For quite some time now people have been talking to me about their concerns regarding how their services are used by their clients and last year I asked this question at a P&L conference if I remember rightly – what happens if a client abuses liquidity with one of your algos?
It’s a hard one to track because let’s face it, as an industry it’s hard to work out why anything happens as there are so many trades going through – if we can’t, or won’t, pinpoint the cause of a flash event how are we going to work out who is using whose algo to run people over? My answer when discussing this previously has been fairly simple, ensure the parameters and checks and balances are in place when giving the algo to the client, maybe pop something in that get out of jail free card, the disclosure, about the onus being on the client to use it responsibly, and away you go.
The problem is that the barriers between service providers, mostly banks in this instance, and their customers have grown, thanks in part to the paranoia that has grown up around banks knowing anything of their clients’ actions or intentions. More and more clients are demanding access to different services without the bank really knowing what they are doing, so how can a service provider monitor their activity?
I found these discussions at conferences over the past year provided an interesting paradox that highlighted how one group of industry participants were concerned about their responsibility in this area, whereas another group, the multi-dealer platforms, were at the same time explaining how they cannot possibly regulate behaviour in the market because they only see one piece of it. If nothing else, this highlights the paranoia (that world again) that permeates banking circles in FX at this time, while at the same time indicating that platforms are less concerned about reputational risk as long as predatory trading (in whatever form it takes for you) cannot be laid at their door.
It could be argued that the platforms are taking a more realistic view of the world here, for how can a service provider really control what its clients do with the tools it provides, especially if the client is demanding anonymity of action? There need not be anything suspicious in this of course, as the 2010 flash crash in equities highlighted, even a TWAP algo – probably the most straightforward of all strategies to understand – can be lethal in the wrong hands, and by the time that is realised it is too late.
The anonymity of action being demanded by clients is the very anti-thesis of the Code.
More nefariously, certain clients may see an opportunity in the widespread provision of algos with the benefit of anonymity, but again, to use the platform argument, unless that client uses one provider’s strategies how can it be spotted? They may be running two strategies aimed at manipulating the market in their favour – this very modern day version of spoofing is possible, and probably happens, but again, without oversight of multiple channels how will it be discovered? More pertinently, it can be argued there is clearly an intention to deal, so is this even poor practice? After all, there are some HFTs out there who masquerade under the LP banner when all they are really doing is arbitraging thanks to last look and the speed of their technology. These firms can, and do, argue they are adding liquidity to the market, so would it be any different for a client using multiple and opposite algo strategies?
Perhaps the ‘watermarking’ of algos could be considered in the way that some of the most sophisticated LPs watermark their liquidity to spot clients who are actually recyclers, although that again would be a very complex and potentially fractious process. To me, however, there are a couple of very simple first steps that could be taken to hopefully solve the problem, the trouble is few banks or clients will be willing to embrace them.
Firstly, after a reasonable time – and given we are talking behaviour it can be days – a TCA report of how the client traded could be sent to the provider? This would not solve the issue of a client using two opposite strategies but it may highlight strange behaviour and prompt the odd question of the client.
Secondly, and even more simply, perhaps these strategies should not be given to any market participant yet to sign a Statement of Commitment to the FX Global Code? I have to confess to getting a little fed up of the chatter about how certain institutions “may not” deal with non-signatories, but actually do nothing about it when faced with a refusal from a “good client” to adopt the Code, so perhaps this is a first step? More to the point, the anonymity of action being demanded by clients is the very anti-thesis of the Code.
If a client knows they cannot access the strategies they wish without adopting the Code they will at least think twice about their refusal and may even decide to sign. Of course, this doesn’t mean they will behave in accordance with the principles, but they would be running a serious reputational and perhaps regulatory risk in not doing so. And lest we forget, by seeking to use algo strategies to abuse market liquidity and trade in a less than professional fashion, these firms would be in breach of the Code.
This could be a good testing ground for those LPs and algo providers that have signed the Code to push their clients into adoption, because if poor behaviour does raise its head the answer will be regulatory action and that will certainly not be good for the LPs and, given the latter’s likely reaction to it, will be just as bad for the clients.
It is good that people are still talking about the reputational risks in our industry, but perhaps the time for talking is coming to an end and the time for action is upon us? If so, how people use algos could be a very good starting point.