A Tale of Two AIs
The trouble with developing ‘artificially intelligent’ trading systems is that a mass market for the technology has yet to develop.
“Potential clients that might want to use a system like ours usually would want it on an exclusive basis,” said Guillaume Vidal, the CEO of Paris-based startup Walnut Algorithms. “They would want to own us or buy us out.”
San Francisco-Based Tech Trader also found a similar environment when it launched in 2012, according to founder William Mok.
Both firms found it was much easier to rely on the trading revenue generated by their AIs to fund their businesses.
“We view ourselves as a technology company first,” said Vidal. “We use the latest advancements in AI to solve very complex problems.”
Except for setting up their own hedge funds, the two funds approach how, what, and when they trade from opposite directions.
When Walnut’s London-registered hedge fund starts making cash trades later this summer, it will be trading equity-index futures on the Chicago Mercantile Exchange via Interactive Brokers and will hold positions for approximately two to four hours, said Vidal. “We’re fully liquid overnight.”
Liquidity was the primary consideration behind the choice of that asset class, Vidal explained. “We wanted something that you could come out of very easily. If you trade single stocks, some of them might be illiquid and you would have to wait a few seconds to exit the position.”
Tech Trader trades cash equities and exchange-traded funds with a minimum of $1 billion in market capitalization via ConvergEx.
“That is the only constraint as stock selection goes,” explained Mok. “That’s mostly just for the hedge fund. The AI can work with the level of liquidity.”
Tech Trader isn’t buy-and-hold, but neither is the firm among the shortest-term market participants. Holding periods “could range from a day to two years, but on average it is about two or three months,” Mok said.
Neither fund started developing their AIs to mirror specific trading strategies or trading schools.
Walnut’s AI, which has been in development for two and a half years, uses algorithms to identify and exploit individual market configurations that the AI has learned through analyzing 10 years’ of market data.
“I wouldn’t call them visible patterns in the sense that they arise from a number of different data points,” said Vidal. “We look at about 2,000 to 3,000 data points per second, which helps the machine describe the market in a series of snapshot. The algos then say ‘I know this configuration and there’s a strong likelihood that the market will overshoot’.”
Vidal describes Walnut’s trading strategy as the aggregation of all results from its trading algorithms.
Meanwhile, Tech Trader designed its AI uses deductive logic to determine its trading decisions.
The AI mirror the thought processes of a child or an animal, said Mok. “It’s not that smart, but it will do whatever it takes to survive in its environment.”
Instead of recognizing market patterns from the analysis of historical data, Tech Trader’s AI runs a series of what-if calculations in parallel with its analysis of current market data.
“All these quant and machine-learning guys crunch numbers to determine probabilities, whereas our AI does not process that much data,” he said. “It’s more like ‘smart data’ instead of ‘big data’. It uses very little data, but what it does use, it uses very intuitively.”
Both funds are looking for investors. Tech Trader has $7 million under management with another $20 million that is committed for later this year.
Featured image by Yuichiro Chino/Dollar Photo Club
Data extraction and integration is the second stage of a digitization process.
Increased electronification has created useable and accessible real-time and historic trade data.
The fintech uses data so institutions can assess the environment impact of their portfolios.
Working with Riskfuel will reduce reliance on slow and expensive financial models.
The platform uses AI to match experts and users and digitize the research process.