OK, first things first, my mailbox tells me I owe an apology following Monday’s column and although it is not in my nature, on this occasion it is justified. To all Labradors out there – I apologise. I understand from reader feedback that I was a bit harsh on your fraternity, it will not happen again…probably.
I was using man’s best friend in the context of operational risk on Monday and while anyone with a Lab would tell you they are not that good at matching the size of the gap up with the size of the log in their mouth, my point was they know (I think) when things get too silly.
When it comes to silliness, I received, coincidentally on that same day as I denigrated our four-legged friends, an inadvertent insight into another example of poor operational practice – the eagerness to blame the technology when something goes wrong.
I was on my commute for the next two weeks on Monday and treated to a discussion between two people over an FX execution that went slightly wrong. Aside from thinking “you really should be more careful with your conversations chaps” it struck me that while it is good to be evangelical about something, there should be limits to that.
It may come as a surprise given how enthusiastic I have been about pre-trade analytics over the years in our Digital FX Awards, but there was something so unquestioning, so fervent in their attitude to this service that it gave me misgivings. Not least was the apparent attitude that although the execution hadn’t been great, it was the analytics fault and – this is a direct quote – “the TCA didn’t show it”.
Without wishing to get too philosophical about things, the answer to building a successful trading business is not delivered in absolutes. Pre-trade analytics can play a huge role, but it bothers me that more and more people see them as something to blandly roll into an algo execution solution and let it loose on the market. I get that algos and analytics are good for the industry and have generally delivered good performance, I just don’t think they are a panacea – when they go wrong it can sometimes be horrible.
Going back to the conversation that I and half the carriage had to listen to, there was no apparent concern that the execution followed the recommended pattern but went wrong. This begs the question, “where was the oversight?” We should not be blindly trusting these analytics, after all, there are so many examples of how lazy algos or analytics can create problems – and as an example I would use a LinkedIn message I received this month. The message basically said (and this is not the first tie I have received such an alert, “Congratulate John Smith (name made up of course) on four years in their current job”. I clicked on the link, not the automatic “congratulations” box, only to find the “job” in question is actually “looking for my next role”? Below was a line of congratulations messages clicked by people – the only saving grace here is the poor bloke has probably given up looking at LinkedIn!
There are those that see the analytics as the pathway into greater algo adoption – that may indeed be the case – but those analytics are often – and should be – delivered with a disclaimer (even though our train friends clearly never bothered with them). After all, how is this different from that investment advice that takes 10 pages to tell us that past performance is no indicator of future returns? The disclaimer should basically say, “These analytics will deliver superior performance apart from when they don’t” and “these analytics worked in the past but that doesn’t really mean much actually unless we can get enough people looking at the same data”.
My point is that at this moment in time, even the best technology still requires a sanity check and that check still largely comes from a human and what bothered me the most about the conversation I overheard was this apparent attitude that the blame for a lack of operational process could be thrown onto the analytics – it was nothing to do with the people involved.
Modern day FX markets need people to take more responsibility, not less, and I am increasingly concerned that the reliance upon (admittedly very good) technology is allowing certain people to abrogate this, but there is also a more serious issue. While I understand that the technology often goes through very robust testing, until such time as it can pick out the wrinkle in financial markets that is the equivalent of congratulating someone for being out of work for four years, this willingness to trust – and then blame – the technology represents a weakening of operational risk frameworks.
Good operational practice should be proactive and forward-looking, not reactive – and just as important, we should read before we click on LinkedIn.