I'm looking for an interesting topic about Bootstrap Method. Have you some suggestions?

Performance measurement of some strategies. We used it with a friend. We developped a strategy where we took positions rarely (over 15 years, only 200 days). So, it is nearly impossible to compare the performance using usual tools (sharpe ratio with S&P500 benchmark for example). We used bootstrap in order to quantify the performance. The idea is to simulate entrance points for other strategies and to compare the performance. "What would have been my returns if I had taken other entrance points for my positions (you keep the lenght and the order of your positions, but you change the moment you enter in)). You run 20'000 simulations, compute the cumulative return, draw an histplot, place your strategy on the histplot. If it appears in an upper quantile, you can say that you overperform the others. You have really chosen the optimal entrance points.I hope it helps,

You can also try to generate more observations from a small sample. Imagine quarterly data. You don't have a lot of independent observations. So you can use bootstrap to generate more. Of course, it rely on certain assumptions....But it might be quite interesting...

Hehehe... you could try to figure out the exact value of kurtosis in finanical markets... hehehehe.... of course, that underlies certain assumptions.... *gnihihihi* <-- Evil

In my limited reading, I gathered bootstrapping, as a term, to be more generalizable. I have seen other dimensions/randomizations used while still being labeled as bootstrapping. For example, Alfred Cowles evaluated investment newsletters by comparing their picks to random stock sets. He did so in 1934, which obviously predates the well know dart contests in the WSJ. The whole point of the exercise seems to be to control for multiple factors in one fail swoop. Without controlling for these factors, one might be lulled into assuming good absolute performance is sufficient to accept or reject a trading strategy.Check out Brock, Lakonishok and LeBaron

For another application see discussions on filtered historical simulation as used by Barone-Adesi et al. I think there is literature on the glorimundi website. I can did out the references to published papers if you are interested.T