PDA

View Full Version : I deem your sample size: INSUFFICIENT


bisonbison
09-14-2004, 01:18 PM
PokerTracker, like most swords and also like most good things in life, is a double-edged sword.

We come to the 2+2 forums looking to improve our poker game, and find herein a most generous bounty of advice and sample hands and also the Zoo. And just when you're wondering how you'll learn to plug leaks too and keep track of the vast mounds of cash that proper 2+2 play will win you, someone mentions PokerTracker. And, sure, it's a little expensive now, but everyone is using it and the first thousand hands are free, so what have you got to lose and you install it and you open it up and you find stats. Hundreds of stats, all authoritatively organized into rows and columns like you've never seen, and topped by cryptic acronyms and initials, that you're told mean things like ColdCalledSecondInWithUnderPairToPreFlopRaiserAndF oldedOnTurn (CCsiWUptpfr&FoT).

And so, struck both dumb and stupid by this vast and confusing array of stats, you find you have no choice but to post them here and ask for feedback and approval and commiseration and awe.

Because, you see, you're on a downturn. Or a breakeven non-turn. Or you've been winning and you're not sure if it'll last. Or you've been running over the game and you want to know if you should move up. Or you're just not sure what any of it means, but the squeaky wheel gets the grease and apparently the profanity filter makes it impossible to communicate to everyone how stupid and useless most of these posts are. And thus they continue and someone feels the need to register Sample Size Man and make posts so as to stop the madness and of course it doesn't work.

So... I'm just going to lay out why I hate the two most common types of PT stat posts and then take a nap.

1. I AM ON A DOWNTURN
Dear Sample Size Man,

Over the last one to ten thousand hands, I've lost somewhere between this amount and that amount, and I'm doubting my play and my wife is finding the mailman attractive and even the dog looks at me with contempt and also longing for kibble.

Sincerely, Didn't Owen Wilson's Nanny Turn Up Racing Nebraskans?


Dear DOWNTURN,
Here are two blocks of stats from within my last 3 weeks of play:
-------------------------
BLOCK 1:
Hands: 4152
VP$IP: 15.61%
PFR: 9.80%

Aggression:
A-Flop: 3.14
A-Turn: 2.20
A-River: 1.13

Went to Showdown if Saw Flop: 37.36
-------------------------
BLOCK 2:
Hands: 4134
VP$IP: 13.57%
PFR: 8.18

Aggression:
A-Flop: 2.69
A-Turn: 2.68
A-River: 1.85

Went to Showdown if Saw Flop: 34.91
-------------------------

During one of these 4k stretches, I won 125BB. During the other, I lost 146BB. While you may be able to figure out which is which, the knowledge is useless.

Why, DOWNTURN? Because the stats cannot distinguish between cause and effect, between good cards and bad cards, good tables and bad tables, overplayed hands and underplayed hands. Poker is too goddamn complex to assume that the aggregate of your stats actually reveals how you've been playing.

If you can play perfectly and lose, then losing CANNOT matter. Only the quality of your play matters, and your stats will only describe the quality of your play in the bluntest, dullest and least useful form. Post some goddamned hands.

Yours, SSM.


2. SHOULD I MOVE UP?
Dear Sample Size,

I've been playing Party Poker 2/4 1/2 .5/1 3/6 10/20 for the last 1000 to 30,000 hands and have been beating the game at a rate of 0 to 11BB/100 hands. Should I move up?

Thanks, Why I Never Remember All The Englishmen.


Dear WINRATE,

It's pretty easy to find out how confident you can be in your winrate. There have been dozens of posts (http://forumserver.twoplustwo.com/showthreaded.php?Cat=&Number=838515&page=&view=&sb =5&o=&vc=1), and plenty of explanations about confidence intervals and standard deviations and sample sizes and so on. All the more's the pity that you could care less about that. You don't want to hear that over x hands, with a y of z and a w of f, your true win-rate is h% sure to be between m and n. OH NO. That's not good enough.

You want to hear that moving up won't break you. That you won't lose your bankroll or go through a downturn or look like an idiot or get outplayed. And no one can tell you that. Because variance happens and because your skills will translate only roughly from level to level.

If you depend on loose-passive games and find yourself in loose-aggressive ones, how will you adjust?

If you depend on known bad players and find yourself in a sea of unknowns, how will you adjust?

If you lose half your bankroll due to normal variance, how will you adjust?

There's no y in stats, so I can't tell why your VP$IP is x% or your PFR is y% or your winrate is z/100. Maybe you overvalue low pairs and undervalue broadway off-suits. Maybe you play suited one-gappers in early position and fold KQo UTG. Maybe you call with any gutshot or underplay top pair.

You have to tell us about this. You have to do the heavy lifting here and evaluate your own strengths and weaknesses. If you don't post and respond to hands that reveal your play and your thinking in common situations, there is no way that anyone on this board can help you or tell you if you're ready.

So act like an adult, and risk looking stupid and post what you're thinking. That's what the board is here for. If you want someone to read the chicken entrails that are your stats over the last poopitypooppoop hands, then post in the Zoo or RGP or one of the thousand or so less intelligent forums on the vast interweb. Don't waste space here.

Sincerely, Senor Sample Size.

MoreWineII
09-14-2004, 01:23 PM
Where is "Holy Crap That's a Lot of Words-Man" when you need him?

Surfbullet
09-14-2004, 01:36 PM
Soo....should I move up or not?

MaxPower
09-14-2004, 01:40 PM
Good post. I could post my stats for the last 23,000 hands that I played at Party Poker 5/10. Everyone would say they look great, but that doesn't change that fact that I am down $800 during that time period. Stats are a wonderfull thing, but its all in the interpretation.

Stats are like a bikini, they show you everything but what you really want to see.

In addition to the hand conveter, you should make a pokertracker stats analyzer. It should work like a magic 8-ball. You enter in your pokertracker stats and it will spit our a random answer like, "The future is uncertain", "Looks promising", or "You will run bad"

afk
09-14-2004, 01:46 PM
[ QUOTE ]
In addition to the hand conveter, you should make a pokertracker stats analyzer. It should work like a magic 8-ball. You enter in your pokertracker stats and it will spit our a random answer like, "The future is uncertain", "Looks promising", or "You will run bad"

[/ QUOTE ]

I think the word "NO" would answer most of the questions.

Lost Wages
09-14-2004, 01:48 PM
In block #2 you have become too passive. Hope that helps.

Lost Wages

Bob T.
09-14-2004, 01:53 PM
Thank you, nice post. I was working on something like this, but I see that I can save it for four months from now.

SomethingClever
09-14-2004, 01:57 PM
[ QUOTE ]



Stats are like a bikini, they show you everything but what you really want to see.


[/ QUOTE ]

Just a terrific analogy. Terrific.

Lost Wages
09-14-2004, 02:24 PM
Seriously, I don't completely agree with you. Yes, sample size is important and trying to draw conclusions from 4k hands is ridiculous. Some stats, like VP$IP however, gravitate towards their true value fairly quickly. When someone has 10k hands and their VP$IP is 32%, it doesn't take a genius to see one thing that they are doing wrong. Other stats like "Won $ at SD" take much longer because, out of that 10k hand they probably when to showdown less than 2k times so in effect the sample size is reduced.

When someone posts their stats after 40k+ hands you can begin to make some general observations like, "you are folding your BB to a steal raise too often". The guy might not have ever even thought about it so he hasn't posted a bunch of hands (or any) where he folded to a steal raise. He doesn't know that he has a problem so he doesn't know it needs a fixin (that's how we say it in Texas).

Lost Wages

bisonbison
09-14-2004, 02:28 PM
I hear you, and it's part of the reason why I just included my two least favorite kinds of posts.

Obviously, some stuff converges very quickly, and I don't mind telling people if their preflop stats look normal after 10k hands, but I just think a lot of the posts are uninteresting and counterproductive.

easypete
09-14-2004, 02:30 PM
Agreed. And let's not forget, we often make judgements on our opponents w/ just 30 hands in the database for them.

I've had 30 hand runs where I've been 0% VPIP and 60% VPIP. It's crazy for me to draw any conclusions about my own play in 30 hands, but we still make these decisions about our opponents just after 30 hands.

So in saying that, yes... I'm ready to move up.

Bob T.
09-14-2004, 02:36 PM
30 hands!? Sometimes I make the decision in one hand. I see someone limp UTG with J7o, or 82s and I make a lot of assumptions in a hurry.

Bob T.
09-14-2004, 02:41 PM
Some stats, like VP$IP however, gravitate towards their true value fairly quickly.

But even these statistics hide the true picture. Some players might have a hard time folding AJo preflop, but might not play 86s in LP for one bet. They trade one hand that is unprofitable, for another one that is, and there statistics seem right, but they aren't getting results because they find themselves in difficult situations all the time.

sprmario
09-14-2004, 02:43 PM
I hvae about 20k hands and one thing I noticed is that I'm playing hands like AJo, KQo, KJo, and QJo way way too often. I've lost more money on QJo than any other hand. My cold calls on these hand are scary high. I'm pretty confident that I know I'm missplaying these hands. They should not be the big losers that they are.

One way that I can see that I'm no where near the long run yet is that QQ is my #1 moneymaker over 20k hands. Its slightly behind AA in BB/hand but I've had it 15 more times. Right now the order for me is AA, QQ, AKs, KK, JJ, KJs. Interesting stuff.

chesspain
09-14-2004, 02:45 PM
[ QUOTE ]
I have about 20k hands and one thing I noticed is that I'm playing hands like AJo, KQo, KJo, and QJo way way too often. I've lost more money on QJo than any other hand. My cold calls on these hand are scary high.

[/ QUOTE ]

"Scary high" for any of these hands would be anything over .00.

Stefan Prodan
09-14-2004, 02:47 PM
I think you forgot to switch accounts, bison.

TheHip41
09-14-2004, 05:47 PM
[ QUOTE ]
I've lost more money on QJo than any other hand. My <font color="red">cold calls </font> on these hand are scary high.

[/ QUOTE ]

I've found your problem with QJo I don't think I've ever cold called with this trash /images/graemlins/smile.gif