Monday, March 17, 2014

How to Conduct a Poker Database Review (A Complete Walk Through Guide).

pokertracker database review
Last time I talked about how to conduct a poker session review. Session reviews are important especially for newer players to make sure that they are putting into practice what they have learned away from the tables.

Also it helps them to start thinking about the game more independently. Session reviews are also useful for people who have just moved up and for anyone who is on a lengthy downswing. In all cases the main goal with a session review is to make sure that you are sticking to the game plan so to speak.

But in order to really improve your play you need to take it a step further and do regular database reviews in Pokertracker or Hold'em Manager. If you are unsure of what these programs are, then I would suggest reading this article first.

A database review is something that all players should be doing on a fairly regular basis. While the goal of a session review is primarily just to double check your play a database review involves an in depth analysis of your play and that of others in order to improve.

Our goal with a database review is to consider new lines (strategic plans of attack) in certain spots and to test their viability over a period of time. These new lines can be influenced by our study away from the tables or by reading books, watching training site videos etc. But most people fail to make any use of the absolute best resource that they have at their disposal...the biggest winners at their current stake!

If you have played at your current limit for any length of time it usually does not take long to figure out who the biggest winners are (i.e., the best players). They will usually be the ones giving you the most trouble.

They will most often be playing a TAG or SLAG style of play. They will be the ones 3Betting the crap out of you, constantly raising and floating flops and just generally being as annoying as possible especially if they are running good.

You may already have thousands of hands on these players with examples of how they play in many different spots. Why on earth would you not use this to your advantage?

And by the way, this isn't about copying somebody else's game. This is about looking at what some of the best players at your stake are possibly doing better than you, testing it out and then perhaps incorporating it into your own game.

These programs also provide you with an amazing opportunity to run filters and to find your exact profitability in almost any scenario imaginable. I am going to show you exactly how to do this.

Lastly, in this article I will be discussing how to do a database review in Pokertracker 4 because this is the poker database program that I personally use. However, all of this can be done in Hold'em Manager 2 as well.

And lastly, lastly I must warn you that this article is very long and took a long time to put together. But I have never seen this stuff covered before (which is crazy) so hopefully it is useful for you.

Let's get into some specifics!

Pokertracker Stats Comparison

The first thing that I always want to look at when conducting a database review is my general stats. This is because the raw numbers across a variety of different metrics tell us a lot about the general style of play and potential weaknesses.

They can never tell us the whole story though. Deeper hand history analysis is necessary for that. But they can point us in the direction of any key problem areas which is of enormous importance.

Specifically, I want to know what my numbers are across a variety of different stats and I also want to be able to compare them to the biggest winners at my current stake. I am not interested in the average numbers at my stake because the average player is a losing player.

I am only interested in comparing myself to winning players. In order to do this I find that using the "My Reports" tab in Pokertracker is the way to go. At the top of Pokertracker make sure these options are clicked:
How to Conduct a Poker Database Review (A Complete Walk Through Guide).
Next make sure that you have selected your player name and a report. If you have never used My Reports before then you will probably need to create a new one. Reports are highly useful because they allow you to save custom stats and filters for future use.
pokertracker 4 database analysis

Next it is time to specify what stats you want listed. Simply locate which stats you want included in the report in the "Available Stats" box at the bottom. Double click on the ones that you want included in order to get them saved to the "Report Stats" box above.

pokertracker player analysis

If you have just created a new report then there will be a few default stats already listed under "Report Stats." Often though the default stats/settings in Pokertracker are pretty useless. We want to customize them ourselves.

I use quite a few stats, 30 in total. I discuss many of them in a popular article from last year on how to optimize your HUD for today's microstakes games.

I won't be discussing them in any detail in this article but I will list all of the ones that I use below:

Pokertacker Holdem Manager database review
Pokertacker Holdem Manager database review
Pokertacker Holdem Manager database review
I regularly refer to all of these stats while playing and so it is a good idea for me to get familiar with them in my time spent away from the tables. I want to know two things:
  • What my numbers are for each of these stats
  • What the average numbers are for the biggest winners for each of these stats

Using the Best Players as a Measuring Stick

Knowing what my own numbers are for each of these stats and being able to compare them to the best players at my stake is very useful in figuring out where my leaks may be. I want to be clear though that there are many different styles of play that can win.

Just because my numbers deviate significantly on a certain stat from the biggest winners does not necessarily mean that I am doing something wrong. 

The only stat that actually matters at the end of the day is your winrate. If you are already among the biggest winners at your stake then it may in fact be that all of them that are doing it wrong.

So do not be too hasty to change something just because your numbers don't align with theirs. With that said, the handful of players who are among the biggest winners at your stake are usually a good measuring stick to look it. They are winning in a big way after all so they are probably doing quite a few things right. 

How Do You Find the Biggest Winners at Your Stake?

I should probably finally discuss how you are going to get the stats of the biggest winners at your stake! Besides an educated guess as to who they are are as discussed above we will probably never really know for sure.

The reason for this is because it can take as many as 100k hands to say anything conclusively about someone's winrate. You are simply unlikely to ever have a sample that large on anyone. 

So I often just use filters instead to narrow it down to the players fitting the stats which most of the biggest winners typically fall between. This is really simple to do in Pokertracker.

On the left side of the screen on the "My Reports" tab you will see "Filters" shaded in blue.

Filters > Edit Existing Expression Filters

This should bring up the "Edit Report Filter" box as seen below:
Poker database review

As you can see I have added a bunch of filters with some of the most common statistics for assessing overall play style (VPIP, PFR, Total AF and 3Bet). I have also added a total number of hands filter so that I can narrow the results to only come back with regs who I have a lot of hands on.

Using standard greater than (>) and less than (<) expressions you simply add in the values and generate the report. The numbers that I have used above represent the ranges that I think most of the biggest winners will fall under in a full ring microstakes cash game.

For 6max microstakes cash games, I would input the following numbers:

  • #VPIP# > 19 AND < 25
  • #PFR# > 16 AND < 22
  • #Total AF# > 3
  • #3Bet Preflop# > 4

This is by no means an exact science though. Breakeven or losing players can still have very good looking preflop and postflop stats. They may just have a major problem with tilt for instance.

But setting filters like this will give us the best chance of getting the results of the biggest winners. I will make note of any stat averages that deviate significantly from mine and then move on to the micro analysis which is discussed next.

Hand History Analysis and Filters

Stats are great and all but they often don't really tell us the whole story. If I told you that your major leak was that you are barreling the turn 60% of the time when you should only be doing it 40% of the time this is sort of useful information but also not so much.

You would know to cut down on your turn CBetting a fair bit but with which hands? Against which opponents? On what boards?

You could just consciously try to bet the turn about 1/3 less of the time when you are playing but it is much more useful to actually look over some hand histories first and find out which spots are profitable and which aren't.

In other words, with what hands, on what boards and against what opponents should we be barreling the turn and which should we not? We need to look deeper now at actual scenarios. The numbers can only tell us that there may be a problem in this specific area.

This is where the scenario specific filters that Pokertracker and Hold'em Manager provide are so valuable. Similar to studying the play of top winners this is another area in my experience that people do not bother much with. Which is totally crazy.

There is nothing scary or nerdy about this stuff. Let's have a look:

First in Pokertracker make sure that you are now in the "Statistics" tab at the top.
Poker database review
Make sure that your username is correct and set your dates and stakes on the left like before. Click on "More Filters" at the bottom to bring up the "Cash Filters" box.
Poker database review
From this box we can literally filter our results for almost any scenario imaginable including what cards we have, the board texture, the action in the hand, the stack sizes and more. However, when I am using these filters I use "Actions and Opportunities" as shown above at least 90% of the time. 

This allows me to set the action in the hand which is what I am most interested in. Let's go back to our previous example of barreling the turn. 

In order to bring up results only in spots where you barreled the turn you would go:

Actions and Opportunities > Actions and Opportunities - Turn > Turn Bets > Continuation Bet > Add to Filter > Save and Apply Filters
Pokertacker Holdem Manager database review
Once you apply the above filter it will now only show spots where you CBet the turn. By definition this means that you were the final raiser preflop, you CBet the flop and got called and then you CBet the turn as well. 

Since these spots do not pop up very often (or at least they shouldn't - barreling the turn every single time at the micros is a surefire way to burn through money) there should not be too many results unless your database is huge.

From here I will ask myself in each spot if it was a good situation to fire again given my hand, the opponent and the board. If you look closely enough you will probably begin to notice some similarities. 

For instance, recreational players don't like to fold especially on action heavy boards when they have a piece. Barreling them with air is often suicide for your winrate. 

On the flip side there are many weaker nits and bad regs who will happily lay down decent hands on the turn to continued aggression so that they can "wait for a better spot." This can bring big benefits to our winrate if we can consistently push them off made hands when we have air.

Calling 3Bets and Playing From the Blinds Analysis

If your database is big enough you can use the filters to run profitability tests in other areas as well. For instance a big leak for a lot of players at the micros is calling 3Bets too much. It is important to have some understanding of the profitability of different situations though before deciding that you have a big leak in this area or others.

Calling 3Bets is one of those spots that is a losing play for almost everybody, even the very biggest winners. However some people lose much more than others. The same thing goes for blind play. Everybody loses from the blinds. So don't freak out if you see a bunch of red for yourself. Instead compare how bad you are losing in these spots to the biggest winners. 

By the way please also do not conclude from this analysis that we should never call 3Bets and should fold every hand from the blinds. We have to play hands in these spots at least some of the time for balance considerations alone. 

You would be insanely exploitable if you decided to take such a harsh response. These situations are like a balancing act. Our goal is simply to lose the least. In fact it can make a big difference to your winrate if you can manage to have just small losses overall in these spots.

Let's get back to the issue of calling 3Bets though. If you have identified that you do indeed have a leak in this area, then you can dig deeper by going through individual hands. 

You should ask yourself if a 4Bet or a fold was more profitable in these situations. Often that is the case but sometimes it does indeed make sense to just flat the 3Bet. Assuming the flat was correct are you making any attempt to bluff raise them post-flop in these situations? 

If not then you are just throwing away money by calling preflop. You simply won't make enough when you finally hit a hand to make up for all of the times that you folded. From what I have seen this is the case for the the vast majority of people with the "calling 3Bets too much" problem. 

Sometimes it isn't even that they call 3Bets too much but that they never try to win the pot when they miss. These players will need to find the spots where they can push players out of pots. Once again it is often the weak/tight bad regs who they should be targeting. 

I can't provide you all the answers here because I don't know your specific game. But that is the point of this entire article. By running filters and comparing your game to other good regs, then you can hopefully fix a few of these problems yourself.

Split Testing 

Now that you have compared your stats to the best players at your current limit and identified specific scenarios where you might be able to improve, you need to prove that these new lines are in fact more profitable! 

Once again, this isn't about copying someone else just because they are having success from what you can tell. We still have to test their line over a large sample in order to prove that it is in fact more +EV.

To do this we employ a commonly known method of getting results in the SEO and blogging world called "Split Testing." Split testing is simply the act of comparing two or more different actions based on the data that you have collected on each of them. 

In the blogging and SEO world this might mean comparing the placement of certain ads on a website over several weeks in order to find out which spot gets the most clicks. 

With regards to poker, split testing means gathering and comparing data on two or more different actions as well. It is not always as simple with poker though because with the need to balance our range in many scenarios versus good thinking regs we simply aren't looking to take the same line all the time. 

Hence, even if we know for a fact that a certain line is lower EV than another we will still take it sometimes for meta-game balance. A full discussion of this topic is beyond the scope of this article but I discuss it more in this post and a lot more in my 2nd book as well.

But at the very least we can use split testing to improve the frequencies within our ranges. For instance if we know for sure that check raising with our strong hands is the best line to take versus certain types of opponents then we can exploit that by increasing the frequency with which we take that action. 

But for range balancing purposes against good thinking opponents we will still need to lead out or check/call against them from time to time as well. 

At the very lowest stakes this is not an issue at all though because most of the regs (and certainly everyone else) are totally clueless about what a range even is.

For instance those who have read my first book will know that I do in fact advocate just check/raising with your really strong hands 100% of the time against all the bad players at the lowest limits. This is because I know that this is the most +EV line to take since it gets the most money in the middle versus players who won't fold. 

And also, I know that I don't need to balance my range versus these beginner levels opponents. It is over their head and means nothing to them. So split testing is incredibly useful at these stakes. But it also applies perfectly very well to many preflop spots in particular at higher stakes and for modifying range frequencies as well. 

Split Testing in Practical Terms

In practical terms split testing in poker is actually pretty simple. We already have a vast amount of data in most cases on one approach (the way that we have been doing it all along). 

Now we just need to build a set of data on the new line that we have decided to take. It is definitely useful to document the date that we changed our approach and exactly what it is even if it is just changing up the frequencies within a range.

Like everything in poker the results will of course not come over night. In fact it will often take weeks or months before we have reliable data in some areas.

River 3Bet bluffs for instance will take an eternity to create a reliable sample on because how often does that spot even arise in the first place? Altering our 3Bet range though can give us results much faster because we get an opportunity to 3Bet much, much more often.

I am not a statistician but of course more hands is always better. A very large amount of poker players (like literally all) are simply unable to fathom the long run in this game. It is much, much longer than almost all of them think it is. 

Play more tables and play every day if you want to get there faster. If you follow the advice in this section but play only a couple of tables semi-regularly then expect it to take months or even years on some metrics. 

This stuff is for hardcore grinders who are very serious about their results in this game.

Mock Scenario of the Entire Process

1) You notice that the large majority of the biggest winners at your stake have a 4Bet ratio that is twice as high as yours.

2) You use filters in Pokertracker to review hands of 4Bet spots from big winners where their hand gets shown down so that you can figure out what they are doing it with.

3) You decide to change your range. For instance, you used to only 4Bet with AA and KK and sometimes with AK or QQ. You will now always 4Bet with AA, KK, AK, QQ, JJ and on occasion with a few bluff blocker hands like A4 or A3. You note the day that you started this and you stick to this new range every single session.

4) You come back at some point in the future when you have put in at least 10's of thousands of hands utilizing this new range and look at your results now when filtered for 4Bet winrate.

5) Filter for the 4Bet winrate before the changes were implemented and compare.

6) If the results are better than before then you should continue using this new range. You should scrutinize the sample size a bit more though just to make sure that it wasn't just a bunch of run good (you coolering the heck out of them a bunch) that caused the change.

7) If the results are worse then you should go back to your previous range. It is important to once again scrutinize the sample size though before coming to any conclusions. This is why we need 10's of thousands of hands at a bare minimum. Run KK into AA a couple times here and watch your winrate suffer in a spectacular way. 

Final Thoughts

I hope this article gave you a better idea of the sorts of strategies and tests that I run when conducting a poker database review. 

This is such a large topic that I could literally write half a book about it and there are many areas that I just don't have the space to go into as this article is over 4k words already. 

The best way to get started is to simply start messing around with filters and scenarios yourself though. 

I think far too many people use these programs (Pokertracker or Hold'em Manager) and then don't even use them at all other than to check their graphs and have HUD stats up while playing. 

This is totally ridiculous given the capabilities of these tools. But this might also be due to the total lack of information on this topic. I do not recall seeing an article like this ever written before.

Get yourself better acquainted with these programs and you will quickly see how much they can do for your game. Both Pokertracker and Hold'em Manager have forums and customer support that you can make use of if you can't figure something out.

You can get started with the trial version of Pokertracker 4 right here.

If you are wondering what a specific stat refers to in Pokertracker, then you can probably find the answer in my massive HUD setup article.

Let me know in the comments below what strategies and tests you use during a database review. Let me also know if you have any questions about anything that I talked about above.

Lastly, this article was exceedingly long and required a lot of work. If you enjoyed it then please do me a quick favor and "Like" or "Tweet" it below!


Pokertracker filters and database review


  1. I've been looking for exactly this article at this stage of my development. Thank you for continuing to turn out great content.

  2. Worked through adding the above states in a report, has the name of 2bet PF & Fold changed in PT4?

    1. Ya it is a bit confusing because PT4 lists it is "Fold to PF 3Bet After a Raise" in the Available Stats box. However, once added it will show up as "2Bet PF & Fold" in your report.

  3. Hi Nathan,
    I learned a lot from your session and I also did it this time as well. I couldn't find these stats in my report:
    2Bet PF & Fold
    Cbet F & Fold
    Cbet T & Fold
    Steal PF & Fold
    They should be the same on a mac as on pc version of PT4, right? or are they named something different now?

    1. Hey John,

      Ya unfortunately sometimes PT4 calls them something different in the the Available Stats box which is a bit confusing.

      Fold to PF 3Bet After a Raise = 2Bet PF & Fold
      Fold to Raise After F CBet = CBet F & Fold
      Fold to Raise After T CBet = CBet T & Fold
      Fold to 3Bet Preflop After Steal = Steal PF & Fold

      I will add these to the article also to make it more clear.

    2. Thanks Nathan,
      In my PT4, Fold to 3Bet Preflop After Steal is shown as Fold to PF 3Bet After Steal in the Available Stats. Maybe it shows different in your stats?

    3. It is Fold to 3Bet Preflop After Steal in the Available Stats for me and Steal PF & Fold in the report.

  4. Nathan, this is perfect! I have a question though. What WR would we might need to have, in order to consider ourselves in the "Biggest Winner" category now a days.. Here's my scenario, I keep on always trying to improve, always trying to look for some new profitable lines(I didn't know where to start til this post), but sometimes when I see my WR I feel like I'm doing just fine. I play at NL2 with 16bb WR over pretty good sample, somewhere around 50k+.. And always ask my self if this is enough? Or can I even looks for better lines to somehow improve this? Every time I look at my database these are the question arises in my head..



    1. Hi Jay,

      16bb at NL2 is probably among the top winrates today especially if you are playing a lot of tables at once. I think you could customize your game more to specific opponents at this level to squeeze out a little more EV. However, ultimately the same robotic lines versus the same beginner regs and terrible fish are what work the best. I would focus on moving up instead where the stuff discussed in this article becomes more important.

  5. Wow, great stuff here Nathan. Thank you for continuing to come out with such awesome advice for the micros.

  6. Hi,

    Hope you can help, I was following your instructions about finding the biggest winner in my stakes. I was stocked on the part after editing/adding expression filter. I mean after I add it what am I supposed to do? The filters were saved but it just goes back on the page where my stats are showing instead of players within the filters we set. Do I need to click on some 'Search' Button or something before it starts searching?


    1. I did btw un-checked the 'Filter on Active Player' option and it still shows only my stats instead of Player's stats.

    2. The only thing I can guess is that perhaps you used filters that didn't actually filter anyone out.

  7. hi Nathan
    I've done everything on this page and when i add the custom filters(Donk flop,Donk turn,3bet fold to a raise) the right side of my screen where my stats where goes also using the trial version of poker tracker you know any reason why this happens?

    1. pardon my english.i meant the stats dissapear and the place they used to be located turns blank

    2. I am sorry but I have never heard of anything like this happening. You should send an email to PT4 support.

  8. can you do stats comparison for hem2??

    1. I use PT4 so I don't write articles about HEM2. You can do all of this in that program too though.

  9. This comment has been removed by the author.