Are there any guidelines for minimizing selection bias when deciding on which assets and algos to include in a strategy? There's the temptation, of course, to select only those with the best backtest, but that bias already starts to overfit the strategy. So what to do? What things should one look for when assessing whether to include an asset?