Somehow twitter has it in its algorithm that I want poker in my feed even though its basically a trigger for me I mean to avoid. So in that sense I'm forced to come across a V Selbsts tweet about the WSOP FT and I'm all too much reminded of the narrative her and other poker stars pros carried and sold to the players:



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.

But even before that I found something strange about the whole poker ecology. For me I played mttsng's full time (low/mid) and a lot of them. 30-40K games a year for a few years playing as many days and hours as I could and upwards of 30 tables on regular. I had some of the volume reasonably possible.

At some point, I wanted to prove to my family it wasn't a gambling addiction so I tried to apply variance calcs to the results and I realized something that I'm not sure many players really understood. You can't just take your actual winrate and apply a probability distribution to it and discern your true expected winrate.

Put another way, just because your winrate is high doesn't mean you are more likely to be a player that wins at a high winrate. (correct me if I am wrong here). I was confused but more so because none of the players would entertain a dialogue on the subject. Then I started to look at all of the winrates of all of the 'top' players in my field. It turns out that there weren't any real consistent year over year winners. Moreover and for example, collin moshman's wife (moshman wrote the book in sng's) wasn't winning much, how does that work?

Some of the top winner's had bad strategy, either observably or noticeably from 2p2 posts or skype chats etc. But what was even more telling and alarming was these top players and coaches didn't understand the limitations of variance observations. And they laughed at the subject. I wonder now if the Overton window has shifted.

I call it moral poker to work together for the greater good of the game, outside of a hand. Variance observations and understanding can be this.

What I had believed that I had discovered in my field was that you cannot reasonably play enough games to ascertain a reasonable confidence in your winrate. I extrapolate some things from this. Players that don't know this are not good coaches to have. Poker sites are aware of this fact and set the games up like this on purpose.

And if its true that my games suffered from this, then mtts suffered worse. If games suffered from this 10 years ago then because of the decreased profitability (from increased average skill) there is even higher variance to face.

Trying to keep this short it will seem like a leap but the reason I think of this reading Selbt's post is it reminds me of the time when all of these 'pro's' who were supposed to be ambassadors of the game were actually profiting off of spreading ignorance about it.

In a game that is clearly tending towards a ceiling of strategy equilibrium her comments make no sense. Just like a commentator saying someone made a good fold because their opponent had a better hand but ignoring range.

I think only the best will agree with me and the rest will laugh.

Is my understanding of variance and the available data wrong?

also critical point of my argument I forgot..back around the times immediately prior and after black friday especially, wouldn't it be natural for us intelligence to be gathering intel from this site?

These are interesting topics to discuss. Honestly I couldn't even get through all your post it rambled so much maybe you need to practice condensing your thoughts and becoming more concise. Literally none of the topics you addressed that I read refer to your twitter quote/title.

I'm a firm believer that the "long run" which allows a lot of what poker analytics use when dealing with variance is overused and rarely if ever applicable in the ways it is which is a lot of what it seemed like you were referring to. Unfortunately it's the best tool we got kinda.

I'm also a firm believer in what Selbst refers to which GTO is a unbeatable strategy but not optimal against every strategy especially as your opponents deviate from GTO it becomes increasingly noticeable. Unfortunately it's more difficult to teach adjustments so people have focused on oversimplification especially training sites recently.

well to condense the most important talking point I'm suggesting that its provable for at least my field you cannot prove that the games are reasonably profitable even for the highest possible volume/sample. And that in contrast most well regarded players believe that their track record proves their ability-it can't.

ya hands/sessions are not non varying independent simulations which is required for the rule of large numbers to occur. But like I said it's literally impossible to model situations like this. You have to approximate and make assumptions to create any meaningful conclusions and its the best tools available in statistics which is why its used.

well consider an mttsng coach that says 'here is my winrate and here is my sample size' and you notice its impossible to show skill via winrate given reasonably available sample sizes...

can this coach be good at teaching profitable poker?

It's not impossible to show skill its just not perfectly accurate. A large sample size with a large winrate creates pretty high certainty they were consistently implementing winning strategies. Just cause someone is highly skilled doesn't mean theyre a good teacher either. You use the best information you have in life/poker/everything and make the best decisions you can. If you have a better way please enlighten us. You seem obsessed with this system but provide like zero solutions/suggestions/ways to improve just complain about the system.

I'm not sure if my claim is clear. I'm suggesting given the possible sample size one could have it is absolutely not possible to show who the profitable/winning players are in my mttsng field, which was one of the highest volume fields in the industry. If I am correct then it applies across the board. I'm suggesting this can be shown by math not sentiments


Thats a backwards corollary. What I am saying is if they don't understand the principle above then they can't understand what good strategy is. Is that controversial?

This is what I mean by overton window, regardless if I will or mean to present a 'solution' the conjecture could still stand true and be proven.

Your first part is simply wrong as stated. It is clearly possible given the data to show who the profitable/winning players are. Do you mean to say that the sample size is such that there is no guarantee that these are the players with the "best" strategy as opposed to the players that have been most fortunate? In actuality assuming even luck a player playing perfect GTO will have poorer results than a player playing an effective to that game at that moment exploitive strategy. The disadvantage of the exploitive strategy of course is if you don't change gears properly all exploitive strategies are exploitable as well so you wind up being exploited for more than you are exploiting.

Why not be more specific... I mean yes large field MTTs (100+ entrants) with highly skewed payout tables probably wouldn't normalize in a sample size that is reasonable for any players really and have too high variance and too low winrates to really have a high confidence is results.... It's still better than nothing especially if the volume is high though just wouldn't have 95% confidence intervals that are winning just probably like 80-90% confidence intervals that include only wining predictions. You correlating that to all poker including cash games with much higher winrates and lower variance and not skewed distributions is absurd though.

That's a ridiculous claim. You're essentially saying the better you are at understanding math principles the better poker player you are. Well not to that extent but basically that you have to understand the underlying process of complex math principles to be good at poker. That is really not the case. Poker honestly has more to do with pattern recognition and memorization now a days then mathematical understanding in my opinion.

What you refer to in a round about way has already been proven in the math world. I think it's actually a theorem that law of large numbers breaks on heavily skewed distributions (as simulations go to infinity it won't normalize). Any mathematician would think your a child for even suggesting its revolutionary. The issue you have is you mischaracterize and misapply the concepts and create no solutions. THERE IS NOT A BETTER WAY TO MATHAMATICALLY CHARACTERIZE RESULTS. If you don't have one which I'm assuming you don't as you keep circling around to other random points instead of addressing this your points are useless and so is this conversation.

Adrian is likely better than Vanessa at poker now, Jorstad probably is too given she's been out of the game and not studying for a few years, I don't know why we'd take her opinion to mean more than a random person who used to be an ex pro. Obviously Adrian played great the last few days, made a bad read or whatever heads up and that was that.

Obviously there are a ton of spots to play exploitative rather than GTO live even deep in tourneys (for example against Duczak who was clearly a nitty recreational player) but Jorstad probably isn't one of them given he's quite a strong player too and Adrian is obviously very good so Jorstad can't really 'adjust' to play ABC nit because Adrian is capable of hero calls or whatever or Adrian will run over him

They both played well to finish first and second and well done to them

The idea of what correct fundamentals are today is vastly different to 7-10 years ago back when I used to play full time and when Vanessa was actually an elite pro, which she was for her era. I'm sure with proper study she could be an elite pro today, I probably could as well if I dedicated a year or two to studying the game, but given neither of us have put in that work for half a decade any ex pro plus telling Adrian or Jorstad what they did wrong is pretty ridiculous because they're very likely both better players Vanessa right now (i'm not talking skill ceiling wise necessarily I just mean with zero additional study playing today)

Obviously exploitative play is better than GTO against outright terrible players you'll have at your table on Day 1 in the main but there weren't that many of them left deep in the main this year

Your point here is pure sentiment. It contains no math. You have said its better than nothing and I am claiming that it actually is not better than nothing and that in my mttsng field you can't get any information about a players ability from their winrate. I did the math. Did I do it wrong, show me.

Poker has a game theoretically optimal strategy. We should think of the field as generally trending towards it. We need to know that strategy to understand where balance is or where equilibrium deviates from. Over time, as fields get harder, there is less and less to the game but this fact. But how far along are we? How important is exploitative play in today's game in which the field is better than 10 years ago?

I don't think non-quantified guesses have merit here.



Writing in caps that we don't have a better way does not in any way whatsoever preclude the point. I have no idea why you think that it should. Why does it upset you to notice this?

k we can put our money where our mouths are. Give me a distribution of win rates and sample sizes from any game site. I'll pick 100 players that you have to crossbook minus rake for MTTs. I'll set up the escrow when you're interested. I'll pay twice what the net loss if it is. u pay 1x what they net win if it is over a year.

Well there was something else this inquiry lead me to understand which is relevant and important. I started to think about variance and sample size in regard to hud stats. I would have had one of the biggest databases for my field compared to even most of the grinders.

Of the top volume regs I think I had maybe 10k hands max with them. And for most regs probably more like 3 to 5k hands in mttsngs. (trying to remember)

What decisions can we confidently make with these sample sizes?

Well if you break an mttsng down by icm/cev and then down by stack sizes until starting stacks and then consider a preflop raise from a certain position one clearly doesn't have enough of a sample size to determine an opening range.

It might seem far fetched, but the top players will agree I think. You can't even really determine if a regular player is loose or tight in a certain spot let alone a rec.

There's no confidence. You will have very few related hands to observe from statistically (if ur hud can even break it down for you in real time) and any further reductions like post flop cbet get absurd to consider.

My backer would get upset for not using a hud. But if my observations are true then using a hud would dramatically skew ones decisions towards random and away from equilibrium.

And then the players joined an argument that huds are an unfair advantage. I think that mttsng players that use their huds don't understand poker.

People still think this?

Vanessa was, is, and always has been terrible. Just goes to show everyone how soft mtts were back then.

gotta take Selbst opinion with a grain of salt, even back then when she had “success” she was mostly a fish on a heater, that said it’s true that if you spot a clear leak in your opponent’s game you should exploit it and not stick to GTO so she sort of have a point still

She's spot on - GTO assumes your opponent isn't highly exploitable and in most live games (especially lower stakes) you're playing against opponents who are highly exploitable. You don't want to play balanced against fish - you want to exploit their strategy by being exploitable yourself. I didn't follow the final table so no idea how everyone played, but against fish she's right.

Too much hate for Vanessa. She would destroy all of you in a random MTT and she hasn't really played in years.

You seem to agree there is GTO and subsequently that you need to exploit with a hud, whilst claiming that you cannot have large enough of a sample to determine who is winning in EV and what is the best strategy, correct?

It seems to me the error is that you agree there is an objectively perfect strategy, assuming villain plays perfect, as you agree the game trends towards equilibrium and you disagree with Vanessa; yet then you leave that place and move to the argument that HUDs are imperfect and one's adjustments are imperfect, with the conclusion being, therefore people can not know whether they win due to variance or because they're good.

You seem to be giving your own answers, and they're both right, but then the conclusion ought to be that your GTO strategy is based on conjecture, as you said too; so the final strategy that you know will be +ev, is the GTO one that assumes villain plays GTO and the way you create that strategy is by having good explanations, i.e math based explanations, for your strategy. You conjecture it from the ground up without any experience or database. This is what you seem to think, but then you move to empirical evidence and observations and state you cannot have a large enough sample of a specific spot to know what they do. Given the first argument of good explanations and conjecture, it seems we can prove that you do not need this sample to prove that one person's strategy can be better than someone else's, and that's all you need to disprove the claim in OP of 'poker is just variance and we can't have a big enough sample'.

Am I misinterpreting parts of what you think?

It feels like people are inferring things from my points and saying my points are wrong because they don't like what is inferred.

In poker one of the great mistakes is to make a decision based on information you don't actually have. If we are inclined to believe my observation about mttsng sample size, we can determine a good or bad player using reason and watching them play to some extent, but we can't do this by using their winrate-this means players that do are bad.

"Hi I made 25% roi playing 100,000 mttsngs I will coach you"

That's a coach that doesn't understand poker.

Something that a lot of these discussions on GTO vs non-GTO blah blah miss is the viability of reliably identifying an opponentÂ’s consistent deviation from GTO and in turn devising an executable counter strategy, while also being aware of the exploits you open yourself up to and countering those.

Like how do you even know if your opponent purposefully deviated from GTO or just randomly clicked a button? Are you really going to make wide sweeping changes to your strategy after 1 showdown?

Imo, a reason good players study GTO is because itÂ’s a tangible strategy that you can work towards implementing. It performs against all opponents and exploits players who deviate, without you having to change your strategy.

Like I get what Vanessa is saying but I think that line of thought is too idealistic. Sure, if you have highly reliable reads that your opponent consistently deviates in consistent spots, then yes exploitative play is better if implemented effectively and not counter exploited. ItÂ’s sorta an obvious statement but also an improbable reality.

Vanessa has definitely earned her right to comment on poker, but imo thereÂ’s clear reasons why her breed of poker player canÂ’t compete with the GTO kids anymore.

As for your other point, yeah there’s an insane amount of variance in poker and a lot of winning players are probably just on a heater and not rly beating games for much. There’s a reason so many poker players go broke at least once in their career, regardless of how much they might have won at one point.

These two points are self contradictory. This is what I want to illuminate.

Here I am also thinking of the live television announcing where you get someone like selbst (a poker influencer) saying things like "oh can this player get away from this spot" since at home we can see their cards, but really the opponents cards don't actually matter whatsoever in regard to making the correct decision.

The pros are selling a lie.

I understand your point and agree to an extent. Vanessa isn’t here offering coaching. Just stating her opinion which I think holds SOME weight given her years of experience as a top pro. Not because she won a lot of money.

But yeah, present day Vanessa selbst probably doesn’t have too many relevant opinions on current poker strategy.