Why your last 5 trades are useless as feedback (and what to measure instead)

Took me some time to get this: you can't judge a setup off five trades. A breakout that wins 55% of the time at 1.5R can still hand you 3 losers out of 5 and look dead. Two things that fixed it for me: stop grading trades by P&L (a loss from a properly sized, stop-defined trade is a good trade that just caught variance), and be the house not the gambler (your edge shows up across hundreds of trades, not the next one)

How many trades do you give a setup before you call it broken?