Variation is a tough thing to quantify and there are no hard and fast rules other than the more you play the more likely you are to uncover the unlikely.

As an extreme example, let's suppose you're playing a player HU who has a strategy to play every pot to the river but fold when faced with a bet. This is obviously a bad strategy and you are kicking his ass. But, on average, once out of every 50 hands, he accidentally calls a river bet. He might not even win the hand when he calls, but he does. That will add some "variation" to your strategy, but you'll still be a winning player against him. You'll probably never experience a "losing" streak per se - you may have to play more than a million years before you ever did lose consistently against him for a day. But, in all likelihood in your lifetime, you just may go through spurts where you don't win as much.

So, too, in practice when we play against different players and games and so many variables, there is a "range" of variance. Maybe that variance includes 10 buy-in down swings, maybe only 4, maybe 20, maybe none. It all depends on your skills, your opponents' skills, and the nature of the game in terms of stack sizes, aggressiveness, and however the chips fall on that day.

The main point to remember though that the long run is long and you do need large sample sizes to even begin to know what is possible. Theoretically, a monkey can randomly type out the Complete Works of Shakespearean ... if given enough time.

online is directly comparable to live, it is just math and a function of winrate and std dev

winrates and std dev are higher live so a "good sample" will be lower, hence why the 1,000 hours will likely put you winthin 1 std dev (or so) of your true winrate (this is me ball parking) is likely +/- 33% of whatever you have assuming you are running "around" expectation.

bayesian statistics would lead me to believe that if you are near the top of what is generally accepted at 1,000 to be "top tier" that it is far more likely you are running good and your true winrate is between (.66)*x to (1)*x where x is your winrate after 1,000 hours.

happens all the time and that is why the games are still good

lol and none of this is meant as some sort of Farha "lol you aren't a huge winner!" it is just more of what the thread is about. I mean you are obviously crushing but it is pretty much SOP for people to over estimate their skill edge in the game.

Thread probably has a built-in bias, in that most people posting / bragging about their winrates are big winners. Needs more losers plus winning players experiencing 10 BI downswings obviously.

This debate has nothing to do with me or my own winrate.

We've had a few guys with even bigger samples (and likely a lesser wr) than my own come and say they haven't had that big of a downswing. Admittedly squidface has admitted that he has had a downswing that big and I'de imagine his sample is bigger than all of ours. I just think there has to be a couple a of big mistakes that people wouldn't normally make. When we run that terribly, we play worse than normal, we're all human.

My argument is not based on a "poker robot" that makes the perfect decision all the time. Obviously none of us make the best decision all the time. I also get that your saying standard spew plays are factored into winrate, I get that.

The point I'm making is that during the 8-10BI "downswing" you are likely running bad and playing worse than you normally play.

Which equals you losing 8 buyins. You lose 4 to variance and 4 to bad play that you don't normally do. Or 5 to variance and 3 to bad plays you don't normally do. The people who post these probably only remember the losses directly correlated to variance.

Playing longer with a terrible image is also a mistake that no doubt helps this downswing increase.

see you are saying that somehow a winning player goes from being +EV in a game to -EV which does happen to people who "tilt bad". What DOES happen to a lot of big winners is that they go from +EV to less +EV so they go from their A game to their B or C game. However, many big winner's B or C game is still profitable just not as profitable as their A game.

This is basically impossible to over come since we are humans and your only real option is to get up and walk away.

You make it sound like somebody gets stacked and just LOL SHIPLKSJF shippit tilt monkey, but it is more likely that they do something like lose the ability to make a big fold they might otherwise make, but that doesn't really effect their long term winrate because it is such a marginal spot that occurs so infrequently and likely won't occur during their "tilt".

I mean tilt is a relative thing and is anything that brings us down from our A game (or theoretically an optimally exploitative strategy).

Also people who are tilt monkeys like you describe likely aren't winners at all, you really can't recover from donk shipping off 4xbuyins but even if you are shipping in combo draws/draws with no FE, generally with dead money in the pot and the equity you have you aren't on average losing 4x BI. even that would be running below expectation.

but yeah obviously there is a correlation to running bad and playing bad, I play bad all the time I just don't lie to myself about it

Checking back top pair they normally wouldn't. 3betting in a marginal spot and being more apt to barreling.

This is not WTFSHIPITTILT. They are marginal mistakes that wouldn't normally be made by someone who isn't losing.

I'm not saying people go on mega monkey tilt and ship overbet everything.

I'm saying that they go from +EV to less +EV or -EV which in turn, causes variance to go up which makes it possible to actually lose 8 buyins.

Or being less likely to fold (especially preflop) which adds up.

Running bad = playing worse than normal which makes it possible to lose 8 buyins where it would near impossible (mpethy ) to lose that if you were playing your A game at all times.

but what you just described isn't tilting off 4xbuyins, it is losing small amounts of value and that your true winrate is a function of your "skill distribution" which is built into your winrate and standard deviation.

then running bad is just an outlier event that occurs because of your winrate/standard deviation and that statistically you will in fact go on a large down swing. haha just that if your winrate is low enough and standard deviation is high enough you will have more frequent downswings and your outlier events will be much larger. In fact, it is 100% going to happen in the long term.

Great players running bad = breaking even
Good players running bad = big downswing
Break even players running bad = losing routinely
Losing players running bad = losing at a higher rate than what would other wise occur

This is ridiculous. Poker math is poker math, whether live or online. The odds, for example, of losing 8 flips in a row, are exactly the same. The probability that you'll go 0 for 5 on flush draws is exactly the same.

Higher win rates live do mean that, on average, downswings will be smaller live than online. But that's accounted for in the fact that we are talking about swings in the neighborhood of 8-10 buy ins, which online players hardly even notice. Online winning players routinely experience downswings of 20, 30 or 40 buy ins, because their edges are smaller.

I didn't get that from your post at all, but, ok.

There's no way you're on the negative side of variance.

This whole section just shows that you have no idea what you're talking about.

A downswing is a downswing; it's an after-the-fact evaluation of what happened in the last X hands you played.

Second, if I tilt shove 2 flush draws, I didn't spew off 2 buy ins, I spewed off 1.3 buy ins--DUCY? JFC.

Third, even if I grant your definition that a downswing excludes tilt, an 8 buy in downswing is still guaranteed in a big enough sample. Hell, online, I had a 4 buy in downswing in under 20 hands--I was 12 tabling, got set over setted on 3 of them simultaneously, and within an orbit ran KK into AA AIPF.

Online, I had a -35 buy in downswing in 35k hands that included losing 13 of 15 times I got all in with a set on the flop or turn against an overpair.

Live, in June of 2011, during my 11 buy in downswing, I got all in 19 times. I never had worse than 47% equity (AK against JJ), and only won 3 of the 19.

If something like this has never happened to you, it's just because your sample is ridiculously small.



You really just don't understand variance. "Highly unlikely," by definition, is exactly the same as "guaranteed to happen in a big enough sample."

So again, the fact that you haven't had a big downswing means either that your sample is small or you've run good. It could mean both.

<shrug> In your place, I'd be confident I was a winner, but I'd have ~0 idea what my actual win rate was.

But I didn't come in this thread to challenge your results, or your WR or to tell you you're running good.

You just said something that is mathematically silly--that a -8 buy in downswing is impossible for a big winner. And I'm just saying that that statement is ridiculous.

yeah and like breaking even would include large downswings just be much more infrequent for a great player

Absolutely not, because the math is the same whether online or live.

The way variance works, is that the probability that you will experience a downswing of x size, is just a function of your win rate and your standard deviation. The formula is the same for live or online, the inputs are just different.

Live win rates being higher should cut down on the size of the swings you'll experience. But it doesn't cut them down to -8 being unlikely. Online players with good solid win rates in the 3 to 5 bb/100 range routinely experience huge 20, 30 and 40 buy in swings. Online, an 8 buy in downswing isn't even worth talking about--I had probably a dozen -8 buy in days in my online career.

So the smaller win rates online leading to bigger swings is already built into this conversation, because we're talking about a fairly puny down swing, anyway.

also sample sizes required to make statistically valid inferences are smaller live due to higher winrates and smaller std deviations

Ok, that makes sense M.

However, solid live regs could probably (???) make in the 10 bb/hour range, which is what, 30+ bb/100? Are there lots of stats available for 30+ bb/100 winners and how many 8 BI downswings they've had? (it's an actual question; I'm absolutely clueless on on-line stat compilations / etc.)

My local game is 300 max or 100% of big stack (yes 100%). My sample size is extremely small and I am running like God over my first 89 hours. $5,747 profit for an hourly of $64.57. 19 winning sessions and 4 losing sessions. I am aware this is completely meaningless but I wanted to brag anyway. My question: What kind of win rate is sustainable for a good and great player at these stakes? Assume that I buy in the max whenever I think I am considerably better than the players with more than $300 and that I always buy in at least $300.

Edit: Also, strong join date to post ratio.

Multi-quotes and internet acronyms ftw.

Their edges are like 3-4 times as small.

Poker math is poker math. Playing live poker is not like playing online poker. Boredom and playing sub-optimally come a lot easier live which can cause someone to play way worse than they are capable of playing over a super-short sample of hands that can result in a huge downswing. You don't see this online because edges are smaller and large sample sizes are way easier to reach where variance gets evened out.

I literally said "lol" after I said that, in the same exact post, but, ok.

Do you think it is impossible to have a 15bb/hr winrate live? Do you think it is impossible for me to be a theoretical 15bb/hr winner? If so, how do you know this? Do you know the quality of the games I play in?

My argument/question to you is what caused that downswing? How much is related to variance and how much is related to bad play as a direct result of variance?

ICWUDT OMG

Yes, I see the equity argument your making here.

You are not seeing that I am saying that by shoving those 2 flush draws your putting yourself in a -EV spot you wouldn't normally be in. You are making bad plays as a direct result of running bad, which is exactly what I'm telling you. JFC

Are you saying that if I have a major tilt problem and blind-shove 20 hands in a row, come out -8 BI after everything that this is a downswing? This is terrible play on my part, not a downswing.

How much of a sample of live poker do I need to have for this to happen to me? 3 lifetimes? It's possible. But highly unlikely to happen live. Guaranteed to happen over a big enough sample, which is pretty hard to get in live poker.

.

Every other hand was played the exact same as if you weren't losing, I'm sure.

Your not wrong. The problem is that it might be impossible for me to reach a big enough sample given the slow pace of live poker.

Thanks. I'm sure it's possible my true winrate is that I'm a 15$/hr winner.

I think it's highly unlikely though.

There is a way to find out

Was this aimed at me? I don't know my SD. Cool link though - thanks.

I'm not aware of any such stats. Certainly not for online play--the best online players were at 8bb/100 or so. Nobody was anywhere near 30, lol.

But you don't need stats. You can use the poker variance calculator to figure it out and look at what swings are possible for any given combination of win rate and standard deviation.

I would have posted the results of such a run ITT in response to Rob Farha, except it seems to be down at the moment.

for long stretches of time everything goes as it should. Then boooooom crazy crazy stuff just seems to happen out of nowhere. It has been said time and time again. You will run worse than you think possible - and then you will run worse. In very short time I lost my 2 biggest pots of the year. As you know it is really tough to get 400 bigs in the center. I lost a 4k pot to 1 outer (set over set)...then lost a set of 10's to fishes AA we got it in on the turn, 3700 pot (I believe cushlash saw this 1...I actually stood up and said phuc me - something I rarely do) when he rivered his 2 outer. This impacted my w/r this year quite a bit. During my losing stretch crazy crazy crazy stuff happened where these people had no bizness winning. JJ lost aipf to 25s whale monkey squeeze gone wrong, KK<88 aipf etc etc etc etc etc.

I am glad you have not yet had this happen...but I promise you it will. When it does, hit me up - I will give you moral support (no joke - for real)

One could surely go on an 8 buy in downswing without making any strategical errors.

It's pretty simple, really.

Suppose you get your money in the middle in 8 'coin-flip' AI situations. The specifics don't really matter-- all that matters is that you're all in, it's a coinflip, and you haven't made an error (a -EV play).

So this situation happens 8 times. What are the odds of losing 8 times in a row? It's 2^8, or 1 in 256. Of course, 7 times in a row is just 1 in 128.

As we all know, these numbers are really not that inconceivable (don't be surprised when you lose those flips).

Haha, I'm with you on that one.

lol you're so naive

or you're somehow using a different definition of "impossible"

this one