We’ve been working for several weeks now to make some changes to the MTT virtual opponents (bots) here on OPT. The goal being to make them play as realistically as possible in some MTT situations that come up all the time. I want to highlight five of those changes in this article.
1. Smaller bet sizes pre-flop
In a typical PLO cash game, most pre-flop raises are roughly the size of the pot. As you get deeper into an MTT, the pre-flop raise sizing tends to decrease (it takes a smaller raise to get the job done and pick up the blinds, as players are trying to conserve chips and survive). As you get deeper into a tournament, you’ll notice our bots start to do this too.
2. All-in Raising when short stacked
When one of your opponents has a short stack, they will go all-in if possible instead of making a smaller sized raise, as would be correct in a real MTT. This usually occurs later in a tournament when one of them has, say, only 3.5 big blinds left. There’s no point in raising a “typical” 2.5x when that would just leave him with 1 big blind. It also can occur when 3-betting pre-flop.
3. Bubble Play
On the bubble, your virtual opponents will now adjust their play based on stack size. The shorter stacks will play fewer hands pre-flop, up to 50% fewer depending on how short they are, trying to survive to make the money. The big stack will shift into “bully” mode and play more starting hands, and more aggressively, against shorter stacks.
For the technically curious, we are defining the bubble as 10% more players than the number that get paid, rounded up. So if a tournament pays the top 27, the bubble would be 30 players or fewer.
4. Hand Selection when short stacked
This part still needs some adjustment, but we’re working on their hand selection pre-flop when short stacked. Hold’em players understand that something like 7-6 of diamonds is a fine hand to play deep stacked, but when you only have, say, 10 big blinds left, you’d much prefer a starting hand with high cards.
The only problem is, in PLO it’s more complicated because pre-flop equities run much more closely together. For example, a hand like 9d-8d-7s-6s (the PLO equivalent of a “suited connector”) actually plays quite well against a wide range of hands when all-in pre-flop. So we’re continuing to research this.
5. Reduction of pre-flop limping
While limping some hands at the start of a tournament when the stacks are deep is ok, later in a tournament our bots on the Easy skill level were limping way too much. We are now slowing dialing this back as the blinds get higher. It seems like even beginners understand that these days.
The main exception is in blind-vs-blind situations. When it is folded around to the small blind, limping (completing the blind) is actually part of the GTO strategy.
Most of these changes were suggested by our users who pointed them out to us. We’re always happy to get more suggestions, just fill out our contact form.