I’ve heard about these clever opportunistic algos… What’s so ‘smart’ about them?
Smarter adaptive algos have been emerging over the past couple of years, differing substantially from the previous generation of strategies which broadly planned the trade in advance and then kept to the plan.
Brian Schwieger, head of EMEA algorithmic execution at Bank of America Merrill Lynch, explained to me that an adaptive algo can put the plan aside, depending on the conditions it finds itself in and the signals it is receiving.
Effectively, the algo can say, ‘wait… something different is happening here and I don’t think the trade plan is appropriate for the moment’. A good adaptive algo will slow down or speed up or change tactics as it sees fit.
“What we are seeing in action is the next generation of algorithmic intelligence,” says Schwieger. “Algos are being developed which behave closer to the way that a human trader trades, only with much faster reaction times and often greater sensitivity.”
Isn’t that how algos are supposed to behave?
Up until now, a lot of algorithms have been able to use an intelligent plan but they couldn’t shift from the plan in a dynamically responsive way. The latest adaptive algos say, ‘while I’m happy to have the plan, I will put it to one side if market conditions do not develop as expected’.
And while in the past, using algorithms required a fair amount of knowledge and skill to master, adaptive algorithms are becoming intelligent enough that a trader doesn’t have to micro-manage them.
“These are true productivity tools which you can turn on and turn away from with confidence that they will get the job done without constant monitoring,” says Schwieger. “Productivity gains also come from the fact that if you possess a truly adaptive algo, it reduces the number of algos you need in the toolbox.”
Much of the customisation that has taken place in recent years stems from the last generation of algorithms needing tweaks to work the way you want them to.
Take implementation shortfall algorithms. Because many of these weren’t adaptive, they were often customised to deal with the different tactics and styles required to trade in different markets and market caps. An adaptive algo adjusts to these differences seamlessly, reducing the need for customisations.
But aren’t opportunistic algos simply aggressive algos?
Algorithms that aggressively hunt liquidity will always have a certain level of popularity with traders, but that’s just one of uses of smarter, opportunistic tools.
“Aggressive algos are a great alternative to market orders because they use prevailing market conditions and signals to grab liquidity at the most optimal time, wherever it may be,” says Schwieger. “These algos are able to respond to high-frequency trading tactics and shifts in liquidity far more quickly than a human trader.”
Should I always be aggressive, I mean, opportunistic in volatile markets?
Schwieger explains that when markets are volatile, traders tend to go in one of two directions – VWAP or aggressive. VWAP feels safe because it takes the average price weighted by volume over the course of the day, smoothing out the volatility.
At the other end of the spectrum, some traders decide that they like the price they see on the screen and will want to get done immediately.
“We often find that clients trading large orders in volatile markets will split a large slice of the order into a VWAP algorithm and then optimise performance by picking their moments to send smaller slices into the market using the more aggressive liquidity seeking algos,” says Schwieger.
As markets continue to become more complex and fragmented, and brokers continue to develop their algorithmic strategies, the pace of algo development suggests buy-side can expect to continue to benefit from improved performance and productivity over the next year.