On a 30,000 hand downswing.....

[krazyace5]

[ QUOTE ]
I'm looking forward to Team Donkey's calculation. It is the only way to answer your question.

But in the mean time, your graph shows that you are down apx 0.02 bb/100. That is a very shallow (although long) downswing. So, I expect that the probability of this happening by chance, if you are a truely 4bb/100 player, is not terribly unlikely.

[/ QUOTE ]


I actually went back into pokertracker and I was running just over 13,000 hands at NL$100 6max at (-5.28)bb per 100 hands and then since I dropped down about 18,000 hands at NL$50 6-max at around breakeven.

[ QUOTE ]

I actually went back into pokertracker and I was running just over 13,000 hands at NL$100 6max at (-5.28)bb per 100 hands and then since I dropped down about 18,000 hands at NL$50 6-max at around breakeven.

[/ QUOTE ]

That is useful, and the stakes should be evaluated seperately. It is a further bit of evidence that your game is not strong enough for the higher stakes.

The bigger picture is, as others pointed out, that we post to improve and no one plays perfect poker.

Just study study study... I'm in a 600 BB downswing (six buy-ins), and I lost 6 of 7 all-ins with the best hand when the money went in. This is through ~3k hands, so I hope I can turn it around anyway.

[Cosimo]

[ QUOTE ]
there are statistical ways to answer your question. If you want, i can calculate the % chance you've been playing losing poker over that stretch vs just being unlucky, and also the probable range your actual winrate is in. PM me your BB/100, standard deviation/100, and actual number of hands.

[/ QUOTE ]

There is a 68% chance that you'll be within 1SD of your expected winrate, 95.5% that you'll be within 2SD, and 99.7% that you'll be within 3SD.

1SD for a given number of hands is (SD/100)*sqrt(#hands/100). That is, if your SD/100 is 30, and you've played 10,000 hands, then (30)*sqrt(10000/100) = 30 * sqrt( 100 ) = 30 * 10 = 300. If you expect to win at 8PTBB/100h, then you expect to be at +2400 BB, and there's a 68% chance that you'll be between +2100 and +2700 BB. The 2400 is your EV, and the 300 is your standard deviation for this many hands.

You have a 16% chance of being 1SD below your expected winrate, 2.3% of being below 2SD, and 0.135% of being below 3SD. Luckily, you have these same chances of being above your expectation by the same amounts.

Sqrt( 30,000 / 100 ) ~= 17.3, so for OP 1SD is about 519 BB. He's down about 800 BB, so that puts him at about 1.5SD below 0. If he's a good player that should be winning at 8PTBB/100h, then he should be at +2400 BB. But he's down 3200 PTBB from that winrate. This is about 6 standard deviations. The chance of a strong winning player being behind by 800 BB after 30,000 hands is about one in never.

The chance of a moderate (4PTBB/100h) winning player being at -800BB after 30k hands is 1:13821 (3.8SD off expectation). Likewise, the chance of someone who is -800BB after 30k hands is actually a moderate winning player is 1:13821. In other words, OP is not a moderate winning player.

The chance of a marginal (2PTBB/100h) winning player being at -800BB after 30k hands is 1:288 (2.7SD off). This is fairly far out on the edge; there's maybe one poster in this forum (assuming 288 regular posters/readers) that is running this badly over the last 30k hands.

The chance of a breakeven player being at -800BB after 30k hands is about 6.2% (1.54SD off). If you play 40k hands per month, you'll have a result like this once a year. In other words, there's a 1:16 chance that anyone (with a SD/100 of 30) will be 800BB below their expectation for 30k hands.

The most likely explanation is that OP has some major leaks. It only takes about one bad call or bad fold every 10 table-hours to drain this much EV, which isn't much at all. When pots get big, you need good reads and smart sense.

Note: I used these public domain normal tables to compute the probabilities here.

[Morrek]

Nice post.
How much would these numbers change as the SD increases? (too lazy/tired to run all the maths right now) Or, what I'm really asking is what would these numbers be for me, at 42 SD and 8ptbb/100?

[Cosimo]

[ QUOTE ]
The chance of a moderate (4PTBB/100h) winning player being at -800BB after 30k hands is 1:13821 (3.8SD off expectation). Likewise, the chance of someone who is -800BB after 30k hands is actually a moderate winning player is 1:13821. In other words, OP is not a moderate winning player.

[/ QUOTE ]

I know this sounds harsh, but it's what the math tells us. If the mathematical assumptions are accurate, then it's grossly unlikely that OP is a moderate winning player.

But I challenge those mathematical assumptions.

Thing is, every hand is not the same. You only get aces once every 221 hands; pocket pairs once every 17. Suited hands about 20% of the time. You hit a strong hand or draw also about 20% of the time. In order to make good money off of the hand, you need an opponent who also has something, which might be 20% of the time.

In other words, I think most of the profit in SSNL comes in these rare events. We only VP$IP once every 6 hands, hit it once every 5, and have an opponent once every 5, and it gets to the river once every 5. In other words, once every 750 hands, we have a chance to win or lose a lot of money.

OP had 40 of these events. If each event is +- a stack, then OP won 16 of these 40 events. This isn't a complete model, tho, since there's also a lot of money won or lost before pots get this far. TPTK raised on the flop is a big -$, uncalled CBs are a consistent chunk of +$, etc.

My point is, I think these rare events have a standard deviation to their own. If you run bad in these rare, huge pot, big-hand-against-big-hand events then that will carry your entire winrate with it.

There's another "set" of large pot events, more frequent but with lesser payouts. And a larger set of moderate pot events, and then a huge set of small pot events. Every time you raise preflop and it's folded to you, that goes in the small pot pool. Or when you hit a limped pot, bet out, and everyone folds. Happens all the time, right?

The "truth" is obviously a continuum of pot-size events. If you boil it all down to one number, you get the EV and SD that PokerTracker spits out, and we all know and love. But I this "one number" model for SD is weak. It doesn't matter how often you get aces, or if your hitting sets as often as you expect -- profit comes from having a strong hand when an opponent has a slightly weaker (but still strong) hand, and that's a rare event.

This is also why playing against fish that are willing to put their stacks in with weak draws and low holdings will add to your winrate so much. You get big-pot events far more often, and you don't need as strong a hand to take them down.

[Cosimo]

[ QUOTE ]
Nice post.
How much would these numbers change as the SD increases? (too lazy/tired to run all the maths right now) Or, what I'm really asking is what would these numbers be for me, at 42 SD and 8ptbb/100?

[/ QUOTE ]

I'll try to find the point at which you should expect to break even.

SD = EV
42 * sqrt(n) = n * 8
5.25 = sqrt(n)
n = 27.56

So, you have a 16% chance of being negative after 2756 hands.

2SD = EV
84/8 = sqrt(n)
n = 110.25

You have a 2.3% chance of being negative after 11,025 hands.

3SD = EV
126/8 = sqrt(n)
n = 248.06

You are 740:1 against being negative after 24,806 hands.

Let's say you play 2500-3000 hands a week. One out of every six weeks, you'll be behind. In a month, you're about 40:1 against being behind. After you play 25,000 hands, you are very unlikely to be behind.

[Morrek]

Thanks a bunch

[Mroberts3]

I would totally agree with Grunch on this one, you can never know for sure about luck, so you might as well look at your game. For what its worth, whenever I play really poorly or on tilt its when I play too agressivly and I suspect you might be doing the same thing. I know the feeling... some guy raises and you reraise PF with KK and the flop comes A high and somehow you find yourself all in on the flop thinking "He can't have A2 again, he just CAN'T" and of course he does.

[teamdonkey]

posted and then totally forgot about this thread. Since I'm a dork, i keep track of this sort of thing in an excel worksheet along with daily play. Calculations are pretty similar to those found in Homer's now famous thread

the upper bound of your confidence interval can be found with the equation:

=J-NORMSINV((100-L)/200)*K*(1/SQRT(I/100))

where J is BB/100, L is your confidence interval (is use 95 for 95%), K is your standard deviation/100, and I is your number of hands.

the lower bound can be found with the same equation, except it starts with J- instead of J+

so if i've seen 5BB/100 over 20,000 hands with a SD of 40BB/100, i can say with 95% confidence my true win rate is between 10.54BB/100 and -0.54BB/100.

What i've learned from doing these calculations is, they don't tell us that much. The OPs case isn't much different. He obviously isn't killing the game, but there's about a 1/3 chance that he isn't playing losing poker over that stretch either.