Mike, I haven't played a single on-line hand, but I'm fairly certain most on-liners would roll their eyes at any suggestion that 1000 hours x 30 hands/hour = 30,000 hands ain't a drop in the bucket.

GbutIdon'treallywanttogodownthatroad;asyouwereG

I would definitely not call myself a LAG by any stretch. My last ~100 hours I have ran pretty good as well.

Online and live games don't compare at all. Live players are much worse making win rates much higher. The higher the win rate, the shorter time frame needed to hit your true win rate...or be close to it more consistently.

GG,

So after 1000 hrs your win rate was 9.6BBs and after 2000 hrs your win rate was 9.4 BBs (meaning your 2nd 1000 hrs had a win rate of 9.2BBs). You think that's a large variation in win rates? Your win rate dropped 4%. That's basically the same number as far as poker in concerned. Of course you're not going to hit the exact same number.

IIRC, your win rate dropped farther in the 3rd 1000 hr segment but your rake went up also right? When you account for that, it sounds to me like your win rate is pretty consistent but you're arguing the other way.

People who have win rates that vary all over the map in 1000 hour blocks have serious work to do on their game.

Approaching 250 hours so far this year which is about average monthly volume for me as serious weekend warrior with day job.

Overall ok results, God damn have never experienced anything like the last 50 hours of 1/3. Went 1 for 5 winning sessions for just over -$3,400. Never been kicked in the teeth like that over 4,000 hours of tracked play. Combination of some horrible beats in the big pots followed by some tilt.

2 hands last session:
1) arrive at 8am on a Sunday buy in for Max 500. 2 orbits in drunk dude sits down with 20k in 1k chips. Gets change to play table Max. Plays awful is down to 300 quickly. Asks who wants to flip for stacks several times. 1 orbit later I 4bet jam over his 3bet with AK suited and lose to his KK. Buddy then leaves to go back to baccarat.

2) lose a 1k pot with J10 to Q9 on board of Q,9,8,7 when 9 comes on the river. The big money went in on the turn.

On bright side, I'm sun running at 2/5. Less hours because I only jump in when the game looks good.

Though maybe 1/2 is where I should spend all my time

On to some graphs. YTD overall, 2/5 , 1/3 yikes, and 1 session of 1/2.


Sent from my SM-G975U1 using Tapatalk

Doesn’t it seem like you should just play 2/5 while your confidence is high for that game? If you feel like you’re running well in that game it should help your game too.
Additionally, your game might just be better suited for 2/5. I think this is possible for certain players.

I'm still very confident at 1/3. My lifetime hourly is still very good. More just showing how over a big enough sample, there will be short periods of wild variance.

I don't agree about some being better suited for a bigger game. That makes no sense to me.

Sent from my SM-G975U1 using Tapatalk

It makes plenty of sense to me. My game at 2/5 is far superior than at 1/2.
I can’t focus or take 1/2 seriously enough to play it well for any prolonged amount of time.

What's your lifetime hourly? Whats the buy in and rake structure?
Props for being able to work a ft job and still get in so many hours.

Not only did the rake go up, but he always says the players in the game got better. So it's not a matter of variance being some monster over 1000 games. Its simply that his game conditions changed and each block of 1000 hours probably accurately reflects his win rate in the game at that time.

Lifetime tracked is 7.86 bb/hr over 3,977 hours across all games and stakes.

Buy in and rake structures vary wildly as I travel a fair bit.

Sent from my SM-G975U1 using Tapatalk

Well played Zippy!

Just to clarify, I considered my winrates for the first 1000 hours and second 1000 hours identical.

But see my giraffes that I posted a couple of months ago here. The next ~1300 hours didn't go nearly as well, like 4 bb/hr. And those giraffes show stretches of ~12 bb/hr play versus ~4 bb/hr play and everything in-between, all over 1000+ hour stretches.

Yes, I have lots of (valid) excuses (increase in rake, less hands per hour, blah blah blah). But you can still produce some pretty wildly different 1000 hour stretches, and remember, this is coming from me where my nitty overall giraffe viewed from space looks like a constant 30 degree angle straight line.

GcluelessvariancenoobG

Congrats, especially on the impressive weekend warrior volume!

GcluelessvolumenoobG

And I've said here many times that this isn't true. Since we just had this argument I'm not going to get into it again right now, but I still feel obligated to point out incorrect information when I see it.

All of this short-stack vs. full-stack and Lag vs. Tag nonsense doesn't really matter. We can argue forever about who generally experiences greater variance, but what really matters is your win-rate and standard deviation. A Tag can have a higher standard deviation than a Lag, and vice versa. IIRC both your and gg's standard deviations are very low, so you both can have relatively high confidence in your win-rates over the period of time you measured. You are incorrect that gg is experiencing this variance due to his play style. I would agree that it may be due to changing game conditions, although it really could be simple variance. Variance can be quite severe.

Confidence in our win-rate in absolute measures doesn't have anything to do with the magnitude of our win-rate. It depends on our standard deviation. However a 2BB/h confidence interval is relatively much smaller to a high 10BB/h player than to a 2BB/h player.

Browni, We are going to have to agree to disagree. No need to rehash the whole discussion again.

As far variance in general, especially over something like 1000 hours, Id say its greatly exaggerated by bad players and players who play a super aggro style that is totally unnecessary to win at a high rate.

As far as GGs variance, I say what variance?

Actually, browni made a different point than the one you disagreed on before, which is about objective SDevs. You can agree to disagree about what style is more likely to produce what kind of SDev, but the SDev's of individuals' recorded results is what it is.

And yes SDev and hours are the inputs to get confidence interval, which does not need or include the observed winrate. I can tell you with 95% confidence that my "natural winrate" is within 4BBs/hr of my observed winrate, and that is true no matter what my observed winrate is.

Higher winrates reduce risk of ruin, or even major downswings, considerably. They do not reduce overall variance, though. They just make variance easier to handle.

Well, we are right back to where we were before which is different people meaning different things when they talk about "variance".

If Player A has a high win rate and his downswings are almost always relatively small and he tends to win somewhat close to the same amounts each time over, lets say each 250 hour block of time....that to me is low variance. Their win rate doesn't "vary" that much and their "long term" is much shorter than most people. They may rarely win more than 2 buy ins in any session but they win a high percentage of their sessions.

That's what I mean by "variance" as it relates to poker.

Then there are guys who win or lose 3-5 buy ins every time they sit down. They routinely go on 10 buy in downswings but when they get hot, they log huge winning sessions. To me that's high variance.

I dont care what the technical definition of "variance" is.

That's because the guy who wins or loses 3-5 buy ins has a huge standard deviation. It has nothing to do with their winrate.

Please look up what standard deviation means before arguing about this anymore.

Im going to make this point and then shut up. Your playing style has everything to do with StnDev and variance. A good LAG will have a lower StnDev and thus lower variance (more consistent results). A good LAG (high win rate) is in more pots so he has more opportunity to use his skill advantage and wins more pots without showdown. A bad LAG (low win rate) spews and clicks buttons. His results are all over the place. He will have a massive StnDev so Yes win rate does play a part.

A TAG is in less pots and is more at the mercy of the cards. It takes him longer to get to the "long term". He will experience more short term variance due to being in less +EV spots. If the TAG plays one hand in every 5 hour session, hes going to do very well or very badly that day. His results will vary greatly day to day.

The long term isnt measured in hours played. Its measured in number of hands played in a +EV manner.

Think of a bell curve. The thing winrate determines is where the center of the bell curve is; standard deviation determines how "flat" or "tall" the bell curve is. Variance is just stddev^2.

Yes, the thing you're describing "feels" right because what's happening to the good LAG is that their bell curve is shifted far enough to the positive that downswongs don't hurt as much (in fact if they're a true crusher, the downswong might look more like winnings a little bit instead of winning a lot).

Stddev and variance is gigantic -- the long run is very long, and the long run is just a measure of what your true winrate is -- but that winrate is nebulous because it packs your A/B/C/D game all rolled up into one.

Of course there is some level of correlation, but correlation is not causation, and whether TAG or LAG has more variance was not even remotely the point of browni's post.

He was speaking of GG's observed SDev, and it is very low. That is not because he is a nit, though it may be because he avoids high variance spots post-flop often.

Generally speaking, I agree that LAG tends to be lower variance, but in GG's specific example, the math shows that his observed difference in winrates in 1000 hour sample sizes is not just due to variance. As browni correctly observed:

and

That's my point. If his 1000 blocks show win rates of something like 8.4BBs, 3.4BBs, 10.6BBs, 4.5BBs.....and he felt like he was playing the same way then THAT would be variance and would help show that 1000 hours means nothing. But thats not the case at all. Not even remotely. I stand by my point that 1000 hours of live poker is plenty big enough sample size.

You could (should) easily be having increasing win rates over each 1000 block as you get better, up to a point....but you should not be seeing crazy up and down variations in win rates over 1000 blocks.

Would you say there is any merit to this stat?

StnDev/win rate

Mine is 4.95

What do you mean by "merit?"

If you add BR to it, it could help you figure RoR.