Let’s say I have 10 strategies, and all of them are sort of useful sometimes in history and sometimes not. My questions is: how can I construct a backtest to inform me that which one of my strategies will more likely to win in the foreseeable future. Any hint or reference is welcome.
Maybe let me try this: If I find a way to pick the strategy from the candidate pool that works for one week and it also works for the next following week. Is this going to work if I apply it in the future ? Is there any logic flaw here?
Your evaluation period is way too short for statistical significance. High frequency traders (avg holding period: a few seconds) expect to use at least a month or two of testing to find a promising strategy. Traders with holding periods of around a month need decades of data. You are probably somewhere in between these two extremes. GAT
Usually in the case of a multi-day Trading System, tests of at least ten years of data are carried out (In sample), and then other tests are carried out in subsequent periods (Out of sample) to check its robustness. Usually tests are also performed with the Walk Forward Analysys and with the Montecarlo tests, after which you could continue to test the strategy for a few months with the so-called paper trading; if all goes well, then you can start trading with real money. This is the pattern published in a book by a trading professional:
Hi, there are excellent books to explore the valuation of Trading Systems: - The Evaluation and Optimization of Trading Strategies (Robert Pardo) - Trading Systems 2nd edition: A new approach to system development and portfolio optimisation (E. Tomasini-U. Jaekle) - Trading System that Work (Thomas Stridsman) - The Ultimate Trading Guide (G. Pruitt, J. Hill)
Try to google something related to factor momentum: https://alphaarchitect.com/2019/06/03/is-factor-momentum-really-everywhere/ What does it mean? Research suggests that there is a short term continuation/momentum in a performance of strategies which had the best performance in the past. So if you have multiple strategies, that you are comfortable with, you can try to rotate among them. But, it's not a 100% solution. You may have a set of strategies when it is better to diversify among all of them and not rotate and chase performance - check second chart (and research paper related to it) in the following article: https://quantpedia.com/how-do-investment-strategies-perform-after-publication/.