Someone recently challenged me on my assertion that the CLOB model is stagnant in the FX markets. I am glad they did as I love a good debate, but it became clear that either I had not been clear enough or my challenger had missed the point. I don’t think the CLOB model will disappear, merely that in volume terms it will remain static and therefore any change amongst the providers will be at the expense of competitors.
It is interesting, and related, that this issue of Squawkbox has two stories about algo launches because this reaffirms the sense that the day of the algo has finally arrived. As is often the way I think that just around the time proponents were starting to give up on algos ever going mainstream, that very thing happened. Everyone I am chatting to at the moment is talking increased interest and activity using their algos.
This creates a potential problem in itself of course – if we all focus on the algos, who is going to make the markets for them to execute against? The simple solution is internal risk books and streams, but if more firms step back on market making we end like the equities markets and have liquidity pools with no real inventory to lean on and greater price volatility as a result.
Happily I don’t think that is going to be an issue, but there could well be a change in direction when it comes to the platforms. Algos like predictability and the most unpredictable thing in FX trading emanates from last look. Most analytical tools that back up the algos prioritise firm liquidity – and that often includes proprietary streams with a next to zero reject rate – therefore it seems inevitable to me that more flow will head that way. Given a key value of algo execution is lack of market impact it would stand to reason that market makers would like to be on the end of that flow and as such would benefit more from pricing to firm liquidity pools.
The minute the LPs work this out the algos will pick up on the increased interest and we could have a virtuous circle. Genuine LPs will benefit from the decision to price firmly, recyclers who hide behind last look will be relegated to venues that support them, who will in turn suffer accordingly in volume terms. In effect the computers, with their unemotional logic, will be voting with their business on last look and into the bargain doing something that humans seem incapable of.
There will, undoubtedly, be challenges, specifically if certain hyper-aggressive clients get hold of the algos and start abusing liquidity through a provider’s name, but hopefully the platforms will also supply the relevant details to allow LPs to make a rational and data-driven decision on whether they allow the algo to access their liquidity in the pool. If the platform is willing to police it – and I have to say certain firms’ willingness to do so is not clear to me – then algo performance at those firms who allow potentially disruptive trading practices to exist, will be denigrated.
Of course, if I am right in my outlook here, then the opportunity exists for my core assertion to be proven wrong – and regular readers will know this is not the first time I have argued with myself! The thing is though, I find the whole last look business too pernicious – the vested interests in not only keeping, but promoting it are too strong at this time. Used correctly, I think last look offers LPs valuable protection against predatory trading, but too many still use it over-liberally and they generally sit in the recycler camp.
This growth in algo volume could be coming at a very propitious time for the firm liquidity venues, and any growth will come at the expense of those that support last look. I sense the platforms know this and are preparing for it, but I think something else is also in play, namely the LPs collective reluctance to support an overly-fragmented market structure. For sure the algos’ coming of age may well help the CLOB model, but I suspect the winners’ circle will be less populated that some think.