Repping Strength Based on Stack Sizes

[Proofrock]

[ QUOTE ]
Edit: BTW, I made a post a long time ago about small sample sizes and stats. TBH, I think you are dead wrong if you say these sample sizes are too small to be useful. In the case of this hand, the BB is ~2.5 times more likely to have a true VPIP of 10 than of 20, ~7.5 times more likely to have a true VPIP of 10 than 30, and ~2.9 times more likely to have a true VPIP of 20 than 30. I'll try to search for and bump that old post.

[/ QUOTE ]

8 hands? Yes, I'd like to see your post -- I know many of the wonders of statistics, but 8 hands is less than 1 full orbit and definitely isn't enough to say with any great degree of certainty that BB is a nit -- the uncertainty of your assessment has to be pretty large. For example, I usually run 16/11 at full ring MTTs, but it's not uncommon to find stretches where I fold 20-30 hands in a row. Yes, you can tentatively say "he's most likely tight," but need to be cautious about making a close move that relies on him being extremely tight.

[Sherman]

[ QUOTE ]
[ QUOTE ]
Edit: BTW, I made a post a long time ago about small sample sizes and stats. TBH, I think you are dead wrong if you say these sample sizes are too small to be useful. In the case of this hand, the BB is ~2.5 times more likely to have a true VPIP of 10 than of 20, ~7.5 times more likely to have a true VPIP of 10 than 30, and ~2.9 times more likely to have a true VPIP of 20 than 30. I'll try to search for and bump that old post.

[/ QUOTE ]

8 hands? Yes, I'd like to see your post -- I know many of the wonders of statistics, but 8 hands is less than 1 full orbit and definitely isn't enough to say with any great degree of certainty that BB is a nit -- the uncertainty of your assessment has to be pretty large. For example, I usually run 16/11 at full ring MTTs, but it's not uncommon to find stretches where I fold 20-30 hands in a row. Yes, you can tentatively say "he's most likely tight," but need to be cautious about making a close move that relies on him being extremely tight.

[/ QUOTE ]

Well, the search function on here sucks so I couldn't find the old thread. But I can re-post some specifics here.

The logic works as follows. If someone's "true VPIP" is .05, we know that the probability that he folds any given hand is .95.

So the math is quite simple. What is the probability of a person who folds 95% of the time, folding 2 times in a row? The answer is .95 * .95 = .90. That is, 90% of the time he folds twice in a row.

Now let us compare this person to someone who has a "true VPIP" of .10. This person has a .90 probability of folding and a .81 (.9 * .9) probability of folding twice in a row.

Given this, we can divide .90 by .81 to express the likelihood-ratio that a person is more likely to be a .05 than a .10 true VPIP given that he folded two hands in a row.

It works out that a person who folds two hands in a row is mathematically 1.11 times more likely to be a true .05 VPIP than a .10 VPIP.

Now that may not seem like much of a difference, but consider that we have gained this information off of only 2 hands! Beyond that, we also know that the difference between a true VPIP of .05 and .10 really isn't that different effectively. So let's compare a person with a true VPIP of .10 (very tight) to a person with a true VPIP of .30 (somewhat loose) after folding two hands in a row.

It turns out that a person who has folded 2 hands in a row is 1.84 times more likely to be a .10 true VPIP than a .30 VPIP. This same person is 3.6 times as likely to be a .10 true VPIP than a .50 VPIP!

All that information gleaned from only 2 hands! I think that is quite impressive.

In this case, we have a person who has folded 8 hands in a row. If his true VPIP is .10, he has a .43 probability of folding 8 hands in a row. Whereas if his true VPIP is .30, the probability of him folding 8 hands in a row is .058. That means it is nearly 7.5 times more likely that the BB in this hand has a true VPIP of .10 than .30! Pretty impressive IMO.

Now admittedly this analysis doesn't take account for position and many other things, but even if it improves your read on an opponent ever so slightly on average it is helpful. I wish I could find the original post. Maybe later today I will make a re-post that includes all of the analyses I have done on the topic.

Sherman

[Stumpy]

Sherman,

My only comment would be position is pretty key for a lot of players. I'm sure I raise 2x or more from LP as EP, so taking my UTG,UTG+1,MP1 folds and deciding I'm 10% VPIP is pretty dangerous I think.

I keep track of my table stats, and I'm surprised how much they vary table to table. For really tight players this would be less true, but still significant.

Edit: It may just be the Search user that sucks.
Making reads based on minimal information

[Sherman]

[ QUOTE ]
Sherman,

My only comment would be position is pretty key for a lot of players. I'm sure I raise 2x or more from LP as EP, so taking my UTG,UTG+1,MP1 folds and deciding I'm 10% VPIP is pretty dangerous I think.

I keep track of my table stats, and I'm surprised how much they vary table to table. For really tight players this would be less true, but still significant.

Edit: It may just be the Search user that sucks.
Making reads based on minimal information


[/ QUOTE ]

Stumpy,

That's you. You might be characterized as a relatively good player. There are still a lot of people out there that don't pay attention to position.

Beyond that, something is better than nothing 2 times out of 3. So I'd take the minimal information that we do have, and apply it to the best of my knowledge. Just dismissing a small amount of information as nothing because it is "small" is just dumb.

Lastly, thanks for the link and for being a dick about it. Everyone knows that the "search" function that these forums use is subpar relative to standard search engines "google, etc."

[Proofrock]

Thanks for the link(s). I particularly liked the uNL discussion because I'm a math nerd.

[hamnegger]

interesting bc ive been using this too.it is a +ev play bc even if ss goes ai we can suk out and bb is prob very strong if he plays back so we have easy fold

[Stumpy]

I didn't mean for my comments to sound too negative.

Much like a lot of things in poker, you tend to remember the bad results?
I've decided to fold in spots because a guy has a tight image after 10 hands and then he goes nuts and I wonder if I should have put as much weight on the stats as I did. Your post really pushes for the opposite, and makes a pretty convincing argument.

If I could do it easily, it'd be interesting to plot my VPIP stats for 10 hand segments and see what the distribution looks like.

[Sherman]

[ QUOTE ]
I didn't mean for my comments to sound too negative.

Much like a lot of things in poker, you tend to remember the bad results?
I've decided to fold in spots because a guy has a tight image after 10 hands and then he goes nuts and I wonder if I should have put as much weight on the stats as I did. Your post really pushes for the opposite, and makes a pretty convincing argument.

If I could do it easily, it'd be interesting to plot my VPIP stats for 10 hand segments and see what the distribution looks like.

[/ QUOTE ]

Yeah, the disadvantage of my method is that it does not include any measure of variability (i.e. Standard Error, SD, variance, range, confidence interval, etc.).