A week and a half ago, during our In the FICC of It podcast, Galen Stops and I briefly touched upon a comment from a speaker at one of our events, that clients were looking at deploying an algo that deliberately lost money for the first 30 seconds. At the time we chatted away, as we do, and I casually mentioned that to me, best execution could involve a child order which goes against the parent in order to throw off the order sniffers in the market.
I thought little more of it at the time but since then feedback I have had suggests this is quite an emotive subject in the industry. Almost universally, everyone I spoke to in the banking world stressed how they could not even consider being a part of such a strategy, let alone deploy it, and while I think this says as much for the paranoid nature of that industry as it does for the value of the strategy, the peremptory nature of the dismissal was still a little surprising.
It really comes down to a question of transparency, who it works for and who should it work for. I have stated so many times that the real reason exchange-traded FX is unlikely to dominate at any time in the foreseeable future is because most customers, especially those from the real economy, do not want their order to be made public. This is why they choose risk transfer and are, I am told, pushing back against bank efforts to sell them an algo execution solution (and as a fee is payable “sell” is the appropriate word here).
There are other hedgers out there who like the idea of having more control over their execution and who do embrace the concept of algos, of course and although I think true transparency would involve those selling the algos to make clearer that the customer owns the market risk, most of those using algos at the moment do understand this and they have a strong interest to minimise market impact. It is here that I think we are missing the point on best execution.
Too many people, thanks mainly to an overbearing compliance function, see best execution as ticking the right boxes on the TCA report – be it independent or vested. It should be nothing of the sort. A huge part of best execution is signalling risk and slippage, with one directly leading to the other. By using an algo there is a risk that the order could be sniffed out thanks to the sophisticated techniques used by some LPs and prop trading firms and therefore the cost of execution will rise.
If you want a pretty obvious example of this, look at the London 4pm Fix, where dozens of firms lie in wait to see which way the algos go. TWAP and Fix-replicating strategies have next to no flexibility, which is the way many like it, however this lack of flexibility is what is causing the signalling risk and raising the cost of trading.
If a user, or an adaptive algo, can be given the option to reverse course for a moment in time (at the user’s risk of course – and that would be a decision not taken lightly), then these sniffers can be thrown off and the overall cost of execution reduced. It won’t always go right, but I suspect it would go well often enough to make it a viable option for the client – and this is what it is about, delivering a better outcome for the client. I hear and see too many instances of obvious market impact being given the thumbs up in the TCA report when more sophisticated analysis would suggest things could have been done better.
At the moment the only defence mechanism is pausing the algo, but that doesn’t create new market data to throw the hounds off the scent – a few child buys in a parent sell order would.
I am sure there are more than a few of you out there asking whether this would be morally correct – and you have a good point. The thing is though, is this any more morally bankrupt than wilfully ignoring the signalling risk emanating from using certain strategies at certain times in certain markets…or using a strategy deliberately designed to lose money at the start? Equally is it any more morally difficult that showing a customer a price and then last looking it for longer than average because the client’s flow is generally toxic…or machine gunning several LPs under the name of best execution? There is so much in the business that could be viewed as morally uncertain but we live with it because generally speaking the people who are trading FX are consenting adults who understand the risks involved,
Ultimately this is a trading decision that should be taken by a responsible adult – and if they can’t take that decision they should not be taking on market risk. Unlike the aforementioned instances I am struggling to work out who will be hurt by such a strategy beyond;
- The customer if it goes wrong (but that is a risk they need to acknowledge before deploying the strategy)
- An order sniffer who would get caught out (and if they are true traders they will accept the occasional loss)
I absolutely accept this is tricky ground for a bank to step into and if I were at desk head level I would have nothing to do with it until either I was directed (in writing – this is a paranoid business now) by senior management or, more preferably, the practice was signed off by compliance (which is unlikely as so few understand the business well enough).
It is, perhaps, less tricky ground for the client to tread upon, especially as so few have signed the Global Code (which as far as I can tell doesn’t cover such an instance, it being very ‘niche’). For many of these clients, being able to confuse other participants trying to jump in front of your order would improve execution quality – the challenge would be measuring or monitoring it – and that where the transparency in this issue is important; in the decision making process, especially if it is an adaptive algo. If the client does the appropriate due diligence on the algo and understands the type of circumstances when it might try to reverse signal, and accepts that, then we have the market data to prove the decision was arrived at appropriately.
Some have, in the past, described what I am talking about here as spoofing, but I would contest that because in this circumstance there is not only an intention to deal, but actually dealing is a crucial element in it – market risk is being taken.
So there we have it – again there is an instance when one person’s best execution could be viewed as another’s market abuse, but if this really is about achieving the best outcome for the client when acting as an agent then perhaps we need to embrace the subject more? It would need a strong compliance and data framework around it, but techniques – as I am told almost daily – continue to improve, so where’s the problem? (answers on a postcard)