Lets say I have a system - a system with an edge. One reason an EA might not be such a good idea is when a signal forms on one currency pair - it tends to happen on a few at the same time. For example, a short on AUDJPY may happen at the same time as a short on CADJPY, NZDJPY, NOKJPY. Opening all 4 sells is too much exposure to Japanese yen weakness. The other (real) reason is I'm bombarded with fundamental analysis which strongly influences which signals I take and which I don't. It's a technical system with a high degree of discretion. Obviously, with a risk:reward ratio of 1:1 my win rate needs to be greater than 50%. And I need to calculate position sizing to accommodate tick normalization on different currency pairs. It isn't uncommon to have losing trade - occasionally have a losing week, rarely have a losing month - never had a losing year. -yet. One of the main indicators used is a statistical indicator. When I started trading emotions were detrimental and took a while to identify. So I always understood the importance of letting trades play out, and never to close them prematurely. To never close out positions in profit before they hit tp - because that would obviously offset the statistics & change the whole r:r which is so important. But given the large discretionary premise of choosing which signals to take and which not to, would it not make some sense to close some profitable positions pre-maturity, based on the fact I may not have entered that position in the first place if some new piece of information changes my opinion which makes me regret opening the position in the first place (but i can close it now and get out with a profit but that would alter statistical r:r - and ultimately - win rate). Source.