CL-2018-000297, CL-2018-000404, CL-2018-000590, - [2025] EWHC 2364 (Comm)

In his oral evidence at trial, Mr Baker of Lindisfarne said that he did not think of this settlement process as a loop, because “I don’t see any start or finish in it, I see things concurrently”, and “there is a certain connotation with a loop that … nothing changes and everybody is the same … [but] this is different. I think people are taking on contractual obligations. I think there is risk there.” I agree with Mr Baker that there is no start or finish to the loop. To spell that out:

I do not think that makes it inappropriate or inaccurate to think of this as a loop. There is then, as Mr Baker said, a certain connotation that nothing changes, but as regards share ownership indeed nothing changes. At no point, before, during (if that is meaningful) or after settlement, does any of V, L or B own, acquire or transfer any shares. I agree with Mr Baker nonetheless that it would not be true to say that nothing at all has changed. The position after settlement is not identical to the position prior to settlement, or prior to trading. That is because if this internalised settlement by the custodian is performance, not cancellation, of the individual transactions, then V and L now owe to L and B, respectively, a stock borrower’s contractual obligation to transfer securities at the end of the stock loan that (other things being equal) will have to be covered in due course, and I come back to that in paragraph 74 below.

For such an internalised, share-less settlement to be a contractual performance, all of B, V and L would need to have agreed to it or to ratify it after the fact. But then, taking account of the agreement to allow that kind of settlement (or the ratification of it after the fact), what might otherwise have been a sale by V to B requiring a share transfer was in fact a trade that V could perform without any share transfer (and likewise for the stock loans).

Mr Baker’s observation that the settlement loop has no beginning is an important one for that issue about the contractual arrangements: prima facie, V’s sale to B cannot settle unless V has shares to transfer; V will have no shares to transfer unless its stock loan from L settles with a transfer of shares; but that cannot happen unless V’s sale to B settles with a transfer of shares. On the face of it, therefore, the equity sale and both stock loans should all fail to settle since no party has any shares to transfer. That is what a contract term or ratification as to settlement methodology would have to overcome.

The simplicity of the basic analysis I have now set out can be clouded by jargon. Talk of ‘traded positions’ in contrast to ‘settled positions’ (or ‘depot positions’) may cloud the difference between a contract to buy shares and a shareholding. Talk of whether a custody record represents a ‘settled balance of shares’ may cloud the reality that a custody record in the books of a custodian with no shares cannot be a shareholding, and cannot make the custody client a shareholder.

In his written evidence in chief, Mr Dhorajiwala noted that the idea at the centre of the transaction structures in this case, on the cash side, was that “it might be possible, as a custodian to create leverage synthetically via [as he put it] netting transactions between buyers and sellers of shares in the same omnibus account of a custodian”. I would prefer to say via the internal settlement of transactions, between buyers, sellers and interposed stock lenders, by equal and opposite account entries at a common custodian; but Mr Dhorajiwala was right to say that the transaction funding was synthetic, that is to say artificial. On the equity side, where all there ever were or would be were book entries in the accounts of a custodian with no shares, likewise any ‘shareholding’ evidenced was synthetic, not a holding of any shares.

To explain my quibble over terminology in the previous paragraph, I would now not say that an ‘omnibus account’ or ‘netting’ was used, and I acknowledge that means my own use of that language in SKAT (Validity Issues) was not as precise as it might have been:

I therefore disagree with Messrs Jain and Fletcher, who in one of their written submissions for closing said that “SKAT proposes that there was no delivery (gross discharge of primary obligation) if the net balance within an omnibus account at the end of the day was zero”. Firstly, that description treats ‘delivery’ (of a shareholding) and ‘gross discharge of primary obligation’ as one and the same, whereas SKAT’s allegation was, and my finding is, that the settlement methodology utilised here treated the primary obligation as discharged even though there was no delivery. Secondly, SKAT’s point was not that on the settlement date there was a nil end-of-day omnibus account balance (although no doubt that was true), rather its point was, and my finding is, that any omnibus account that existed at all was empty (devoid of any shares held by anyone for anyone) at all times.

In the trading models considered in this judgment, there was always a second phase of trading, an ‘unwind’ phase. Where stock lending had been used to settle the initial phase, there had to be an unwind because, as Mr Rabinowitz KC put it on Day 1, “otherwise the short seller is left with an obligation to return shares that it doesn’t own and the buyer is left with an obligation to return cash collateral that it doesn’t have.” It suffices to say, in broad terms, that:

In cross-examination, Mr Horn suggested that there was no requirement, when each set of initial trades was put on, for there to be an unwind set in the future. He was correct about that only in a very narrow sense, namely that the individual trade terms for the initial equity purchase and stock loans did not provide for it. However, in line with Mr Rabinowitz KC’s pithy summary, the practical reality was that stock loans would come to an end and the participating entities would have no means to perform the obligations then arising without a managed unwind of the type just described. Furthermore, it would not have been rational for those entities to enter into their respective initial trades unless they were confident that such an unwind would be arranged. Mr Horn’s attempt to quibble about that was not, in my judgment, because he was trying to be precise, making just the narrow point with which I opened this paragraph. He was trying to lay the ground for a line he intended to spin (and which he did spin the following day), in relation to Solo Model trading in Belgium, that the equity buyer was somehow free to do whatever it wanted with its entitlement to recall its stock loan, whereas in reality, as (I find) Mr Horn knew, the equity buyer was only ever going to recall the loan when told to do so by Solo, on terms determined by Solo, as part of Solo’s management of its trading model.

None of what I have said so far is affected by whether, or as from when, as a matter of market practice, a trading party might think of itself as ‘owning’ ‘shares’, or by what exactly they might think they meant by that if they did. Any market practice of that kind cannot convert book entries at a custodian that holds no shares into shareholdings.

It makes no difference of substance in paragraphs 61 to 76 above if, rather than a direct sale by V to B, there is a matched pair of sales, by V to C4 and by C4 to B (as illustrated by Mr Wade’s diagram reproduced at paragraph 65 above); and it then makes no difference whether, if that is the position, it is created by two trades concluded with C4 (V selling to C4, C4 selling to B), or by trades matched through a broker (V selling, B buying) where the broker gives the trades up to C4 for settlement, or at all events uses C4 as custodian for settlement of the matched trades. It is also no difference of substance if, where a broker is used, a different broker is used on the unwind.

Such additional complexity does not change the fact that, in the settlement loop described above, reversed on the unwind, even if, as a matter of contract:

In that simple description, only a single set of transactions is assumed, in which B buys and agrees to lend X%, V sells and agrees to borrow X%, and L agrees to borrow and lend X%. However, it again makes no difference of substance if, instead, “B”, “L” and “V” are sets of “buyer-lenders”, “borrower-lenders” and “borrower-sellers”, respectively, each of whose own positions is matched as to share volume, so that, for example, each buyer-lender, for its own part, contracts to lend whatever volume it buys, and all transactions are brought together for internalised, share-less, settlement at C4. Similarly, it makes no difference of substance if there are additional links in the transaction chains, putting greater distance between “B” and “V”, for example:

The more complexity of that sort that is introduced, the more challenging it may be to draft the contractual terms required for C4 to be entitled to operate a share-less settlement process equivalent to that which I have described for a simpler structure. But if that were achieved, still the upshot would be that: at C4, where everything balanced out, all transactions could be settled internally without any party ever having, acquiring or transferring any shares; even if under each contract, any seller or stock lender is treated as having delivered rather than defaulted, still the whole operation can and does occur in a world disconnected from VPS and DanCo, as illustrated by the diagram at paragraph 61 above, and thus involving no shares, that is to say, so far as shareholding (share ownership) is concerned, in a world entirely synthetic.

Sanjay Shah illustrated a more complex loop within a set of PowerPoint slides he prepared for use in his criminal trial in Denmark and adopted as part of his evidence in chief before me. The slides were proffered as an encapsulation of how Mr Shah says he understood things at the time. They included the following diagram to illustrate a settlement loop relating to a TDC dividend declared in March 2015: