Most important is to find a way to evaluate entries without taking into considering the closing transaction. Compare those entries with random entries. I scatterplot MFE against MAE from entry time to the end of the day for day trades (end of the week or month for swing trades). Once you find entries with a considerable edge you can find some reasonable ways to get out. For example, If 80% of your good entries are +5 ticks never to return in 15 minutes, you could have a rule to move your stop to +4 after 15 minutes if it is +5 or better, and get out otherwise. etc. etc. etc. Another important feature is to plot your live entries in real time against what the are supposedly supposed to do, and continue to measure them with your entry metric against random. This is how you know if you screwed up your research or when the market changes.