[Gonso]

With only 62 hands over the 60bb range, I really don't think it's enough to evaluate. I mean most of the samples are decided by one hand...

Second, win % isn't necessarily equal to preflop AI equity, since a small number of those hands can go multiway. You should have a generally downward slope here with enough data as the effective stack sizes increase, and it's not really close there yet.

Next, you also have to factor in that there will be some money in the pot when it comes time to decide where to call the AI (for example with a standard raise, three-bet, shove line).

With some allowance for multiway and pot odds, you'd likely still be +EV even if you were only 40% win at showdown - although again I'm sure it's not going to be that low in a decent sized sample.