Record Keeping & Data Mining

Poker Economy, Strategy

September 16, 2007

Keeping accurate records of your play is important for all poker players. It’s essential for tax purposes to know how much you won or lost in each session, but there’s lots of additional benefits to be gained by having comprehensive records. Specifically, good records will help you improve your win rate over time by highlighting situations where your results are consistently different from your expectations. This process will enable you to develop your game selection intuition.

I keep my records in the following format, first on paper at the cardroom and then on a spreadsheet on my computer: date, day of the week, location, game, structure, buyin amount, add-on amount, cashout amount, start time, end time, hourly rate and notes.

For tax purposes, all you really need to know is your net result and the date, but I find the rest of the information to be invaluable for later analysis. The notes field in particular is imporant. I use it to record my gut opinion at the end of a session as to why I won or lost. Usually, right after playing, you have a fairly good idea of what you did right or wrong and how the cards broke. But a couple of sessions down the line you’ll forget that critical information and just remember the $ result. So recording it saves a key piece of information you’ll want to have when you try to make sense of the data.

I take an additional step and color code each entry in the spreadsheet – green for a win, red for a loss.

Data Mining

Once you’ve built up a good collection of entries in your records, it’s time to let Excel (or the spreadsheet of your choice) go to work. A good starting point is to sort on the hourly rate field. Start looking at the games where you made the biggest (hourly) scores. Are there common elements that jump out at you? For example, I found out this way (to my surprise) that my hourly rate was higher at stud than at limit holdem, and higher at PLO than at NL holdem, when playing at similar stakes online. That came as somewhat of a surprise as my level of experience and evaluation of the my skills in those games would suggest the opposite result. But as we know game selection is a dominant fact over skill, so I shouldn’t have been too surprised. The key point here is that good records allow you to re-adjust your expectations to match reality.

If you want to take this data mining exercise to extreme lengths, you can have the spreadsheet calculate separate hourly win rates by location, day of the week, etc to get a very fine grained picture of how these vary. Be sure to raw calculate these hourly win rates (ie. add up the results and divide by the number of hours) rather than just averaging the session hourly wins.

I mentioned earlier that I color code each entry in the spread sheet. By sorting the sheet on any non-results column, the colors provide an instant visual picture of how that variable affects your results. The human mind is very good at spotting patterns in colors, and often times you’ll see things this way that you might never think to look for in the raw numbers.

It shouldn’t take too much imagination to continue you data-mining exercises. If you play the same game at multiple casinos, compare your results and see where you do better. Then plan on playing there more often. Look to see if your notes indicate you’re getting good results against maniacs or bad results. Figure out if your win rate goes up on weekends. There are tons of permutations, but the key point is that if you have accurate records, and spend time examining them and looking for patterns, you will find out important information that will help your game selection.

Like this article? Subscribe to the CardSharp RSS Feed

Comments are closed.