Quote:
Originally Posted by Monorail
Can you expand on the above, in general terms if need be? You don't have to give specific #'s here. But please do give an example of something that you guys tried to convey to them that they simply did not understand (or did not make an effort to understand.) Or something that they presented that, to use your words, you knew not to be true.
Those are important observations, but understand that they're rather difficult to weigh for anyone who wasn't there. There must be a way for you to further unpack those statements without disclosing the sensitive numbers that you can't talk about.
I've been thinking about how to answer your question and I think it's best to do it through a hypothetical example.
DISCLAIMER: I do not know advanced statistics or data analysis methods. The kind of stuff we are dealing with is complex and proper analysis goes beyond my understanding - so don't ask me what I think the proper way to analyse the result end of things is, because I don't know.
Having said that, the knowledge that I have is still considerable and I can identify when things are off, even if I do not know how to fix what is off (it will be more clear what I'm talking about after you read my whole post).
------------
One of the most important things with regards to rake changes (or cutting VIP rewards, which is the same thing) is determining how much PS makes from running certain games, and how much the winners win (take out) of these games. This is particularly important in high stakes games, because high stakes games need to be beatable in order to run. The argument that games will still run even if there are no long-term winners does not apply to high stakes. One of the reasons is centered around liquidity: you can't find enough net depositors to fill a 25/50 table or make a 1k SNG go off. And even if you did, what would happen if you forced an all-rec table is that some of the had-been recs are now winners.
Figuring out if a game is beatable, by whom, and for how much is hard, especially when the only thing you have access to is results.
But let's take a look from the other end of things through and example.
Suppose you are dealing with a 1kNL ecosystem.
From this ecosystem, you pick a subset of players which consists of 100 regs of equal skill. For simplicity, assume the games are rake free to begin with.
These regulars all have an expected winrate of 1bb/100. Their standard deviation is 100bb/100.
They all go ahead and play 500k hands of 5/10 each.
Naturally, the following happens:
- On average, they win $50k each
- About 1 in 4 are down money
- About 1 in 4 are up over 100k
Now, if you're dealing only with the result end of things, things kind of get messy, because the results will lead to some funky conclusions, namely:
1. The top 15% of players are crushing pretty hard, making a whopping 2.4bb/100 or more.
2. We see that the winrate of the winners is substantially above 1bb/100.
3. If you increase the rake by 1bb/100*, the top 50% of players are still going to be profitable.
*this increase of 1bb/100 roughly corresponds to cutting all VIP rewards for a high volume 5/10 player
I don't know enough to be able to say what the proper statistical methods are in order to avoid these types of biases. But I do know that all three conclusions are absolute bogus. Clearly, there is a systematic bias that ignores players that happened to lose despite being profitable, and overestimates the profitability of winners. Depending on what % of the "top regs" from this sample you wanted to look at, you could easily overestimate how much this game is beatable for by well over 100%.
Worst of all, in my hypothetical example we were dealing with pretty huge 500k hand samples. If you are dealing with a game that does not run very often - say 50/100 the errors would be huge. If you assume that there are a dozen regulars who play 50k hands of 50/100 per year and they all have a 2.5bb/100 winrate, it would be an unlikely result if at least one of them did not post a winrate that's about 3x higher than that. It's kind of disastrous to then conclude from this that "winners are winning at too high of a rate."
Monorail + everyone else, does that help answer your question?