What I'm saying is that drooler you are talking about, has one of the best mttsng scores over the longest sample size available online. They are offering coaching to you at the live tables. do you accept?

when "5bps of edge" isnt enough...
https://youtu.be/k3REB7RkBbU?t=622

Yes. From a player's perspective 'reasonable' becomes a match of expectation to income etc. We aren't at that question yet.

Reading from your tone, my quess, if you go look at the data available, you will come back and say you agree with me. You are assuming that the sample sizes are 'reasonable'. They are QUOTE OBVIOUSLY too small.



I'm not using this point. Ignore this point for now.

Is there a typo there? This doesn't go against anything I have said. The relevant point is that you can hard core skew your stats in mttsng if the tracker site lists "5-15$ games" and you binked a few 15$ games and plyaed the rest in 5$ games.


This is speaking to/against my argument B. It's premised on A. Show me the data available IS sufficient. Or tell me you looked at the data and its enough.

If you have 10 games sample, you cannot make a good decision. if you have 100 games sample also can't. if you have 100k games sample you CAN make a good decision. If you have the ACTUALLY available sample you cannot make a good decision.

Let's call the available sample x. You didn't check what x was before you spoke. Go look at x. The highest available sample is too small.

You can make a good decision on 10 samples for a personal definition of 'good'. There is no absolute definition/threshold for where 'good' starts. There is some that is used as convention in scientific papers (e.g. how many standard deviations you need before you can claim a discovery) but those thresholds are just convention - not based on any kind of real qualitative difference. An 85% confidence interval is just a quantitative difference to a 95% confidence interval.

What your personal definition of 'good' is depends on what your tolerance for risk (i.e. being wrong) is. Going all-in with QQ against AK is a good decision - but it still turns out 'wrong' around 44% of the time. Yet we would all be satisfied with having made a good decision, no matter how it turns out.

Some side note on coaches: Besides looking simply at the win rates of the coach you could also check on whether the win rate of students of a particular coach have improved, which you could basically add to your sample size.
...And then again there are some things that coaches teach which are independent of the coache's own aptitude for using it in a live setting. Much like there are in any sport coaches who are worse at the sport than the ones they coach. The reason they still provide value is because they understand the math/methods/psychology and can teach those without necessarily being able to employ them under time pressure themselves.

I don't think using 10 games to judge the likelihood of a players mtt skill is a good idea.

This is WILDLY not on topic.
If the sample size of the entire field isn't big enough, how would considering only 2 players from the field be big enough?
Right ic. So if the coach doesn't understand the game or the math of the math game, then their attitude or something makes up for it in poker? I'm a little confused.

What I'm really not understanding is our inability and unwillingness to comment on the actual available data.

edit: put it another way. instead of saying "oh im blind to the stats and im blind to the math involved and blind to what is a good sample or a bad sample. Let's all suggest what might be a reasonable sample for 180 mttsngs and then i'll show you that the available sample is nowhere near what you thought.

Fair enough, thanks for the response. You sound educated enough that you can understand why I want to check methodology and why I ask for someone to show their work. (I still don't know what Dozier's actual win rate is, but at least you confirmed you based your statement from her SNGs specifically.)

Incidentally, you can credit/blame my best friend growing up for these questions. He co-founded an organization that promotes transparency when it comes to research in an effort to improve integrity and reproducibility. As one would guess, many a "scientific study" comes out with some surprising statement, and the trigger-happy media tends to report those findings before anyone gets a good look under the hood. Whether it's due to bad science or bad reporting, unreliable conclusions often make their way into the zeitgeist.

Consequently, he has me and most of his friends pretty well-trained at asking a lot of questions at almost any claim. (And as a PIO, I already have a trained eye/ear for media bias, etc.) So apologies if I sound like I'm attacking your integrity or something.

Fair enough...maybe...I think I was a little un-motivated to sort of prove that, expecting people want to check for themselves anyway etc. I forgot I had my original thoughts still in a past blog so I have the details I was going on. I still feel like no one will comment on the numbers but will just say I'm a crazy idiot. Do you mean to call me out but not really care when I show my homework anyways?:

And I agree. But it would still be better than judging someone on 0 games. I was not disagreeing with you that a small sample size is problematic and that someone shouöld buy their coaching just based on that. I'm saying any sample size is better than no sample size. Whether it's good enough is up to the one wanting to sink their money into said coaching.

...because what you are arguing is that even though some people are having a win rate with a small sample size that they MUST be outliers (i.e. that none of them is possibly a winner due to edge)..and that's an entirely different thing to prove.

I played poker full time for some years. There is no decision I can think of that a 10 game sample size affected and certainly not one that is worth bringing up as a 'counter point' to anything I have said. I am talking about useful sample sizes versus sample sizes that aren't useful. 10 games is not useful for anything else give an example please.

I have no idea how you would think I am arguing that and can only think that you have created a stawman.

Think of 3 types of profitability a player can consider:

1) whether or not a player could reasonably afford their own expenses at their winrate. (at no point here am I arguing sites owe players this and I will reference this bolded if someone responds that I am)
2) whether pots and prizes are awarded/distributed/paid fairly etc ie whether or not sites cheat the prize pools-this we aren't considering.

3) whether or not the players playing the most profitable strategies have an expectation of profitability (ie is the rake and effective rake so high that no players have a profitable expectation).

We want to consider 3) and the range/degrees of the field/time considered. I'm saying you can't measure/observe/analyze this even in the most liquid games of the most liquid times.

It's not something to be an opponent of me for saying, or call me crazy. If you disagree...I posted the numbers...reference them at least.

"More specifically I'm thinking of a time when Daniel Negreanu headed a poker stars campaign that tried to convince the players that higher rake was better for the game. The basic idea being there is some subset of players that YOU the reader belongs to, that benefits from 'low rake seeking players' leaving the site for more profitable games."


Daniel said this because he's dog s hit. He knows this is stupid. He basically sold out
to the poker world massively. Probably as the highest endorsed poker player in the world
at that time, or at least top 5.

But only a few players on 2+2 and Doug Polk called him out.

When I guy whose not top 100 in the world, and is probably top 5 in endorsements,
sells out like that. I mean you may as well give up hope. Imagine what this guy
would do or say if he really needed money and was broke!

So its pointless to police or call out anyone. The average human is just too pathetic.

Ah, no.

Very productive discussion itt. I love to read endless opinions about variance and gto/exploitation

Bobo Fett casts doubt I believe. And yes there were only a few of us. I hope people remember what Polk did back then (and Joey iirc). I do remember as well a type of poster that would always break into the general complaints against raising rake to proclaim: "POKER SITE DON'T OWE YOU A LIVING". I maintain that such a player archetype is self-inconsistent and cannot exist from a game theoretical point of view as a (game theoretically-rational) player. Not that I mean to say poker sites do owe players a living, but rather that the best players should expect non-zero non-negative profits(else where does the money go?!).

I've put up the numbers, I put them in a spreadsheet, I "parsed" them to the best I could, and I summarized them. Then I summarized my summary.

Since we are poker players and interested in reducing probabilistic type data into finite conclusions/decisions I think that my point "A" should stand at least until someone references the actual data I put up. Or at least conjecturally we can start to understand how this would shape our understanding of the game as it has evolved.

Putting this another way, by now we should be able to have (or at least want) a sample size to view empirically whether or not Dnegs claim that raising the rake would make the game more profitable for the player archetype he was referring/preaching to (ie 2p2 regs!). But of course we run into an intrinsic difficulty here in that the games must be less liquid than at the time when mttsng 180's were at their peak (or we can test this) and if that is the case then if "A" is true, that there was at the highest liquid times not enough data to prove profitability of the games for the best players, then it must be that you also cannot discern this for today's game.

Then we would have shown that there is no way to prove Dnegs claim wrong, even if it were wildly so.

Here tho is where we again highlight that line that is alluded to in the OP:

And we can see how it creates an assumption for us that is implicit but useful for the observation that there cannot actually BE an (empirical) observation.