So in my last blog entry I touched the surface on the Great HUD Debate, you can read that HERE.   I thought it stimulated some great conversation in the comments, which is what it’s all about! 


In that blog, I promised I would followup with some tips and pointers for using your HUD stats so you’re not making common mistakes and misinterpreting the data, leading to bad decisions.  So without further ado, let’s get rolling.

1. Sample Size

By far the number one thing people mess up when interpreting their HUD stats is the factor of sample sizes.  The smaller a sample size, the less relevant the statistic tied to it.  Let’s start with a basic example… suppose someone says “The villain is a tight-passive nit, only playing 6/0 over 16 hands.”   Is a VPIP of 6% and a PFR of 0% tight and passive?  Yes.   Is it a fair read to characterize them as a tight-passive nit then?  No, certainly not, and the key here is sample size.  16 hands produces these %’s when the villain has voluntarily put chips in the pot 1 time, for a call rather than a raise.  Have you ever gone 16 hands without picking up anything remotely playable?  Of course you have.  Maybe the 1 call was calling a raise with 33 to set mine, or calling with AJs in position, totally reasonable actions.  Because poker players will go many stretches over the course of their playing career where they fold 16 hands in a row, this stat line produces virtually no read at all.  I wouldn’t even go so far as to say they might be tight/passive on such a small sample.  About the only thing I’d feel comfortable assuming is the villain is not a maniac, as I’m sure a maniac will find more than 1 hand in 16 to throw some chips around. 

This is kind of the basic level of the sample size phenomenon, and most people once pointed out to them, get it.  They almost invariably continue to make the sample size mistake however.  The key is in the layered nature of the sample sizes.  All HUDS display # of hands of data on a villain, but people tend to make the mistake of using this sample for all reads, when that’s not accurate.  Certain action opportunities occur more than others.   VPIP comes to a relevant sample faster than any other… every time your opponent is dealt a hand they will be confronted with a decision to voluntarily put chips in the pot or not (save for if they are blinded all in).  Other stats could be rather obscure and take a long time to come to a relevant sample.  For instance, something like “folds to river raise”… in order for an opportunity to occur, the player must have gotten to the river, bet the river, and then been raised, a sequence that happens relatively infrequently.  A villain who we have a 216 hand sample on we can use their VPIP as a somewhat reasonable loose/tight preflop indicator.  Their opportunities to fold to a river raise over those 216 hands they were dealt will almost certainly number in the single digits, and might even still be 0 times that they’ve bet the river and been raised.  So someone might say “I decided to bluff raise the river because the villain’s fold to river raise was 100%”, and if the total hands that read was based in is 216, it’s highly likely that the 100% they’re using for their read is actually 1/1 or maybe 2/2 in terms of actual opportunities… there may be good reasons to bluff raise the river in a spot, but that stat in this case is not one of them.

Any time you are using a stat to garner a read, be sure to consider the opportunities for that actual stat to get a gauge for relevance.  The more broken down a stat is, or the more obscure a situation it addresses, the much larger a sample of total hands you need to provide enough opportunities for the stat to carry weight.  Below is an example of where I’ve hovered my mouse over the “3-Bet vs. Button” stat to display the sample size… this stat reads as 13% in the HUD and total hands on the villain is 133, but the hover over reveals inadequate opportunities to comfortably use this particular stat in a read and doing so might lead me to making bad decisions:

2. Positional/Situational Relevance

This one comes up sometimes too… most general stats are all encompassing numbers.  So for instance take PFR… this stat is generated by 100% of all opportunities a player has to raise preflop, regardless of their position or situation.   Meaning, if a player 3-bets from UTG+1 vs. an under the gun open raise, or if they min-raise on the button when it’s folded to them, those are both preflop raises and both count equally into the PFR stat.  They are obviously quite different situations however, with the range of the former being significantly stronger than the range of the latter.  The key in this case is to consider the position or situation at hand when faced with your decision, and use any stats accordingly.

3. Incorrect Stat Usage

What I mean here can best be illustrated with an example.   Someone says “Villain is a LAG, playing 38/29 over a 1K hand sample.  They raise and I call on the button with ATs.  The flop comes T93 rainbow, they c-bet 2/3rds pot and I call. The turn is a jack, they bet again 2/3rds pot and I call.  The river pairs the 3 and they fire again for 3/4 pot.  I decided to call them because I can beat all bluffs and they are very LAGgy so I think I’ll catch them bluffing a lot... is this a good call down?”  The answer is, I don’t know… our hero has used a preflop set of stats as the basis for their river decision to call and showdown a bluff catcher hand strength.  What we really need to be considering at this decision point is post flop stats of some sort that are relevant to the situation.  So for instance if this villain c-bets the the flop 89%, the turn 88%, and the river 85%, then they are a frequently triple-barreling with what necessarily must be too weak a range and I think we may well catch them bluffing more than normal.  My normal default to exploit this will be to call them down more with decent bluff catchers such as this.  If however the villain c-bets the flop 89%, the turn 52%, and the river 28%, then while they are loose-aggressive preflop and fire the first bullet religiously, they are slowing down significantly on the turn (probably to made hands and draws mostly), and only 3-barreling post flop with a select part of their range… which probably does not include enough bluffs to profitably call them down for the river barrel and we can feel more comfortable folding.

4. Incorrect Stat Interpretation

It’s important to remember that most stats are just a frequency that something has or hasn’t happened.  In my experiences, most players fail to consider the context of the situation and how it might impact a range.  Let’s take a look at preflop 3-bet % as an example.   This stat is certainly one that’s impacted by positional and situational considerations as listed in #2 above… someone is often 3-betting an UTG open with a tighter range than they might a button opener for instance.  However beyond that, people tend to miss-interpret the range distribution as well.  Most people do something like this:

Villain is 3-betting 7% so their range looks like this:


They have basically taken the slider and moved it over to near 7%, grabbing the equity calculator’s estimate of the top 7% of all starting hands, and they stop there.  In reality though the range distribution may be quite different. Someone 3-betting vs. a solid player might have a more polarized range of strong and weak hands that produces an entirely different 7%, which may look something like this:


As you can see, these ranges look quite a bit different, and they will interact with different flop textures differently as well.  This is also why when someone asks me a question like “what’s a good 3B%?”, I cringe and hesitate to answer.  I feel like if I tell someone who’s 3-betting 4% preflop that a good % is 7% (or whatever), that they will try to force their 3B% up to 7%... but in doing so, be picking a lot of wrong situations.  A better method is to examine what their 4% currently looks like (it’s probably value heavy), and then work towards identifying additional opportunities to 3b profitably that they are currently only calling or folding.

I hope this helps you in your own quest to understand and apply HUD stats to your decision making process.  If you can start by getting your head around these 4 things, I think you’ll find your decisions improve dramatically.