hey nasher, looks like you bought a little bit of more RAM
I understand limping big hands to be able to play more crap hands that otherwise would have been folded. However, I wonder if the use of different opening sizes is due to non-convergence. For instance, a 4bb open doesn't make sense as we are commited even when knowing villain only reshoves aces. I'd imagine that an unabstracted nash strategy should only call, shove and maybe minraise (this might be dominated too, but not raise any bigger)?!?
Furthermore, the 8bb/100 difference is huge for such a small game. Does the base strategy know how to handle limps? Here (
http://www.daimi.au.dk/~bromille/Papers/aamaspoker.pdf) the authors did compute a similar game and report a minimal difference between PoF and unabstracted game.
Maybe one can test it without the issue of not adequatly handling limps this way:
case 1: your old strategy is always in the SB and PoF, your old strategy is always in the BB.
case 2: your new strategy is always in the SB and PoF, your old strategy is always in the BB.
Then one can compare the EV from the SB between old and new strategy.