Applying evolutionary ideas to algorithmic trading
Gregory Fishman (pictured) is one of the managing partners of Automated Intelligence Systems, an independent scientific institute established in Saint Petersburg in 2006. The team, which comprises some of the top Russian minds in mathematics, physics and IT, takes its inspiration from nature. By studying complex ecosystems, and how they evolve, they have developed a sophisticated algorithm based on artificial intelligence to trade the financial markets.
Evolutionary theory, says Fishman, is “one of the methodologies we use to decide what is reasonable to research and what isn’t”. In a similar fashion to David Harding’s Winton Capital, which famously takes a rigorous scientific approach to the markets, AIS has applied its collective expertise to build, in effect, a sentient algorithm that evolves as market conditions change.
This has allowed Fishman to build an incredibly strong track record of performance, delivering significant returns due to the small amount of proprietary money that has, to date, been traded. The strategy – a high frequency trading platform – now trades with USD10million.
“We made some algorithms to do high frequency trading. The main algorithm is based on lots of maths and physics, a mesh of ideas. That’s our core strength: designing complex systems using a framework of maths, physics and IT. Each idea does something different, but they are all connected,” explains Fishman.
As mentioned, the strategy currently uses proprietary money to trade, although there are now plans to open it up to external investors. So successful has the algorithm been in recent times that last year AIS won a competition hosted by Micex RTS (Moscow stock exchange) as the best trading strategy, netting Fishman USD30,000 in prize money.
“Our daily trade turnover is about USD1billion, roughly 10 per cent of the Russian market turnover,” confirms Fishman. “Between September and December 2011 we publicly made 8000 per cent.”
The team at AIS essentially views the financial markets as an ecosystem; an organism that self-regulates, adapts and evolves. It’s the ability to find patterns in nature that has helped build the algorithm, which uses pattern recognition to trade.
“Yes, we view the financial markets as if it were any other ecosystem. If you go to a forest you’ll find an ecosystem and you can build lots of parameters to model it: we’re applying a similar approach to the stock markets. Every idea we develop is based on evolutionary systems. It’s basically artificial intelligence: our algorithm is like a robot that can trade the markets and think for itself,” explains Fishman.
The strategy trades Russian stocks, derivatives, the FX spot market, and also trades Russian paper on the London Stock Exchange. Fishman believes that once the strategy is established into a hedge fund structure it will have an initial short-term capacity of USD500million.
For any HFT platform, speed is important but as Fishman points out: “We have co-locations in Moscow, London, Frankfurt, New York and we also have partners in Chicago. Latency is important but is not critical. The algorithm itself is the critical point. This is the competitive edge that we have.”
- By Category
- News from other sites
- Special Reports
- Partner events