There is an old quote which says that "past performance is not an indication of future outcomes". Modern day trading is no exception. The fact is while the market may surprise the trader - which is incidentally why monitoring moved from t+1 or more, to real-time - it shouldn't be a shock that the majority of successful traders are slaves to their discipline. They trade the way they always did, until they don’t.
When they don't, do we want to be the statistic that is a headline about a disaster brought about by one rogue trader? Wouldn't it be more prudent to have the analytics to spot the activity before the trader earns the nickname 'Rogue'? Today's problems aren't typically the bad trade stuffed in a drawer, or the trades being 'lost' in an error account for weeks or months. They are more likely to be a machine that runs amok or a trader who has moved away from his usual disciplined pattern.
We are all watching the proscribed 'stuff'. We are checking the defined parameters. We are (or some of us are) logging the trades, checks and market events. We are able to gather the alerts, other account parameters and build a view of the current day's events. However, these alone won't tell us if the account has remained 'within its limits’.
Gathering data from disparate systems and marrying it with other equally or more contrasting platforms is part of everyone's daily vocabulary - thank you Mifid II. However, the pitfall is that we run the risk of being buried under an avalanche of data without knowing its value.
“There's money in them thar hills," writes Mark Twain in The American Claimant. Whilst this is a little bit of trivia to think about, there is indeed a lot of money in the hills of data. Like the miner prospecting for gold - a great deal of shoveling, sifting and sorting needs to happen before a strike.
How do we know this? Simple. A lot of the trading accounts either make money for us or pay us commission in the process, which we want to keep persistent and indeed grow.
So we try to dig through the data and build a picture of activity - trades, market prices and profit and loss perhaps. A bit like trying to understand why the best skier won the downhill by accounting for variables such as weather, snow, equipment and time of day, but ignoring how previous runs were executed and what the key elements of the risks taken were.
Trading is like the downhill skier’s course. Defined track, changing and varied terrain, temperature fluctuations, wind changes, skier fitness and so on. All of this is equivalent to direction of the trade(s), existing market conditions, market depth, data delivery, network speed, competition for the trade and existing account limits or profits. Monitoring all these different key values based on previously determined parameters in both situations gives the manager a sense of the next success or failure. But imagine what the change in outcomes could be if you modify the elements under your control based on a deeper understanding of the participant’s interaction with the environment.
Behavioral analytics is in almost every industry today - but we are only starting to see this employed in the trading business. From the independent trading company trying to understand what makes a successful trader, to the clearing firm that needs to be certain that it only has the ‘best’ customers. In the age of KYC, we may know who the customer is and whether he or she is qualified to hold an account and trade. But do we know whether we are taking on more risk than we'd like to?
Using data to understand the dynamics of the markets is already very much in the purview of the trader today. They are experts at designing trading strategies, tailored to behave as market actions require them to. What is starting to gain broad traction is the desire to have the ability to replicate this in customer management strategies, while staying aligned with the firm’s mission and objectives.
Similar to the function of an aircraft black box, post- Mifid financial markets are the ‘black boxes’ which contain more and more data. These are not only restricted to market data and trades, but also include data derived from orders, validations or rejections based on parameters enforced or monitored by systems. All of this builds a much more detailed and meaningful insight of the landscape, where the analysis of the users’ behaviour creates a safer, more precise and profitable paradigm.