This is really confusing me. I first noticed that my results in Pio were overbetting the flop (SRP IP) in spots where GTOW doesn't overbet at all. I reproduced one of many of these spots in GTOW as closely as possible in Pio, expecting strategies to be very similar again after inputting perfectly matching ranges, stack sizes, etc, so I could figure it out from there. However, after solving to 0.5% accuracy, there were still differences, see below.


Here are the ranges from GTOW -- 100bb 6-max, BB calls UTG raise. I reproduced these very, very closely in Pio, but as I no longer have a subscription, couldn't just copy the range directly.



Here's the GTOW flop strategy for IP after BB checks -- note that there's virtually no overbetting, and the 1/3 pot sizing is the most frequent.



Here's the flop strategy for the same spot in Pio -- note that we're overbetting 11% of the time, and the 2/3 pot sizing is the most frequent.



Finally, here's the tree setup from Pio -- BB always checks, and IP has three sizing options. The starting pot I noticed after has an extra big blind instead of a sb, but this can't be that significant. I also figure that although in GTOW BB is sometimes donking, that absence of donking in the Pio tree should just slightly cause IP to check more, and perhaps favor a smaller sizing.

You would have to have all the settings be the same to have an objective comparison. The allowed bet sizings, raise sizings, the ranges themselves, rake, donk betting, etc. and it looks like even the effective stack depths are not equivalent between then two. You would also probably be better off solving to a higher accuracy (0.1%) before you compare.

Why doesn’t my solution match GTO Wizard?

I set the stack depths in Pio as 100bb in a 2/5 game, with a 2bb open ($10) as in the GTOW sim. They should be equivalent. The strategy differences seemed more drastic than I would have expected for the differences in allowed bet and raise sizings, rake, as well as the solved accuracy and so on.

Oh it is the same, but basically everything else was not... so... yeah.

edit* actually 4.5bb and you doing it as 2/5 (which Idk why you would do it like this) would be 22chips, you have yours as 25.

I ran your sim with the correct ranges, 10 chip BB, correct SPR, and 5% 4bb cap rake

Solved to 0.22%







Since BB flats so tight at 50nl rake they should have a donk range. Your sim basically node-locks them to never donk flop and you can see that decreases our nut and range advantage

Just wanted to add that it is also possible to have multiple equilibrium solutions. In some cases you can have multiple strategies that are rather different, but have equivalent EV.

Exactly.

From the link I posted: