Hey everyone, I've been working on a news feed lately, and I think this is an appropriate place to tell about it. I collect the latest news, whitepapers, and blog posts that would be of interest to systematic traders. It's a curated list, so you won't waste your time wading through dozens of unrelated SSRN papers every day. Let me know of any good blogs or news sources I should start monitoring. http://www.quantnews.com
This is good stuff that will set a new standard of thought for system traders-- . https://faculty.fuqua.duke.edu/~cha...Papers/P116_Evaluating_trading_strategies.pdf
Thanks, I just added a link to it. The topic of multiple testing is important to me. Two days ago, I had my qualifying exam, and part of my proposal included some ideas I have about incorporating randomization/permutation testing into data mining.
David Bailey has a related paper out, but with a more provocative title: Pseudo-Mathematics and Financial Charlatanism: The Effects of Backtest Overfitting on Out-of-Sample Performance. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2308659
Here is issue in terms of curve fitting that no one has discussed here, that is the premise of the system. The theory behind it, for example many of my intermarket systems. If we say utility stocks can predict T-Bonds for example that system is less likely for curve fitting than a triple moving average crossover. Another example would be a S&P500 model which uses PE ratio. Yes these correlation can decouple but if they make sense , they should in general hold up. Curve fitting in the case of these types of system will make them have flat out of sample equity curves or slow bleed, they don't normally blow up. Decoupling can cause a bad year but then the system will should come back.