Converting ev into bb/100
Posted by Trinity
Posted by
Trinity
posted in
Low Stakes
Converting ev into bb/100
Some coaches like Uri Peleg like to convert EV as shown in a solver into bb/100 and implicitly comparing this number with a typical overall winrate in an attempt to stress how important some exploits can be.
Example: He compares a simplified strategy with a complex one, see the ev difference is 0.04 bb and he says this gap corresponds to 4bb/100, which is correct in the sense that every 100 hands we get into that situaion and pick the simplified strategy we are losing 4bb of ev if we're playing against the solver.
But we're getting into that situation just a small fraction of the times, so my question is: Is this conversion of any use?
I mean, we're folding around 75% of hands preflop, and an overall winrate includes every single of them. So my intuition says that comparing the two is misleading.
Loading 7 Comments...
If I am reading this correctly, you are saying the following:
He runs a GTO sim with several bet sizes on all streets
Then, he runs a simplified sim with, say, one bet size on all streets
He looks at them side by side
The GTO solution, on the flop node, the entire range in question has an ev of 4.04
The simplified sim, on the flop node, the entire range in questions has an ev of 4.00
The "range bet" strategy began because someone realized just how little ev it costs them to play that strat vs dwelling on a strategy where the complication just is not worth it. If the two ev's were more like 4.04 (GTO) and the simplified strategy had an ev of 3.2 (simplified). The strategy would be horrible.
For all intents and purposes, the simplified strategy in the first example costs us about 1% of the pot. This is miniscule in the grand scheme of things.
From what I have seen, coaches will often compare a GTO sim next to a simplified sim, look at the ev at the node in question (it is often the c bet opportunity IP). They look at range equity and establish if IP has range advantage and then they typically compare the ev's.
I do agree that it can be misleading. The good thing is that the simplification is usually going to over perform since a human is going to come nowhere close to how a solver defends.
I think most coaches use a 3 to 5% threshold between two ev's. The advocate simplification when under the threshold and would lean more towards more mixing when over the threshold.
The comparison i mentioned was not about the 2 sims, it was about comparing the ev gap (0.04bb = 4bb/100) to the overall winrate. You do agree that a 0.04bb of ev is minuscle, but of course a 4bb/100 in a player's winrate would be massive. So it's not like picking one strat over the other is causing a 4bb/100 differnece in the player's winrate. This was my doubt since I witnessed Uri Peleg make this conversion multiple times.
Was there a specific video he discussed this in? I am curious about it now :)
It was his course on Upswing, and maybe some of his youtube videos
Ahhh Ok I have that. I a planning to rewatch it again. I will look for that concept in there.
EV in solvers is displayed in "big blinds per hand". You multiply by 100 to get bb/100. This is the expected value of some strategy given the assumption that you've run into this spot.
Yes, the impact of that one spot would be minimal on your overall win rate. But consider what happens when you repeat that experiment for every possible spot.
One thing to be cautious of is that most sims are not run to high accuracy. For example, if you solve to 0.5% of the pot, and the starting pot is 6bb, then your sim is only accurate to +- 3 bb/100. A lot of people try these EV comparisons and neglect to solve to an adequate accuracy.
Thanks tombos, i love your work for Gtowizard
Be the first to add a comment