In a previous post, I discussed the difference between two types of market trading systems: predator and prey-type. The key difference between the two types as I defined them is that predatory trading systems induce changes in the market, and prey-type trading systems react to the history of the market. In other words, prey-type trading systems hope to profit from past price patterns, and predatory trading systems hope to profit from price patterns they create. Prey-type systems hope for infinite liquidity, predatory hope for finite liquidity. Just like in the real world, both predatory and prey-type trading systems can "win," but predatory type systems, to the extent they exist at all, are harder to develop for the reasons I previously explained.
How would one develop a predatory trading system? <pure speculation ahead> First, you must have sufficient cash to move the markets and induce the changes in the market. Second, you would have to develop a theory of the behavior you induce and would need to put real money on the line in a real market to know if your theory works. Maybe you get your theories from the field of behavioral finance. You probably can't just dip your toe in, because after all, you must induce changes in the market. You can't likely test thousands of theories either, you have to test a select few, because this isn't fake money that can be lost. On the bright side, there's probably less chance of experiencing the pitfalls of datasnooping than with prey-type systems for those developing them.
Third, backtesting, if used at all, probably can't simply rely on the typical price bars, tick and volume data. You would need to have all the market microstructure data. With such data (see also here), you could make an informed guess with regard to your initial impact on the markets by examining liquidity in the limit order book, but from there on, it's increasingly guesswork. You would then need to know how the microstructure of the market works and how liquidity reacts to your actions. Predatory backtests, if used at all, would be based on the assumption of finite, morphing liquidity. This means that a predatory system must adapt and evolve over time to changing liquidity and price patterns, because if they don't, pretty soon the prey is going to be them.
It doesn't sound easy, but perhaps the rewards for predatory trading systems are rich.
Comments