The Last Look…

It’s a cliché that the foreign exchange industry is a fast-evolving world, but nonetheless true, and the pace of evolution is leaving a trail of unanswered conduct-related questions over what the right approach may be in different circumstances. One of those areas regards the use of data.

I will confess to raising an eyebrow when I heard, and read, that the Global FX Committee’s Data Working Group is looking at “improving disclosures related to the use of data generated by clients on electronic trading venues”.

As with so much regarding the ongoing discussion around data, we probably need to understand specifically what data we are talking about, because to me, its use in one area would justify closer examination, in another it would be a travesty that panders to those clients who are losing a sense of proportion.

If we are talking about data relating to resting orders, in other words up or down the book that have yet to be executed, then I think it is more than appropriate that disclosures around their use are studied and, perhaps, improved. What role does a resting order play in an LP’s price making? Obviously it needs to play some role, especially if and when the market gets to the level, but what about when it is close to, but not at, the market?

I have heard from several traders over the years about how careful they have to be around orders, to the extent that many bemoan the fact they have to watch them at all – just as pre-hedging or last look is seen as a free option for LPs, so resting orders are for the buy-side.

This has worsened in recent years, however, with the growth of correlations and, more importantly, the concentration of risk books in fewer hands. I have heard, for example, from Cable traders who can’t touch the market because they also run the euro book, and from Aussie traders because they also trade the Kiwi – in all cases the presence of an order inhibited how they acted in the market, stopping them taking a position, and inhibiting how they price.

I should be clear, and this should be obvious to anyone in the industry, bidding or offering just in front of a client order is just plain wrong, and should be a disciplinary offence, but my understanding is that just about all disclosures now cover this. More to the point, the level of details some clients are required to give around partial fills and slippage is much higher than it was just a decade ago.

That said, if there are genuine concerns over how order data is being used, then we should indeed take a close look and make the appropriate recommendations to avert future problems.

If two parties trade, how can the data “belong” to just one?

I view with irony GFXC chair Gerardo García’s comment to me during an interview that end clients are “increasingly inquisitive” about how their data is used – we probably need to understand where that data is no longer “owned” by the client. After all, if two parties trade, how can the data “belong” to just one?

I assume that these end clients have signed a Statement of Commitment, but inevitably my mind casts back to the ECU Group case where the customer left large, simultaneous orders with multiple banks, and then tried to sue one of them for front-running them (pre-hedging anyone?). I would argue that leaving orders in such a fashion is unfair and unethical practice, but then at the time the Code didn’t exist and outside a commended few, hedge funds haven’t signed up to the Code thus far anyway.

Where I think there would be an over-reach is if clients believe the data is “owned” by them post-trade. If the clients are asking for more transparency of how their data is handled post-execution, I would suggest it has little to do with them. The FX Global Code is clear that LPs shall not use trade data in the last look window, so hopefully that isn’t happening (but sadly it probably is in some dark corner of our industry, well away from the mainstream), but once a trade is done, then surely the data is public?

I spoke to someone about this and they suggested the concern was aimed at so-called “full amount” trading, which is, of course, nothing of the sort, rather it is just another example of modern society’s penchant for obfuscation – much like “pre-hedging”. This person observed that the concern was over how dealers used the earlier “full amount” trades in their ongoing pricing.

I may be missing something here, and am happy to be corrected, but to my understanding there is nothing written down that says a client has to keep hitting the same “full amount” stream. In other words, yes, the client may hit a “full amount” stream for 20, but the LP doesn’t know if they will be back. Even if the client does come back, the LP doesn’t know the final amount, so any trade could be the last clip.

If that client hits a multi-dealer platform, that platform’s data will reflect the trade, so my sense is this is really a single-dealer platform issue in the clients’ minds. It is a tricky one I suppose, because the LP may have a very good idea that more trades are coming and start skewing, and yes, while I know they probably won’t win the trade, they will probably have signalled to the external world that there’s a buyer/seller out there.

The only way that signalling risk/market impact can be allayed is if the LP warehouses all the risk, but again, how do they know how much is still to be done? A book may have a 200 million limit (in many dealers’ dreams I suspect), so what happens if the parent order is 250? Do they take the 200 on board and then go “oops” and start selling the last 50? The dealer has to have a chance to make money on the trade, but in these circumstances, it’s a tough ask because, again, they are being inhibited in how they trade.

Surely a trader has the right to adjust their position and pricing according to the risk they are accumulating? If the client is looking to do a big ticket, perhaps they need to use an algo? Of course, the algo doesn’t hold risk, so the signalling and market impact will come to the fore at some stage, unless the client is discovering the secret to unlimited liquidity – being wrong!

Are these concerns around data are actually a reflection of what has become poorer execution quality generally for clients?

This is an old theme in these pages, but if the client is worried about their trade data and market impact then they should work with the LP, preferably by voice with the relationship team, who in turn will use electronic means wherever possible to trade and process the deal. That is the only way to control the level of information dripping into the market – because even the best algos are occasionally going to get caught out.

The only way to avoid market impact for a client’s trade is to give it in true full amount to a dealer, but this seems to have gone out of fashion, which is a shame, because unlike a decade ago, the checks and balances and surveillance is in place to ensure that, with the major dealers at least, the order is being handled fairly and with maximum discretion. If you achieve that, then you shouldn’t have to care about your data.

One thing I have been wondering is if these concerns around data are actually a reflection of what has become  poorer execution quality for clients generally, because, as I wrote last week, the risk absorption mechanisms just aren’t there? Historically, with a few honourable exceptions, clients are not that proactive around their FX dealing, so why bring it up now? Is it because in an age where the LPs better understand the true value of the client’s business, they are widening some of them? Is it because too many dealers are pushing their algos over their risk warehousing capabilities, pushing responsibility back onto the client and opening some of their eyes to the concepts of signalling risk and market impact?

Whatever it is, this seems to be a issuing that is going to grow, and it would be a good thing to get in front of the curve by establishing exactly what data these clients are worried about.

colin@thefullfx.com