Algos: Human After All?

What defines a human? If it is the ability to think autonomously and to evolve to the surrounding environment, then it is not so difficult to draw comparisons between us Homo sapiens and the latest trading algorithms.

Trading firms in many cases prefer computerized algorithms to human traders, but algorithms are beginning to look more human every day—not least through the mistakes they are prone to make.

Algorithms have had a lot of bad press in recent times, but for Andrew Saunders, managing director of software provider Intalus, a well-designed algorithm is less likely to make poor decisions than a fund manager, providing that it is well-constructed.

“Algorithms base their decisions purely on data, and as long as that data is correct and the algorithm has been well-designed, then I don’t see any reason why it should go wrong,” he says. “What it could potentially do is start losing money if the core market’s fundamentals change, but a well-designed trading strategy should include not only an algorithm for finding trades; it should include an algorithm for coming out of positions.”

Customizable
Gerry Turner, executive director at connectivity provider Object Trading, says the rise of specialist brokers is adding value to the market by introducing a new range of algorithms that are customizable to the needs of buy-side firms.

“Specialist brokers are looking to make algos operate more like humans, rather than be very rigid across all markets. Algos now understand more of what is going on in the market … and this will definitely add to the quality of the price that’s out there.” —Gerry Turner, Object Trading

“Specialist brokers are looking to make algos operate and make decisions more like humans, rather than be very rigid across all markets,” he says. “Algos now behave and understand more of what is going on in the market, like the way the market behaved when I was a boy and it was all human interaction, and this will definitely add to the quality of the price that’s out there.”

One of the driving factors in the push for customizable algorithms, according to William Capuzzi, president of ConvergEx Group’s global execution business, is an increasing desire for transparency across the industry, whereby investors are asking their fund managers to explain the reasoning behind their investment decisions.

“Clients want to better understand how their algorithm is trading,” he says. “It’s not good enough anymore for someone to put in a volume-weighted average price (VWAP) trade and just get fills and understand where they stood versus VWAP over the course of their order. They actually want to understand how the slices are getting executed, where they’re getting executed and what the motive of the broker is to execute in that venue.”

In Demand
The demand for algorithms shows no sign of slowing down, according to Capuzzi, both in terms of the number of firms that are using algorithms and the percentage of total flow in which algorithms are used.

While overall volumes have continued to fall over the past year, he says demand for algorithms continues to rise, despite a lack of notable additions to the pool of benchmark algos.

“The general trends around algorithms are focused more around getting smarter, improving performance, improving client workflow and finding liquidity,” he says. “There’s nothing new in terms of an actual benchmark algorithm, but the focus now is really around improving performance against existing benchmarks and improving the liquidity that I’m accessing and trading against.”

This is a trend also noticed by Philipp Kahler, senior quantitative analyst at Intalus, who likens the switch from manual to automated trading to the way machine-led transportation has taken over from horses and carts over the past century.