellants suggest. On the contrary, the Judge recognised that the findings made in the first tranche of the trial would only bind the represented appellants to the extent that the factual inquiry was similar to that for the representative appellants. The Judge described the court’s approach in bifurcating the trial as follows (at [53]):

120 In summarising the court’s approach to the questions of reliance and causation, the Judge stated at [98(d)]:

121 We would add that external circumstances may support the inference that an investor only ceased to believe in the Conceded Representations after the Cut-Off Date, or that even after realising the truth, the investor was not able to liquidate their UST holdings for some time.

122 Finally, after summarising the court’s general approach to assessing the quantum of damages, and accepting the Cut-Off Time of 12 May 2022, 12.01am UTC, the Judge stated at [115]:

123 We therefore do not accept the respondents’ submission that the Judge had, based on the objective evidence before him, ruled that the court could not envision any scenario where any of the appellants, representative or represented, would have continued to rely on the Conceded Representations past the Cut-Off Time. The Judge described the cut-off date in the following terms (at [104]):

124 It follows from the Judge’s reasoning that, if an appellant was unaware of the falsity of the misrepresentations and continued to act under their inducement, or tried but was unable to sell UST, any further loss or gain could not be described as “purely the outcome of [the] claimant’s personal decision to undertake a risk”. If evidence of an appellant’s lack of knowledge or inability to sell was placed before the Judge and accepted, it seems clear that the Judge would have adjusted the cut-off date applicable to that appellant accordingly. None of the representative appellants placed such evidence before the Judge at the trial below, and it is too late for them to do so now (see [57] above). The Judge’s findings on the representative appellants’ individual claims remain binding on them. But that does not preclude the represented claimants from adducing the relevant evidence at the appropriate juncture when their claims arise for consideration in the next tranche of the trial, which will then be assessed by the Judge together with all the other available evidence.

125 Our view, that the Judge’s finding of the Cut-Off Date and Cut-Off Time is not absolutely binding on the represented claimants without the possibility of adjustment based on individual circumstances, coheres with the parties’ own agreed list of issues. That list makes clear that in respect of the claims for fraudulent misrepresentation, negligent misstatement and innocent misrepresentation, the issues of inducement, reliance, causation and quantum of damages were to be determined on an individualised basis, with determinations in the first tranche of the trial binding on the representative appellants only.

126 Therefore, in our view, the Judge’s finding as to the Cut-Off Date and Cut-Off Time does not necessarily bind the represented claimants, although it binds the representative appellants, and the Judge did not err in his approach to determining the Cut-Off Date and Cut-Off Time.

127 We turn to address whether, even if the Judge was entitled to form a view as to how a reasonable investor would have understood the 11 May 2022 tweets and acted thereafter, the Judge erred in determining on the facts of the case that a reasonable investor would have discovered the fraud and liquidated their UST holdings by the Cut-Off Time.

128 The appellants submit that, in any event, a “reasonable investor” would not have discovered the respondents’ fraud by the Cut-Off Time, given that:

(a) UST returned to the US$1 peg on three consecutive days from 7 to 9 May 2022;

(b) UST had previously experienced fluctuations over four days during the May 2021 de-peg before the peg was restored, which the second respondent represented was evidence of the stability of the peg; and

(c) the context of the 11 May 2022 tweets would have led a reasonable investor to believe that UST would return to the peg, instead of appreciating that the Conceded Representations were false.

129 In response to the appellants’ argument that UST briefly re-pegged over 7 to 9 May 2022, the respondents highlight the lowest price and closing price of UST each day from 7 to 12 May 2022, arguing that this showed that UST was not a functioning stablecoin.

130 The respondents also highlight that the May 2021 de-peg saw UST fall to a lowest point of only US$0.92, which is not comparable to the May 2022 de-peg.

131 In our view, the Judge did not err in finding that, on the available evidence, a reasonable investor considering the entirety of the 11 May 2022 tweets would have appreciated that UST’s value was not stable.

132 First, the evidence of re-pegging that the appellants present is selective. It focuses on the high point of UST from 7 to 9 May 2022, the three earliest days of the de-pegging episode. This ignores the low points and closing values of UST, which show a significant deviation from the 1-to-1 peg, and would have put investors on alert.

133 Second, we agree with the respondents that these deviations were significant and much larger than during the May 2021 de-peg. In May 2021, the de-pegging episode lasted about four days, with UST dropping to a lowest point of about US$0.92. By contrast, UST had hit daily lows significantly below that on three consecutive days from 9 to 11 May 2022, reaching a lowest point of US$0.2998 on the day of the 11 May 2022 tweets. The difference in the fall in UST’s value was stark. Thus, investors should have realised the situations were not comparable, even if the second respondent had tried to suggest that the May 2022 de-peg was “similar to last year”.

134 Third, we agree with the respondents that the 11 May 2022 tweets cast significant doubt on whether UST would re-peg. It is true that the 11 May 2022 tweets contained references – arguably, overly optimistic references – to UST re-pegging, eg:

135 However, the question is whether such hopeful phrases were believable. The Judge (at [110] of the Judgment) rightly noted that the tweets indicated the hopelessness of the UST peg, since they highlighted that UST was trading at a significant deviation from the peg, that the UST-LUNA arbitrage mechanism was having difficulty restoring UST’s value, that any remedy to the “supply overhang” of UST would require solutions at high cost to UST and LUNA holders, and that UST would need to be collateralised. Thus, in our view, the Judge’s finding was not plainly wrong or against the weight of the evidence. Indeed, as counsel for the second respondent, Mr Keith Han, highlighted at the hearing, the Judge had found that the Seventh Representation, which arises from the 11 May 2022 tweets, was not actionable – a finding that the appellants have not appealed against. Their challenge to the Judge’s findings regarding how these tweets would have been understood does not sit well with that position.

136 The appellants submit that (a) there was no available market able to purchase all of the appellants’ (roughly 64m) UST tokens on 12 May 2022 at 12.01am UTC; and (b) even if there was an available market, not all of the appellants would have been able to obtain the price of US$0.8011 for their tokens.

137 The appellants also submit that there is no principled basis for adopting the specific price of US$0.8011, when any hypothetical “mitigation sale” could have occurred at any point within the 14-hour grace period.

138 The first and third respondents submit that the Judge had held that the appellants would liquidate their UST progressively and by the Cut-Off Time, not exactly on the Cut-Off Time.

139 The first and third respondents also submit that the only evidence supporting the appellants’ claim that they would not have been able to obtain a price of US$0.8011 is the Liquidity Analysis, which is flawed. In any event, given our decision to dismiss SUM 29, the Liquidity Analysis is not before this court and thus need not be considered.

140 In our respectful view, the Judge did err in adopting US$0.8011 as the Cut-Off Price, on the basis of the evidence that was before the court in the first tranche of the trial.

141 To begin with, the Cut-Off Price may not realistically reflect the price that an appellant, taking reasonable steps to sell by or at the Cut-Off Time, would have been able to obtain. We accept that there may not have been a ready market for the purchase of all of the appellants’ UST tokens that remained unsold as at 12 May 2022 at 12.01am UTC. The reasoning behind the selection of the Cut-Off Price is that a reasonable investor would have liquidated their UST holdings by or at the Cut-Off Time, but if all investors had tried to act reasonably and did so, the market for UST would have collapsed. This differs from the archetypal situation where an individual claimant should mitigate the losses from their transaction with a single counterparty by entering the market: the entry of one such claimant is generally assumed, as a starting point, not to affect the market price. Even if it is assumed that all the remaining UST tokens held by the appellants at the Cut-Off Time could have been sold, it does not seem realistic that all these tokens would have been sold at the exact time at the exact price of US$0.8011. It is reasonable to think that a significant influx of UST tokens would have depressed the prevailing market price, taken some time to be fully sold, or both. This was a consideration recognised (albeit obiter) by the High Court in Oei Hong Leong v Chew Hua Seng [2020] SGHC 39 at [78], which the appellants rely on. Even if the UST tokens had been sold progressively over a 14-hour period leading up to 12 May 2022 at 12.01am UTC, the increase in supply of UST tokens may still have resulted in depressed prices.

142 We also note that there were significant fluctuations in the price of UST during this period. On 11 May 2022, the price of UST varied between a high of US$0.847 and a low of US$0.2998. On 12 May 2022, the price of UST varied between a high of US$0.829 and a low of US$0.3626. It is unclear if the price of US$0.8011 at 12 May 2022 at 12.01am UTC was reflective of the average level of prices during this period, or if the Cut-Off Price may have varied significantly if the Cut-Off Time had been fixed even just a few hours earlier or later. In other words, the volatility in UST prices may have rendered the selection of the Cut-Off Price based on the market price at the precise point of the Cut-Off Time somewhat arbitrary or unrealistic.

143 Furthermore, we recognise that the appellants did not have an opportunity to address the Judge on whether they could realistically have sold their UST tokens at US$0.8011, since the respondents’ case was based on the average price of US$0.60485. This latter price was calculated as the average of the opening and closing price of UST on 12 May 2022 (UTC).

144 The question then is what the appropriate cut-off price should be. At the hearing before us, counsel for the appellants, Mr Mahesh Rai (“Mr Rai”), urged that we remit the matter for the cut-off price to be individually determined because the appellants were trading on different exchanges. Alternatively, Mr Rai submitted that we should adopt the volume-weighted average price or the lowest price of UST on 11 May 2022, which was US$0.2998. The appellants emphasised that the court adopts a generous approach in assessing an aggrieved party’s conduct in mitigation (The “Asia Star” at [43]).

145 The drawback with the appellants’ preferred approach of remittal is that they would effectively be given a second bite of the cherry, having failed to adduce the relevant evidence in the first tranche of the trial. This cuts against the very considerations of finality and fairness which undergird the Ladd v Marshall rule (AnAn at [23]–[26]). “[I]n most cases it will be unfair to a litigant to subject him to a retrial … because his opponent culpably failed to put all the best relevant evidence before the court at the first trial” (AnAn at [24], citing Saluja v Gill [2022] EWHC 1435 (Ch) at [24]). Likewise, adopting the volume-weighted average price in this appeal would only be possible with the admission of Mr Epstein’s Liquidity Analysis, which we have rejected for similar reasons as explained at [50]–[65] above. Finally, there is simply no principled basis for the adoption of the lowest price of UST on 11 May 2022, namely US$0.2998, which would be stretching the bounds of the court’s generosity even if a generous approach is to be adopted.

146 In our view, the more appropriate approach would be to adopt the price of US$0.60485 that was proposed by the first and third respondents at the trial below. The law does not demand that a claimant prove with complete certainty the exact amount of damage they have suffered (Robertson Quay Investment Pte Ltd v Steen Consultants Pte Ltd [2008] 2 SLR(R) 623 (“Robertson Quay”) at [28]). It is unfortunate that evidence concerning the cut-off price that could have been furnished was not adduced and tested in the trial below. Where precise evidence is obtainable, the court naturally expects to have it. However, where it is not, the court will do the best it can (Robertson Quay at [30], citing Biggin & Co Ld v Permanite, Ld [1951] 1 KB 422 at 438). We do not think that this is a situation where no evidence has been given as to the amount of damage such that it is virtually impossible to assess damages (see Robertson Quay at [27]; POP Holdings at [47]–[56]). Rather, in assessing the claims of the representative appellants, this court must rely as best it can on the evidence adduced in the first tranche of the trial.

147 We bear in mind Lord Blackburn’s classic dictum in Livingston v Rawyards at 39 that when damage is done maliciously or with full knowledge that the person doing it was doing wrong “you would say that everything would be taken into view that would go most against the wilful wrongdoer”. This is not a license to disregard all principles in cases of fraud, but a salutary reminder that especially in such cases, a mechanical approach towards assessing damages is to be eschewed in favour of flexibility (Vita Health Laboratories Pte Ltd v Pang Seng Meng [2004] 4 SLR(R) 162 at [93]).

148 In the present case, we are of the view that the cut-off price of US$0.60485, representing the average of the opening and closing price of UST on 12 May 2022, is a fair and an appropriate cut-off price to adopt on the best available evidence, which primarily comprises the data for the opening, closing, highest, and lowest daily prices of UST. This is in the absence of any expert evidence on the likely prices the appellants would have obtained had they tried to sell at the Cut-Off Time. This price better accounts for the possibility that the appellants would in all likelihood have obtained, on average, selling prices lower than US$0.8011 had they all attempted to sell their remaining UST at or around the Cut-Off Time. We note that this price is also close to the average of the highest price (US$0.829) and lowest price (US$0.3626) on 12 May 2022, namely US$0.5958. Significantly, the cut-off price of US$0.60485 was included in the first and third respondents’ Oral Closing Slides, and the appellants did not raise any of the objections regarding the selling price that they now seek to raise on appeal.

149 On the basis that the cut-off price is revised to US$0.60485, the amount of damages to be awarded to each of the representative appellants who were awarded damages is revised as follows:

(a) Third representative appellant, Mr Davis: The damages to be awarded are the total value of Mr Davis’ UST purchases (US$29,990.48), less the value of his holdings of 29,968.00 UST at the Cut-Off Time, using the cut-off price of US$0.60485 (US$18,126.1448). The total damages awarded to Mr Davis should therefore be revised to US$11,864.34 (rounded to the nearest cent) instead of US$5,983.12.

(b) Fourth representative appellant, Ryan Connell Macquisten: No revision is necessary to the sum of US$28,141.33 awarded by the Judge, which was awarded on the basis that no mitigation sale was required as of 12 May 2022 (Judgment at [151]).

(c) Sixth representative appellant, Xue Yao (“Mr Xue”): No revision is necessary to the sum of US$3,378.81 awarded by the Judge, which was awarded on the basis that all the relevant USTs purchased by Mr Xue were sold by the Cut-Off Time, and thus no adjustment on the basis of a notional sale at the Cut-Off Time was necessary (Judgment at [168]–[169]).

(d) Seventh representative appellant, Mr Arun: We deal with the damages to be awarded to Mr Arun at [163] below, after accounting for other changes to be made arising from his individual appeal.

(e) Eighth representative appellant, Andrew Joseph Che-Bin Lee (“Mr Lee”): No revision is necessary to the sum of US$39,266.23 awarded by the Judge, which was awarded on the basis that all the relevant USTs purchased by Mr Lee had been sold off by the Cut-Off Time (Judgment at [180]–[181]).

(f) Ninth representative appellant, Mr Epstein: The damages to be awarded are the total value of Mr Epstein’s pre-depeg UST purchases (US$1,261,300.26), less the value of his sales from 29 March to 11 May 2022 (US$72,103.13), less the value of his balance holdings of 1,037,445.32 UST at the Cut-Off Time, using the cut-off price of US$0.60485 (US$627,498.801802). The total damages awarded to Mr Epstein should therefore be revised from US$358,099.68 to US$561,698.33 (rounded to the nearest cent).

(g) Tenth representative appellant, Mr Tan: The damages to be awarded are the total value of Mr Tan’s pre-depeg UST purchases (US$66,059.50), less the value of his holdings of 64,246.40 UST at the Cut-Off Time, using the cut-off price of US$0.60485 (US$38,859.43504). The total damages awarded to Mr Tan should therefore be revised from US$14,591.71 to US$27,200.06 (rounded to the nearest cent).

150 We turn next to consider the appeals in respect of four individual representative appellants.

151 The appellants submit that the Judge’s finding that Mr Epstein’s purchase of 3,350,367.35 UST (“8 May 2022 Purchase”) on 8 May 2022 (2.25am UTC) was not in reliance on the Conceded Representations is against the weight of the evidence. First, the 8 May 2022 Purchase cannot have been influenced by the tweets comprising the Seventh Representation, which were made on or after 9 May 2022. Second, the 8 May 2022 Purchase was at a price of around US$0.9967 per UST. This was comparable to, for instance, the seventh representative appellant Mr Arun’s purchases of UST on 7 May 2022 – on which day UST hit a lowest value of US$0.9927 – which the Judge found did not amount to purchases after the de-peg. Mr Epstein’s 8 May 2022 Purchase should have been treated the same way.

152 In our view, the appellants’ submission misreads or omits aspects of the Judge’s reasoning, which when properly understood is not plainly against the weight of the evidence.

153 First, and as the second respondent submits, it is apparent from [185] of the Judgment that the Judge did not find that the 8 May 2022 Purchase was influenced by the tweets comprising the Seventh Representation. Only Mr Epstein’s purchases of UST on 11 and 16 May 2022 were treated as such:

154 Second, the Judge was justified in finding that Mr Epstein’s 8 May 2022 Purchase was not based on his reliance on the Conceded Representations, given the non-recovery of the peg since 7 May 2022, as well as the fact that he invested far more into UST tokens after the de-peg on 7 May 2022, in contrast to his stated motive for investing in UST to yield passive income on a crypto-asset with a stable value (see Judgment at [187]). Mr Epstein’s purchase of about 3.3m UST on 8 May 2022 is in stark contrast to his earlier pattern of purchases, which generally involved far smaller sums.

155 This differs from Mr Arun’s purchases on 7 May 2022, which the Judge found not to be out of place with his prior pattern of purchasing and staking USTs from December 2021 onwards (Judgment at [173]).

156 Mr Epstein’s large purchase on 8 May 2022 supports the Judge’s inference that he was simply speculating on potential profits from any rise in the value of UST after it had de-pegged (Judgment at [187]). This inference is also supported by Mr Epstein’s own evidence, which the respondents rely on. In Mr Epstein’s 2nd Witness Statement dated 1 April 2025 (“Trial Witness Statement”), he stated:

157 During cross-examination, Mr Epstein likewise confirmed that he was seeking to profit from the slight de-peg:

158 It is true that Mr Epstein maintained that he was relying on the respondents’ earlier representations for his belief that UST would return to peg, both in his Trial Witness Statement (see [156] above), and during cross-examination. However, considering that Mr Epstein was buying UST precisely to take advantage of the 7 May 2022 de-peg, the Judge was entitled to infer that Mr Epstein was in fact speculating that UST would rise in value, at a time when the de-peg on 7 May 2022 would already have cast doubts on UST’s professed stability (see Judgment at [187]).

159 Therefore, we reject Mr Epstein’s appeal against the Judge’s finding that the purchase of 3,350,367.35 UST on 8 May 2022 (2.25am UTC) was not in reliance on the Conceded Representations.

160 The appellants submit that the Judge made a clear and obvious error in calculating the amount of damages to be awarded to Mr Arun. Mr Arun had traded UST on two accounts: a Coinbase account in his name, and a Binance account in his sister’s name. Although the Judge had accepted that the transactions in the Binance account should be attributed to Mr Arun – a finding of fact the appellants say this court should be slow to overturn – the Judge had adopted the respondents’ second permutation calculation, which excluded the transactions in the Binance account.

161 The respondents submit that the Judge’s assessment of damages for Mr Arun should nevertheless be upheld, on the grounds that Mr Arun had failed to adduce sufficient proof of his ownership of the Binance account.

162 In our view, the Judge was entitled to accept Mr Arun’s evidence that he had created the Binance account for his sister, but had taken over the account as she had decided not to use it (Judgment at [170]). This finding is not plainly wrong or against the weight of the evidence. Mr Arun had control of the account, as supported by video evidence of him logging into the account using dual-factor authentication. Therefore, the damages to be awarded to Mr Arun should be revised to account for his UST holdings in both his Coinbase and Binance accounts, in addition to the cut-off price of US$0.60485 we adopted at [146] above.

163 Adopting and adapting the calculations supplied by the appellants, the damages to be awarded to Mr Arun would be the total value of his purchases from 29 December 2021 to 7 May 2022 (US$54,654.19), less the value of his sales from 12 February to 11 May 2022 (US$31,381.58), less the value of his balance holdings of 21,850.04 UST at the Cut-Off Time, using a cut-off price of US$0.60485 (US$13,216.00, rounded to the nearest cent). The total damages awarded to Mr Arun should therefore be revised from US$1,990.78 to US$10,056.61.

164 The appellants’ submissions in respect of Mr Davis and Mr Tan are entirely dependent on the Locked UST Evidence. Given our dismissal of SUM 29, it follows that there is no legal basis to disturb the decision below. Their appeals are thus dismissed.

165 The Judge ordered each party to bear their own costs because in his view, “[n]one of the parties has fully prevailed”. The appellants submit that the respondents should be ordered to pay for their costs and disbursements of S$4,231,661.93 on a joint and several basis as they were the successful parties in the first tranche of the trial, having been awarded substantive damages of US$451,451.66 by the Judge. Their claims that were dismissed (for breach of unilateral contract, inducing breach of contract and unlawful means conspiracy) were alternative to their primary claim for fraudulent misrepresentation. Although they were only awarded 5.58% of their total claim for damages, that does not change the fact that they were the successful party, and could not have recovered those damages if not for the suit. The first respondent’s partial success in its counterclaim merely involved obtaining declarations that the information on the first respondent’s websites did not give rise to contractual warranties – which is the obverse of the representative appellants’ unsuccessful claim for breach of unilateral contract.

166 The appellants submit that their costs of S$4,231,661.93 are reasonable and proportionate, considering the significant amount of work done, and by comparison with the respondents’ total costs of S$5,220,000. The work done in the first tranche of the trial was for all the appellants, and it was not reasonable to use the damages awarded to the ten representative appellants to judge the reasonableness and proportionality of the costs and disbursements for all 366 appellants. In the event that the court disagrees, the appellants submit that a maximum discount of 30% or 40% should be applied, reflecting that three of ten representative appellants had failed to establish their individual claims.

167 The respondents submit that there was no successful party in OA 3, because:

(a) The entirety of OA 3 as originally constituted was doomed to fail, until it was bifurcated.

(b) The appellants entirely failed in their alternative claims, their claims in misrepresentation in respect of the Fifth and Seventh Representations, and the claims of three of the representative appellants. The representative appellants were only awarded less than 6% of the representative appellants’ cumulative claim for US$8,085,417.31 in damages (excluding the claims of the represented appellants).

(c) The appellants pursued various claims which were untenable from their inception, particularly the claims of unilateral contract, conspiracy and fraud in respect of the Fifth Representation.

(d) The appellants repeatedly changed their case by amending their pleadings.

(e) The appellants disclosed documents crucial to their claim in an extremely belated fashion, necessitating urgent additional preparation.

168 The respondents also submit that the appellants’ costs were disproportionate, for similar reasons. The appellants’ costs also cannot be compared to the respondents’ costs, as the respondents bore a heavier burden in discovery and the efforts to quantify the damages.

169 Order 22 r 3 of the SICC Rules 2021 states the general rule that “a successful party is entitled to costs and the quantum of any costs award will generally reflect the costs incurred by the party entitled to costs, subject to the principles of proportionality and reasonableness”.

170 In our view, the appellants should be awarded costs, albeit with a significant reduction.

171 In considering who the successful party is, the starting point is to ask which party in substance and reality won the litigation, looking at its outcome in a realistic and commercially sensible way (Comfort Management Pte Ltd v OGSP Engineering Pte Ltd [2022] 5 SLR 525 (“Comfort Management”) at [28]). Where the dispute is about money, the event is typically in favour of the party whom the court has found is entitled to receive money (Comfort Management at [29]). The general rule does not cease to apply simply because the successful party raised issues that failed, but he could be deprived of his costs in whole or in part where he had caused a significant increase in the length of the proceedings or where he had raised issues improperly or unreasonably (Tullio Planeta v Maoro Andrea G [1994] 2 SLR(R) 501 at [24]).

172 In the present case, the appellants are in our view the successful party. They have obtained a money judgment for a significant and by no means nominal sum, even if that constitutes a small fraction of the sum of US$8,085,417.31 claimed by the representative appellants (and excluding the represented appellants). The judgment sum is 5.58% of the amount claimed based on the quantum of damages awarded below (ie, US$451,451.66), and 8.43% of the amount claimed based on the revised quantum of damages awarded on appeal (ie, US$681,605.71). Although the respondents admitted that the Conceded Representations had been fraudulently made, thereby narrowing the scope of the dispute, they did not accept that there was any inducement by or reliance on the Conceded Representations. Seven of the ten representative appellants successfully proved reliance and obtained damages.

173  We recognise that the appellants failed in their claims in misrepresentation in respect of the Fifth and Seventh Representations (the only disputed representations) and in respect of three out of ten representative appellants. They also failed in all their alternative claims, although it is evident from the parties’ submissions below and from the Judgment that, broadly speaking, these alternative claims were much less of a focus than the misrepresentation claims. However, we agree with the appellants that the respondents’ success in obtaining two declarations as part of its counterclaim simply mirrors the dismissal of the appellants’ claim in unilateral contract.

174 Considering that the appellants failed in respect of most of the live issues, it is appropriate that they be deprived of a significant proportion of their costs. Taking into account all the above factors, the appellants should only be awarded 20% of their claimed costs and disbursements of S$4,231,661.93, which yields a costs award of S$846,332.39 (rounded to the nearest cent). This goes beyond the appellants’ submission of a 30% or 40% discount and reflects that the claims of three out of ten representative appellants were dismissed, and accounts for the dismissal of the appellants’ claims in respect of the Disputed Representations and their alternative claims as well.

175  As for whether the appellants’ claimed costs were reasonable and proportionate, the appellants have prima facie shown that their costs are reasonably incurred by providing a breakdown showing the hours claimed, the levels of seniority of their counsel and corresponding hourly rates, and what types of work were done (see Senda International Capital Ltd v Kiri Industries Ltd [2023] 1 SLR 96 (“Senda”) at [73]). The evidential burden then shifts to the respondents to show that these costs were not reasonably incurred (see Senda at [75]). In our view, they have not discharged this burden. It is telling that the respondents’ own costs of about S$5.2m are higher than the appellants’ costs of about S$4.2m – this is perhaps the best evidence of the appropriate level of costs in the particular case (Senda at [75]). We do not think the respondents have successfully explained why they should have expected to incur significantly higher costs compared to the appellants, considering that both sides would ultimately have had to review documents during the discovery process and conduct calculations for the assessment of damages. Therefore, in our view, the costs claimed by the appellants are reasonable and proportionate, albeit subject to the reduction indicated at [174] to account for their qualified success at the trial below.

176 For the foregoing reasons:

(a) We dismissed SUM 29 with costs of S$30,000 all-in to the first and third respondents, and costs of S$20,000 all-in to the second respondent.

(b) We dismiss CAS 2 save for the following:

(c) We partially allow the appeal in CAS 4. The appellants are awarded costs of S$846,332.39 (all-in) in respect of the first tranche of the trial in OA 3, to be paid by the respondents on a joint and several basis.

177 The first and third respondents seek costs and disbursements of S$180,000 in respect of CAS 2 and S$30,000 in respect of CAS 4, while the second respondent states that he had incurred S$110,000 in respect of CAS 2 and S$20,000 in respect of CAS 4 at the time of filing the second respondent’s case.

178 The appellants seek costs of S$205,000 (all-in) for CAS 2 and S$30,000 (all-in) for CAS 4.

179 Taking a holistic view of both appeals, and considering that the respondents were broadly speaking successful in CAS 2 whereas the appellants were broadly speaking successful in CAS 4, we award costs as follows: 

180 We turn to address CA/OAS 3/2025 (“OAS 3”) and CA/OAS 4/2025 (“OAS 4”). OAS 3 is an application for permission to appeal against the decision of the Judge in SIC/SUM 58/2025 (“SUM 58 decision”), in which the Judge made the following primary orders:

(a) the sum of US$41,948,675.49 out of the sum of US$56,948,675.49 paid into court on 28 September 2022 (“Deposit”) be released to the first respondent; and

(b) the sum of US$125,396.13 be released to the first respondent, being the interest accrued on the sum of US$41,948,675.49 between 28 September 2022 and 23 September 2025.

181 The Deposit had been paid into court by the first respondent to discharge a Mareva injunction in the same amount that had previously been imposed on the respondents. In deciding SUM 58, the Judge had adopted the first and third respondents’ calculation of US$13.152m to be retained, which for convenience was rounded up to US$15m, with the remainder of the Deposit released to the first respondent with interest. The sum of US$13.152m had been derived by assuming that all represented claimants were allowed to proceed to the second tranche of the trial, and calculating their claimable loss as the net value of all UST bought, sold and earned as yield in reliance on the actionable representations (ie, only the Conceded Representations), less the value obtained from any sales of UST prior to the Cut-Off Time and a notional sale of all remaining UST holdings at the Cut-Off Time at the price of US$0.8011. The sum to be retained also included the Judge’s award of damages of US$451,451.66 to the representative claimants in the first tranche of the trial.

182 In OAS 3, the appellants raise three grounds of appeal:

(a) First, the Judge made a prima facie error of law in applying the legal test for the renewal of a Mareva injunction pending appeal to determine whether any of the Deposit should be released pending appeal, instead of the legal principles applicable to whether monies paid into court as security should be released pending appeal.

(b) Second, the Judge made a prima facie error of fact which was obvious from the record, in finding that the applicants do not have a good arguable appeal in CAS 2.

(c) Third, the Judge made a prima facie error of law in finding that interest accrued on the sum to be released should be paid to the respondents pursuant to O 26 rr 6 and 7 of ROC 2021.

183 OAS 4 was the appellants’ application for interim stays pending the determination of OAS 4, OAS 3, and the appeal against the SUM 58 decision if permission to appeal was granted.

184 At the hearing for the appeal in CAS 2, the parties agreed that the determination of OAS 3 and 4 would follow the outcome of the appeal. OAS 4 as well as the appellants’ first and second grounds of appeal in OAS 3 as outlined at [182(a)–(b)] above are now rendered moot, since we have heard and decided the main appeal in CAS 2.

185 Part of the Deposit will go towards satisfying the damages awarded to the representative appellants as revised on appeal (ie, US$681,605.71). Part of the remainder of the Deposit will be retained pending the determination of the represented claimants’ claims – the amount to be retained should, however, be revised to reflect our revision of the Cut-Off Price from US$0.8011 to US$0.60485. The remaining portion of the Deposit will be released to the first respondent.

186 We turn to address whether the Judge made a prima facie error of law by finding that interest accrued on the sum to be released should be paid to the respondents pursuant to O 26 rr 6 and 7 of ROC 2021. In our view, the Judge did not make a prima facie error.

187 Order 27 r 6 of the ROC 2021 provides as follows:

188 Order 27 r 7 of the ROC 2021 sets out the relevant rules for the computation of interest on money placed on deposit in court. There are no similar provisions in the SICC Rules 2021.

189 The appellants submit that the ROC 2021 does not apply to the proceedings in OA 3, including SUM 58, pursuant to O 1 r 2(10)(b) of the ROC 2021, which states:

190 We are unable to accept the appellants’ submission. Order 1 r 2(10)(b) applies unless the court otherwise directs and subject to any other written law. Order 1 r 11(3) of the SICC Rules 2021 provides:

191 As such, in the absence of express provision in the SICC Rules 2021, the Judge was entitled to apply O 27 rr 6 and 7 of the ROC 2021 and order that interest accrued on the sum to be released should be paid to the respondents.

192 For the foregoing reasons, we dismiss the appellants’ applications in OAS 3 and OAS 4, with liberty to apply for further directions in respect of OAS 3 if the parties are unable to agree on the amounts of the Deposit to be retained and released in the light of our revision of the Cut-Off Price from US$0.8011 to US$0.60485.

193 Considering the outcome of OAS 3 and OAS 4 in the light of the outcome of the appeal in OAS 2, we award costs of $3,000 (all-in) to the first and third respondents in respect of OAS 3, and costs of $3,000 (all-in) to the first and third respondents in respect of OAS 4.