This post was written in November 2015 when I was Head of Poker at Microgaming.
Many years ago I wrote for a magazine called InsidePoker Business. It was edited by Alun Bowden (@Poker_Hack), and fellow contributors included Kim Lund (@InfiniteEdgeKim), who now runs his own consultancy business, and Barry Carter (@Barry_Carter), who is now a freelance journalist.
It’s become something of a running joke that we were writing articles back then about topics which the industry is only beginning to address now. As an example, check out this article I wrote back in 2011, where among other things I mention the ‘English only’ chat rule and why it needs to die in a grease fire. One of the first things I did as Head of Poker at Microgaming was to eliminate this rule, but that didn’t happen until May 2014, and as far as I know we’re still the only network where you can speak any language you like.
In another delightful example of our collective foresight, Kim Lund wrote an article called End the Spread about cash game buy-in sizes back in 2013. One particular quote stands out to me:
I’d argue that those in the know either min buy-in to try and hit-and-ruin the game or they buy-in max to make sure they’re not at a disadvantage. All other possible choices are moot.
At the time, we didn’t have the data available to test his theories, but now we do. So we set out to determine, once and for all:
- What buy-in amounts people prefer
- How this influences their likelihood of winning
- Which players are more valuable
- Whether we should change the buy-ins – either increase them, decrease them, or do as Kim suggests and take the choice away entirely
The conclusions are interesting, at least if you’re a poker geek like me.
What We Did
First I tried to analyse a couple of million rows of data in 32-Bit Microsoft Excel. It was a painful few hours before I realised this was a bad idea, and upgraded to the 64-Bit version. It was an even more painful few hours before I realised I was better off getting our Business Intelligence experts to do this properly…
Together, we looked at 2,273,022 actual buy-ins on the MPN which took place earlier this year, before we introduced the Grouped Lobby feature.
We filtered out non-Euro games, and the €0.01/€0.02 stake, because this is the only stake level on the MPN where it’s possible to buy in for less than 30BB. We also filtered out auto top-ups, and looked only at the first buy-in to the table (almost all auto top-ups are to 100BB).
At this early stage, we discovered that the players who perform the worst are those who have no discernible strategy when buying in. For example, if you take between 80 and 89.99BB to the table, chances are that it’s the last bit of money in your account. You’re probably on tilt and playing badly, and hence players who buy-in for this amount lose the most money per buy-in of any group.
To ensure that our data wasn’t skewed by players playing just a couple of times and blowing off their last few Euro, we focused only on players who had made 10 or more buy-ins, and who tended to buy in for the same amount, i.e. players whose buy-in was in the same category 60% or more of the time. About 57% of players who buy-in more than 10 times have a clear preference like this.
The Results
A whopping 85% of cash game buy-ins are for exactly 100 big blinds, which is the maximum in the majority of our games. The next most popular buy-in level is exactly 30 big blinds, the minimum in most of our games, with 8%. These are followed by exactly 40BB (3%) and exactly 50BB (2%). There is clearly a strong preference for 100BB play – only 2.5% of players have never bought in for 100BB.
Players who buy in for 100BB are also the most likely to be winners – 47% of players who buy in for 100BB are net winners after rake. 44% of those who buy in for 30BB are winners.
That might seem close, but it isn’t because the amounts won and lost are significantly different. 100BB players make up 82% of players, but win 92% of the money. 30BB players make up 11% of players, but win just 4% of the money. A 30BB player loses over 6 times as much in their lifetime as a 100BB player, and 14 times as much per hand played (100BB players play more hands). This is despite the fact that they are much tighter, putting 56% less into the pot than 100BB players.
There is definitely no evidence here to suggest that 30BB players have an advantage overall. I’m fully aware that in poker theory, there are situations where having a 30BB stack should give you an advantage – I read Rolf Slotboom’s Pot Limit Omaha book after all. But there is no indication that this is actually playing out in practice on the MPN.
So far, all of this could be explained by saying that 30BB players are predominantly casual, and hence more likely to lose. There is lots of evidence, which I won’t share here, that shows 30BB players to be more casual than other groups. Similarly, you can argue that the most skilled players tend to buy-in for 100BB, and so players at this level are more likely to win. So far, we’ve only demonstrated a correlation, and not causation.
But what if you take a 30BB player, and make them buy in for 100BB?
We introduced our Grouped Lobby feature in April this year, and as a by-product we moved some players from the 30BB category to the 100BB category. That’s because the Grouped Lobby has a ‘Play’ button which automatically seats the player with 100BB, and this is the default interface for all players. At the same time, for unknown reasons, a small number of players went from buying in for 100BB to 30BB (perhaps they read Rolf’s book). We were able to pinpoint those players that changed from 30BB to 100BB, and vice versa, and see the impact on their stats.
The group of players who moved from 30BB to 100BB immediately showed better results. In fact, they went from losing money on average, to actually winning money on average. Their P/L improved by a factor of 3, after rake.
Players who moved from 100BB to 30BB immediately showed worse results. They went from winning money on average, to losing money on average, and their P/L declined by a factor of 2.5.
What’s more, 30BB players pay more rake per hand played. This is probably because they are so often all-in before the flop, as this is a key part of the shortstacking strategy. However, because they lose more quickly, their lifetimes are shorter than 100BB players and they are worth less to a poker room.
This is quite conclusive evidence that the buy-in is a causal factor in determining whether players win or lose, and it suggests that 30BB players might have a better experience if they were to buy in for 100BB instead.
The Experiment
Since a lot of players hate to play against shortstacking opponents, and it appears that players who buy-in for 30BB would actually have a better time if they played for 100BB, we’re going to try an experiment.
On December 1st, we will change all Omaha Hi/Lo tables on the MPN to only allow buy-ins of exactly 100BB. We will carefully monitor the impact, and if it is neutral or positive we will then roll out this change, gradually, to Omaha Hi and eventually to Hold’em. If at any point the impact is negative, then we will revert to the current setup.
One side-effect of this change, which we’re already seeing to some extent because of the Grouped Lobby, is that it will force players to manage their bankroll better, and play at tables they can better afford. If you’ve deposited €50, the highest NL/PL stake that you’ll be able to play will be €0.25/€0.50 – you won’t be able to play at €0.50/€1 as you could previously. Our data shows that this will make your money last longer, and suggests that you’ll have a better time.
If all this works out, we’ll have another question to answer – is 100BB the optimal buy-in amount after all? But that question can wait until later.