i looked at $ndx gaps going back 5 years and playing the gap for a breakout is an almost dead-even wash. this is on an opening/closing basis and holds for both buying up-gaps and shorting down-gaps taken individually as well as both taken together. also did not find a significant variance related to the size of the gap, which really surprised me.
what i haven't yet done is looked at playing gap-fill failures for a breakout: ie, if no gap fill by 10:15 (or whatever) then place a bet in the direction of the gap and close at EOD.
imo that could be a very useful bit of information to have.