How you exit matters as much as how you enter. This experiment compared a single full take-profit against scaling out — and the result reframed how we measure a strategy entirely.
Two exit models on the same entries. Model A: full close at a single target. Model B: close 50% at +1R (banking a partial and de-risking), then run the remaining half to +2.5R.
Over 6.4 months Model A made +$0.1k; Model B made +$2.2k — despite Model B having a lower headline win rate. The runners paid for all the small give-backs and then some.
Banking half at +1R turns most trades break-even-or-better early, which removes the emotional pressure and the drawdown from full reversals. The runner then captures the occasional large trend move that a fixed target would have cut short. Asymmetric payoffs come from letting winners run, not from being right more often.
This is the default exit in the MIDAS EA. The experiment is also the clearest demonstration of the course's core discipline: judge a change by the money it makes over a real sample, not by the win-rate number that feels better.