HP-2020-000016 - [2025] EWHC 1451 (Ch)

I do not need to address MGA’s submissions on the Base Case 2 simulation model or the Cabo projections model, given the position reached on these by the end of the trial. The relevant question is the reliability of the Base Case 1 and Hatchimals simulations, as presented in Cabo’s closing submissions. As to that, MGA criticised both the overarching approach of the model, as well as its main input assumptions.

Overarching approach. MGA said that Mr Colley’s simulation approach was based on a hypothesis of market dynamics which did not reflect the reality of the market for collectible toys. I have some sympathy for that submission. Ultimately, any estimate of the profits which Cabo might have made, in a counterfactual scenario which did not occur in fact, is likely to be highly speculative and based on very broad-brush estimates and assumptions. The court’s analysis in such a case is a paradigm example of the wielding of the broad judicial axe. In those circumstances the construction of a sophisticated simulation model carries the risk of turning out to be an exercise of spurious precision. Whether or not its results are robust will depend upon the extent to which the model corresponds to the reality of the market. The more sophisticated the model, the more sensitive it is to the fallibility of its assumptions.

The Bertrand model used by Mr Colley in this case was a model of static price competition in a market with a relatively small number of firms competing on price. That is likely to be an unreliable predictor of market dynamics in a volatile market with a large number of competitors, with significant competition on non-price factors (in this case product differentiation).

An example of the problem of using this sort of model was the fact that the output prices for Worldeez predicted by the model, based on the simulation of the price competition that would occur under the model, were far higher than Cabo was able to charge in fact. For example, the weighted average retail price for the globe predicted by Base Case 1 was over £4 in 2017 and around £3.50 in 2018, whereas the globe was in fact sold for £2.99, and the evidence supported this as the right price for the product both in the actual and the counterfactual case (see §550 above). Mr Colley, when cross-examined on this point, confirmed that he was aware of the actual selling price of the globe, but that the model “comes up with another equilibrium” as a mathematical result. That answer did not, however, grapple with the question of how the court could usefully use a mathematical model which resulted inoutcomes that were on any basis (even in the counterfactual scenario) wholly unrealistic.

In Cabo’s closing submissions, Cabo proposed that this problem could be addressed by overriding or constraining the output retail price in the model for 2017 (only), so as to correspond to the actual Worldeez launch prices, while retaining a higher retail price for subsequent years. The final version of the DTM reflected that price constraint for each version of Mr Colley’s model. This proposal was not, however, addressed in any of Mr Colley’s evidence prior to the trial. His only mention of the point was a very brief comment, on the afternoon of the second day of his cross-examination, that it would be possible to constrain the prices in the model. That was far too late for this to be explored with him and indeed with Mr Parker.

Cabo’s proposed solution was, moreover, internally contradictory. On the one hand, Cabo said that Mr Colley’s model predicted a slightly lower retail price in 2018 than in 2017 because of the hypothesised increased competition from LOL Surprise, which would have required other brands to lower their prices to compete. On the other hand, Cabo’s price-constraint solution implied that Cabo would have increased the price of Worldeez from 2017 to 2018, with Cabo justifying that by saying that in the light of early success it was not unrealistic to assume that Worldeez would have raised its price. Ultimately, Cabo’s solution simply highlighted the artificiality of the model on which it relied.

The selected comparators. Turning to Mr Colley’s input assumptions, it was common ground that since Mr Colley was not himself opining on the prospects of success, all his model could do was to predict profitability if Cabo had enjoyed the levels of success identified as the starting assumptions in the various versions of the model. That in turn depended on the whether the comparators chosen by Mr Colley to calculate his “quality index” for Worldeez were reliable predictors of the commercial success that Worldeez would have achieved in the counterfactual case. I agree with MGA’s submission that the comparators chosen by Mr Colley were in fact wholly unrealistic.

Mr Colley’s Base Case 1 specified as the comparator products five brands (LOL Surprise, Shopkins, Hatchimals Colleggtibles, Num Noms and Mash’ems/Fash’ems) which were by any measure extremely successful toy products – indeed, in the case of both LOL Surprise and Shopkins, phenomenally successful. All five were, moreover, brands marketed by major international toy manufacturers, with a track record of producing successful toys in the UK. Base Case 1 assumed that Worldeez would have achieved commercial success equivalent to the average of the success of those five brands.

The evidence did not, however, suggest that Worldeez could realistically have achieved that level of success. Mr Harper’s view was that Worldeez would not have done as well commercially as Hatchimals Colleggtibles, for the reasons outlined at §520 above and discussed further below. He considered that Num Noms and Mash’ems/Fash’ems were also both better products than Worldeez (see also §520 above). As noted above, Ms Munt said very little about these products. As for LOL Surprise and Shopkins, there was no serious suggestion, at least by the time of the closing submissions, that Worldeez would have achieved anything approaching the commercial success of those products (whatever might be said about some aspects of the comparability of Worldeez with Shopkins, discussed above). Particularly when considered in the light of the extensive evidence as to the operational failings of Cabo, the selection of the Base Case 1 comparator set was hopelessly unrealistic as a measure of Cabo’s likely profitability.

That left the comparison with Hatchimals Colleggtibles alone, or alongside Base Case 1 to produce an average figure. Prior to Cabo’s closing submissions, this comparison featured only briefly in an Annex to one of Mr Colley’s earlier reports as a sensitivity analysis, and was described as such in the first joint economic expert report. In Cabo’s closing submissions, however, it emerged as one of the preferred scenarios alongside Base Case 1. I do not consider that to be entirely satisfactory: unsurprisingly, since this comparison had never previously been relied on as one of Cabo’s primary profitability models, it was not addressed in any detail by Mr Parker, and was not specifically explored in Mr Colley’s cross-examination.

In any event, the Hatchimals comparison was no more realistic than Base Case 1. Hatchimals Colleggtibles was one of a small set of extremely successful toy products in 2017, and was even more successful in 2018. While there was no dispute that, as a product, Hatchimals might be regarded as comparable to Worldeez, Mr Harper’s considered view (in both his written and oral evidence) was that he would have not expected Worldeez to perform as well as Hatchimals Colleggtibles, both because that collectible product was launched following the success of the original (larger and more expensive) Hatchimals product, and because of the fact that it was produced by an experienced international toy company, Spin Master. There was no evidence to contradict this: Ms Munt did not consider whether Worldeez would have enjoyed a similar level of commercial success to that of Hatchimals Colleggtibles.

In fact, because of Hatchimals’ continued success after 2017 the Hatchimals comparison in the simulation model produced an even higher estimated level of sales and therefore revenues across the lifecycle of Worldeez than Base Case 1. If the level of success implied by Base Case 1 would have been unachievable for Worldeez, then the same would have been true a fortiori in relation to the Hatchimals benchmark. This alternative benchmark therefore simply substituted an unrealistically high success model with an even more unrealistically high success model.

Cabo sought to characterise Mr Colley’s comparators as being “conservative”, suggesting that they placed Worldeez (in the counterfactual case) at the bottom end of the successful brands at the time. Cabo’s comparison was presented as a bar chart in its written closing submissions, with an amended version in its post-trial submissions. That comparison, however, was not explored in the expert evidence and suffered from similar problems to those arising with MGA’s comparisons with other products in the breakeven analysis (see §§575–576) above, namely that Cabo’s comparison was based on sales during the entirety of 2017, in circumstances where Worldeez did not launch until June in that year. Even if Worldeez had been as successful as (say) Hatchimals Colleggtibles in that year, a comparison based on sales only in that calendar year would present Worldeez as having lower revenue figures simply because of the shorter sales period. Cabo’s comparisons were also based on sales revenues alone, which did not reflect sales volumes given the quite different price points of the various products being compared. In addition, while Mr Colley’s simulation predicted profitability across a five-year lifespan, Cabo’s bar chart was limited to comparative sales in 2017, thereby masking the fact that Hatchimals went on to be more successful in 2018 product than in 2017. I do not, therefore, think that this comparison provided any meaningful indicator of the reliability of Mr Colley’s comparators.

In my judgment, the comparators used to populate Mr Colley’s simulation model could not provide a reliable indication of Cabo’s likely profitability in the counterfactual case. While Mr Colley said, at the trial, that it would be a very easy matter to re-run the model with a different selection of comparator products (“It is a flick of a switch”), his evidence did not consider any other comparisons, despite his detailed assessment of other similar products for the purposes of his market definition. That was indicative of a lack of objectivity in his approach. Given that Mr Colley was not purporting to give evidence on the level of commercial success which Cabo would have enjoyed in the counterfactual case, it would have been helpful if Mr Colley had considered a broader spread of potential comparators, rather than confining his analysis to a small set of extremely successful products.

Cabo initially suggested that if the court considered that the comparisons used by Mr Colley assumed a level of commercial success that Cabo would not in practice have achieved, that could be corrected by use of the probability adjustment feature of the DTM, i.e. by selecting one of Mr Colley’s models as the starting input, but then reducing the cashflows implied by that model by selecting a probability of less than 100%. While I agree that this would be a possible approach in theory, the mechanism is (as noted above) not a true probability adjustment, but simply a means of reducing the cashflow by a specified percentage. That would be a paradigm example of spurious precision: if the starting point is inherently unreliable and improbable, any reduction of that by the court is a venture even further into the realms of speculation.

A far better approach, if available to the court, is to select as the starting point the model which most closely approximates the likely success of Worldeez, based on the evidence before the court, and to consider what that model indicates regarding the profitability of Worldeez in the counterfactual case. Notably, by the time of oral closing submissions, that point was essentially common ground, with Mr Kuppen agreeing that if the court was able to form a view (based on one of the models) as to the likely counterfactual trajectory of Cabo, there should then be little need for any probability adjustment.

Licensing and international sales. MGA’s other main criticism of Mr Colley’s model was that it assumed that Worldeez would have been able to secure licensing deals, and also assumed very high levels of international sales. As shown above, the international revenues made up the vast majority of the profits predicted by all three versions of the model put forward by Cabo in its closing submissions. Given my findings above as to the suitability of Mr Colley’s model itself and the comparators used to populate it, the issue of licensing and international sales based on that model does not arise. For completeness, however, I agree with MGA that Mr Colley’s assumptions regarding licensing and international distribution were unrealistic and flawed.

For licensing, Mr Colley based his calculations on an estimate produced by MGA in 2017 of its prospective sales of licensed products for LOL Surprise, relative to its LOL toy sales revenue, and a royalty rate of 18% on those sales. Both of those assumptions were wildly optimistic. There is no basis whatsoever to assume that Worldeez would have obtained licensing sales on the level of LOL Surprise, an extraordinarily successful toy marketed by an established and experienced toy brand. Nor was there any basis to assume that, even if Worldeez had concluded licensing deals, it would have obtained a royalty rate of 18%. Ms Munt suggested typical royalty rates of 10–12%, based on indicative rates in a third party licensing handbook, the Brand Licensing Handbook. She accepted, however, in cross-examination that she had not been involved in the negotiation of royalty rates. Mr Harper did have direct experience in negotiating licensing agreements (albeit not in the toy market), and was able to give evidence of his knowledge of royalty rates drawing on his extensive experience in the toy industry. His evidence was thatwithout substantial TV programming support for a brand, its royalty rate on licensing was likely to be around 5%. The one licence deal which Cabo was offered, in March 2018, was a magazine licence which offered a royalty rate of 5–7%, depending on sales. Neither the factual nor the expert evidence therefore supported the royalty rate of 18% used by Mr Colley.

Mr Colley also accepted in cross-examination that his estimates as to LOL’s licensing revenue (from which he derived his predicted licensing revenues for Worldeez) were themselves very significantly overstated. While he had estimated that MGA had made £6.7m in UK licensing revenues for LOL Surprise in 2017, in fact MGA’s internal reporting indicated that it had generated only $2.4m global licensing revenue in that year. Mr Colley’s methodology did not, therefore, provide a reliable basis for an estimation of Worldeez’ prospective licensing revenues, even if Worldeez could realistically have been expected to generate licensing revenues on a scale comparable to those of LOL Surprise (which it plainly could not).

As for international sales, for the reasons discussed above I do not accept that if MGA had not intervened Cabo would have been successful in selling Worldeez outside the UK, the essential problem being the feasibility of any international distribution arrangement given Cabo’s cost model and the working capital requirements for international sales. These were not matters considered by Mr Colley, who simply assumed that international sales in Europe, North and South America, and Australia would have been likely once Worldeez had “gained traction” in the domestic market.

It is therefore apparent from the evidence that Mr Colley’s estimates of licensing and international sales revenues were unrealistic, even leaving aside the problems with the comparators used for Mr Colley’s simulation model. This adds to a picture of a model which could not ever have provided a reasonable and objective estimation of Cabo’s counterfactual losses.