Friday, February 26th, 2016
by GEORGE BOLLENBACHER
Regulators have cast a rather harsh spotlight on the practice of ‘last look’ in the FX. So what does the future hold for last look. If it survives at all, you can be sure it will look and act very differently than it has in the past, says Capital Markets Advisors’ George Bollenbacher.
Several developments on the legal front over the past few months have cast a rather harsh spotlight on a practice in the FX markets called “last look.” FlexTrade’s Ivy Schmerken wrote an excellent review of the issue on this forum (“A Hard Look at ‘Last Look’ in Foreign Exchange”), but what I want to know what the future holds for last look. My conclusion? If the practice survives at all, you can be sure it will look and act very differently than it has in the past.
How it began
The practice of last look actually began rather innocently. Not long after single-dealer electronic platforms began proliferating, the builders of the first HFT programs began monitoring those dealer platforms, looking for quotes that hadn’t kept up with market changes. In short order, dealers got tired of being picked off by the HFTs, so they instituted a “last look” policy, which gave the dealer an instantaneous last chance to review its bid or offer before accepting the incoming order.
As often happens, however, what started out as self-protection became somewhat more predatory. Since no human could execute last look, the process became an electronic millisecond event, and then morphed into a scan of market movement while the order was held for a few microseconds. If the market moved in the firm’s favor during that period, the order was executed – if it moved against the firm, the order was rejected.
Soon, the process got more than a little out of hand. According to a New York State Department of Financial Services (“NYSDFS”) settlement with Barclays, one of Barclays’ customers had “over 300 rejected orders” in one day, and routinely “got rejected … 9 times out of 10.” Needless to say, NYSDFS found plenty of evidence of a process gone seriously bad, including incriminating emails, and ended up fining Barclays $150 million for the practice. And, not surprisingly, class action attorneys were immediately on the case, and Barclays subsequently settled one of those cases for $50 million. Finally, there is no doubt that more banks are in the crosshairs of both the regulators and plaintiffs’ attorneys.
Looking both ways
At first glance, this looks a lot like what Brad Katsuyama discovered in Michael Lewis’s book, “Flash Boys.” Were Barclays’ customers simply the victim of a classic bait and switch, where Barclays figured that, once rejected, they would come back for a worse execution, no matter what the market was doing? And it isn’t just Barclays that is under suspicion – it’s every dealer. It almost doesn’t matter now what was happening behind the curtain, since nobody trusts the Wizard of Oz any longer.
So now we need to figure out if the practice can be redeemed, or have we seen the last of last look? Barclays has published a sort of a mea culpa on its website, indicating that it will now run last look both ways. What that means, though, is that whether the market moves for or against the customer, once it moves the customer will get a NACK (“Not Acknowledged”) rejection message instead of an execution. However, the customer will have telegraphed its intentions to Barclays without getting anything back.
So that’s really not much of an improvement. In volatile markets a customer could spend hours trying to execute on a dealer’s electronic platform, never knowing if the last NACK was because the market moved with him or against him, not to mention flipping back and forth between several dealer platforms, hoping for an execution.
One fairly obvious possibility is that dealers and customers could establish a set price range around the quote on the screen, so the customer gets a little worse execution if the market goes against him and a little better if it goes his way. That depends, of course, on the dealers being honest about which way the market moved – it would look more than a little suspicious if every trade was done at a little worse than the screen price.
Looking out for yourself
The opposite alternative is to do away with last look entirely. In that case, the price on the screen when the order arrives is good for the size listed, no changes allowed. Since most FX dealers use algorithms, not people, for trading, as do many customers, the trade would automatically be executed when the order arrives, as long as that price is still showing on the screen.
We should expect some implications from that approach, of course. One would be wider spreads in spot markets. The second affects the way customers would do business. With wider spreads, the responsibility to get the best price now weighs heavily on the customer, particularly the institutional asset manager. So certainty of execution would come at a price; the question is whether it would be a fair price.
So the last look controversy has all the earmarks of a good idea gone really bad. An arrangement that gave customers high-quality markets while providing dealers protection against sharpshooters somehow devolved into another behind-the-scenes game of three card Monte. Once again we’ve learned, as we did with dark pools, that some people will take advantage of every opportunity, and trusting the system is only for the naïve. I keep seeing the image of that farmer standing over the corpse of the goose, with all those golden eggs littering the ground. And so it goes.
