Within each strategy find some criterion (if possible) to rank your potential candidates such that the ones with expected higher avg profit are at the top of the ranked list. Select only the top 10 of each strategy to trade everyday. Initially, divide capital among all strategies equally, and divide the capital allocated to each strategy among the top 10 ranked stocks for that strategy. Track the pnl of each strategy. If a strategy makes $x in a week or month, then increase the total allocated to that strategy by some amount. If a strategy loses $x, reduce the amount allocated for that strategy.
yes, that's what i have been doing, i.e. selecting a limited number from all the trades triggered - my concern has been that effectively, i am re-sampling in actual trading, i.e. i am taking a sample from my system's generated sample... it clearly changes probabilities, but it seems that it doesn't impact performance in any dramatic way. i am thinking that a better solution lies along the lines of integrating correlations between the triggered stocks, i.e. if highly correlated stocks are triggered, i'd reduce their individual weights. i have not been able to analyze correlations in depth, only to the extent i need to use them for hedging purposes. thanks for your feedback and all the best.
If you can't "optimize" your stratgy to pick the best 10, then take capital/ number of trades generated today. full capital usage without losing trades.