'it depends'. not the answer you wanna hear, but true
for example, I have a few spreads I trade where I set super tight stops. they get hit more often than not, and my win rate is ugly. but my system targets outliers, and while theyre rare, when they happen theyre obscenely profitable and dwarf my tick sized losses. in this case, I follow the adage.
conversely, google proebstings paradox wrt mean reverting spreads. specifically, check the kelly chapter in Sinclair's book. sinclair (and the papers he references) argues that as your spread moves against you, you should be adding to your position to hold σ/2, as your increase in edge > your accruing losses. BUT, this applies only to a point. IIRC the derivation returned a maxima @ σ = √2, beyond which your accruing losses > your increase in edge