Thanks for your comments, Yuval; I also found this thread response to be exceptionally helpful in case someone else goes down this rabbit hole in the future: Which ranking systems do you use? - #10 by chris319 Look for test_user’s responses, which are excellent to identify curve fit problems.
For reference:
- My Rolling tests look great, with very small rules and mostly Ranking systems.
- My subuniverse tests look decent Mod(StockID,5) if I account for Rank > x due to the smaller universe.
- I did not trim for outliers in the universe, but I did add minimum prices to the universe.
- I’ll be running 2001-2010 in Feb sometime due to increasing my subscription.
- Short is an open trade study from a Puts/Shorts perspective, Market Timing Problem, and Potential to Hedge with TWM. Again, I’m looking for the Greatest Alpha with the Lowest Drawdown that is tradeable and has similar performance OOS.
- Finally, as listed above, cutting the top 50 stocks from the performance hampers performance significantly but is still above the benchmark. As such, I’m inclined to believe it’s still okay. Thoughts? I think I read somewhere that the top 1% accounts for ALL the alpha, but perhaps I’m misstating the quote.
Thanks to everyone for your responses; they’re very much appreciated!
Tyler