It is customary to believe that main source of market pricing inefficiency are emotions of discretionary traders.
But I think that most inefficiency today, at least on shorter time-frames, is created by actions of "naive" systematic traders, implementing excessively simple and straightforward approaches. They over-backtest, over-fit, and then over-invest into whatever they found. Then at some point they get ruined by the market game logic, creating shocks, jumps and other kinds of impact sophisticated traders love to exploit.
The problem with the systematic approaches is that they seem so scientific and reliable, that you just can't find enough reasons to doubt the promise you see behind them.
But they are not in fact as reliable as it seems, because the market is not a physical object with fixed properties. It is an interactive game, which adapts so the less sophisticated participants loose money (mostly to transaction costs, though). It doesn't really make a difference are those participants discretionary or systematic, if they are not good enough to adapt faster than the environment changes or see the gameplay deeper than the average participant, they pay.
So if you search for the market inefficiency spots, you may start with a question: "what would a simple minded systematic trader NOT do?"
Another consequence is that this situation is probably not going to change as more and more technology comes into the gameplay.
We all expect the AI to come to the market. The common idea is that it makes all the market inefficiency disappear, eliminating traders as a class.
But again, there will be no AI not linked with the particular person making some decisions about that AI. That means we will have over-simplistic approaches all over the place and over-investments into whatever AI may seem to promise. Which will be breaking at some point, creating shocks and jumps.
So inefficiency is here to stay as long as we have at least some humans involved in the gameplay.
But I think that most inefficiency today, at least on shorter time-frames, is created by actions of "naive" systematic traders, implementing excessively simple and straightforward approaches. They over-backtest, over-fit, and then over-invest into whatever they found. Then at some point they get ruined by the market game logic, creating shocks, jumps and other kinds of impact sophisticated traders love to exploit.
The problem with the systematic approaches is that they seem so scientific and reliable, that you just can't find enough reasons to doubt the promise you see behind them.
But they are not in fact as reliable as it seems, because the market is not a physical object with fixed properties. It is an interactive game, which adapts so the less sophisticated participants loose money (mostly to transaction costs, though). It doesn't really make a difference are those participants discretionary or systematic, if they are not good enough to adapt faster than the environment changes or see the gameplay deeper than the average participant, they pay.
So if you search for the market inefficiency spots, you may start with a question: "what would a simple minded systematic trader NOT do?"
Another consequence is that this situation is probably not going to change as more and more technology comes into the gameplay.
We all expect the AI to come to the market. The common idea is that it makes all the market inefficiency disappear, eliminating traders as a class.
But again, there will be no AI not linked with the particular person making some decisions about that AI. That means we will have over-simplistic approaches all over the place and over-investments into whatever AI may seem to promise. Which will be breaking at some point, creating shocks and jumps.
So inefficiency is here to stay as long as we have at least some humans involved in the gameplay.
No comments:
Post a Comment