The limp re-raise may have thrown me off. I highly debated about shoving turn, but pre-flop threw me, and then by the river, I just had a hard time seeing this guy bluff enough even despite the board. What's the MDA on this line? lol I'll post results later.

I fold pre vs. his sizing/line/current stats.

As played I'd honestly call it off...

Interested in results.

What is the 11/11 sample over how many hands?

Giving what I know about limp re-raise ranges I'd call. That runout is overbluffed as well in 3BP OOP PFR (9 high board river overcard)(flop 2t missed FD).

Only thing that is worrisome is his exact sizing is B25/B25/Jam but the other data points to a call.

Can't decide, but it really looks like he doesn't want you to fold, plus AQ/KQ got there if he ever has those. I would just fold pre against someone I know is big nit.

35 hands at the time I played. That's why I said, likely nit. I don't know for sure at that point.

This is why statistics is so counter intuitive.

I was talking to TBJ about this and he made a very good point. It's way more likely that this guy is just a regular running bad than a nit because there are no 11/11 regulars. And there are no 11/11 fish over 35 hands because

A. There's no VPIP/PFR gap
B. Fish usually VPIP more than regs

So we can deduce that this is either.

1. A regular
2. A bot

I think it's way more likely to be a bot given it's Ignition. Given bot tendencies I think you have to call river although they do have a decent amount of AQo/KQo in their range given preflop. Bots tend to trap strong hands in general.

Very cool hand.

50-200nl data, I assume nits are even less existent at 1k

97.5m hands from players with VPIP and PFR < 30.0
52.6m for 25.0
2m for 20.0
394k for 17.0
42k for 13.0

So villain is almost certainly a 20-29 VPIP reg/bot

This is GG bot limp RR (not as SB) and saw showdown


ACR bots almost never limped, so who knows for Bonition

same but regs

Fwiw my not-a-bot stats a few days ago:



I understand (and applaud) the vigilance, but don't understand the certainty over 35 hnds

You will see some 18-20ish range VPIP tight players that try and feast on whales and gamblers at these stakes. They are mostly tight w/ their pre-flop range. Doesn't mean they won't bluff a lot post flop.

I'll explain my thinking in this hand tmw and post results. I have a lot going on today.

The most interesting part of this hand to me is actually preflop although postflop is also interesting.

If we define a nit as a VPIP<20.

I looked at this hand more in depth in my PGC and came to the conclusion with the help of TBJ's data that it is almost with 100% certainty that this player has VPIP of between 20-29 and not <20.

Looking forward to results tomorrow.

Fold pre; the limp reraise is AA most of the time

As played, not sure how you can fold given the flop SPR. I guess we arrive at this river with a decent amount of QdXd that floated flop and turn due to villain's small sizing.

Sent from my SM-A146U using Tapatalk

I recently took a deep dive into the topic of Bayesian-adjusted HUD stats.

Method:
In short, I take the population data as a prior. Then adjust that prior using the observed sample. That gives us the posterior.

I create priors by splitting the population data into winning/losing players. Then I calculate the mean and SD of the population's HUD stats. Then I use a beta distribution to update the population stats with the player's observed data.

If we sample from the winning players, after 35 hands their 11/11 stats become:

• VPIP 22.8% ± 4.1%
• PFR - 18.7% ± 2.6%

If we sample from the population, then it's closer to:

• VPIP 17.3% ± 5.3%
• PFR - 16.9% ± 3.7%

My dataset comes from a nittier site. Looking at TBJ's data, it seems ignition is a lot looser. So my method has probably underestimated their stats!

Could you walk us through how you arrived at that conclusion, DDP?

Generally speaking, when you have reason to believe you're playing a fish, you should be more likely to believe outlier stats.
When you have reason to believe you're playing a reg, you should be more suspicious of outlier stats.

I am definitely deferring to you here but I used Bayes Theorem so P(A B) = P(B A)*P(A)/P(B)

I wrote it all down here on page 3.

https://forumserver.twoplustwo.com/1...44/index3.html

I used your calculator to get the probability of being a nit and then probability of being reg and then plugged in the numbers from TBJ's data. I'm not sure I completely did it correctly so let me know if any of my inputs are incorrect.

Can you explain what the charts mean in laymen's terms? I appreciate you taking the time to do this.

So if he is a regular aka winning player. Then his VPIP is between 18.7% and 26.9% according to your data right? And if he is again a regular. His PFR is between 16.1% and 21.3%.

Is that right?

Would it be fair to say that the odds of him having a true VPIP of <20 is less than the odds of him having a true VPIP of 20 or higher? I'm trying to figure the odds of this guy being a nit (<20VPIP) or not. I'm claiming that the odds of him being an average regular is higher than the odds of him being a nit.

Also TBJ's data is from 50nl-200nl so on average there are less nits/more aggro players at higher stakes which also makes me think villain is more likely to be a regular than a nit.

Yeah so my thinking is because he has no VPIP/PFR gap over 35 hands we should be very inclined to believe he is a regular, do you agree with that?

Then once we know that he is a regular we need to figure out if he is an average regular or a nitty regular.

It's just Stars but could be more recent/higher stakes than yours

I didn't filter for 6 handed which could skew things a lot

Personally I don't think whether or not he's a nit is as important as whether or not he's a fish.

Why split pool into winning and not winning player when that's one of the slowest converging stats?

If you want to do this analysis, population should be divided by very fast converging stats like VIPP PRF or limping %(something relevant for this example) or preflop/flop sizing they use, so even with small sample you know which group particular player belongs.

Idk if limping from MP is something regs do often on this site, but for me this is one of the strongest indicator it's a rec.

Putting player in the right subcategory is the most important step in the analysis. If you put players in really broad category and you have small sample on him, as result you just get population averages. Making whole analysis almost poitnless.

In this example imho you should put this player into the category of players who have limping range on MP and who play 100bb deep (if he did buy in with max).

Fair criticism Haizemberg. The reason I did it this way was because this tool wasn't built for this question - it's a generic tool meant to handle a wide array of HUD stats.

You could definitely get a lot more accurate by plugging in more known info, like the fact that they limped in EP.


Not really, it depends how strong your priors are. In this case, the posterior is pretty far from the population. It's still gonna converge a lot faster and be more accurate than taking the HUD stat at face value.

If you had to guess future salary of 21yo and you have this info
-he is from US
-he is in Harvard
Would you look at distribution of salary of
A-US in general
B-Harvard students after they fish
C-people who are 6ft tall

Ofc B will give you the best guess, A will give you generic guess with huge variance, C is equivalent to putting this player in winning player distribution he might be he might not be. B reduces population the most (in relevant way).

When you do this type of analysis you want as many constrains/evidence as you can get.

Yes taking HUD at face value is bad, but assuming that guy who limps is random representative of the pool is not ideal either imo.

I understand your criticism but we should just be grateful there is someone on the forum who puts a professional level of time and effort and presentation into understanding this subject.

I said this in my PGC but understanding statistics is a completely separate skill set from understanding poker fundamentals. If you are labeling an 11/11 over 35 hands that limped as a likely nit and playing that person as if he has nit tendencies when in reality there is a <15% chance he is actually a nit then that is the most important concept to take away from the thread.

As as aside, analogies don't work in poker. I hear people make analogies in YT training videos (you might be able to guess who I'm talking about) and they fall short literally every single time.

Poker is about game theory/data/formulas/statistics and to a much lesser extent psychology. Analogies just confuse the viewer/student. Just say what you mean.

TY again Tom for the A+ post.

I guess there is limp/raising going on from bots at 500nl on ignition right now (I would assume 1k also). I don't know what the range construction is though. Looking at this hand again with that information makes me think this was a bot seeing the postflop sizings.

Have you seen more limp/raising pf from "reg looking" profiles?

Seems like part of this discussion revolves around drawing inferences from small sample sizes. Not to oversimplify or discount the great stuff already put forth, but a good rule of thumb is that a player with loose stats over a small sample is more likely to actually be loose than a player with tight stats is to actually be a tight player.

Let's take a player with a 30% PFR over 5 hands and compare that with a player with a 0% PFR over 5 hands. The chances of being dealt a top 30% hand five consecutive times is 0.002%, while the chances of being dealt a bottom 70% hand 5 consecutive times is 17%.

Since it's super unlikely for the 30% PFR player to have actually been dealt those hands 5 consecutive times, they are most likely playing even wider (unless they are on a heater). On the other hand, it is relatively much more likely that the 0% PFR is just card dead and will eventually be dealt more playable hands.

Sent from my SM-A146U using Tapatalk