I wonder why White's BRC only determines whether the best trading strategy is statistically profitable.

What prevents us from comparing the average V of the second best strategy (i.e. square root of the number of observation multiplied by the mean of the strategy) and comparing it with the distribution obtained by using White to obtain the pvalue (just as we do with the best strategy)? Similarly, why can't we do the same with the third best strategy, and so on?


  • $\begingroup$ Hedge funds examine a large number of possible strategies on paper, then (being profit maximizers) select the best one (in terms of P&L, Sharpe ratio or other criteria) for actual implementation with real money. The White RCB is designed to test whether this best strategy is statistically significant or not.. In any case note that if the first strategy is not stat significant, the second one will also be not stat significant and should also be rejected. If the 1st and 2d are stat significant then use the 1st one since it makes more money. $\endgroup$
    – nbbo2
    Aug 6, 2022 at 15:19
  • $\begingroup$ But is it statistically correct to check the statiscal significance of the second best, third best, etc strategies using the same approach ? So computing the Pvalue of the second best strategies using its V ? For diversification purpose, it could be nice to have more than one strategy in the portfolio for diversification purpose. $\endgroup$ Aug 6, 2022 at 18:32


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.