Are longer sessions better or worse? (Non-grumpy content)

In late 2006, Card Player columnist Steve Zolotow wrote an interesting series of pieces on the importance of record-keeping in poker. (Part 1, Part 2, Part 3.) I had already been keeping records adequate for tax documentation, but his writings prompted me to refine my spreadsheets (how did the world get by before there was Excel???).

But for some reason, I never got around to doing the first analysis that Zolotow recommends, which is looking at results by length of session. Well, now I've done it. There's not really any reason anybody should care about the results, but I did the hard work, so I might as well share it.

The first chart above (click to make it bigger) shows data of dollars won/lost versus hours of cash-game play (I excluded tournaments), for the first eight months of 2007; I got tired of entering it after that, and thought that I had a big enough sample.

The first thing you might notice is that there appear to be as many dots in the negative range as the positive. That is so. I've seen and heard other pros say that they book wins something like two-thirds of the time. Well, for whatever reason, that isn't how I run. In fact, I'm at 51%/49% right now. That doesn't bother me, though, because as the chart also shows at a glance, my average win is quite a bit higher than my average loss.

Zolotow advises: "Look at your five biggest wins and five biggest losses. If you played significantly more hours during those losses than you did during your wins, you have a major discipline problem. If you perform this simple exercise and change your negative pattern, it will be the most important thing you can do to improve your bottom line."

I'm pleased to see that I pass this test. My five biggest wins are in sessions varying from 2.75 to 6.25 hours. My five biggest losses are between 1.75 hours (that was a positively brutal, intense beating that session!) and 4.5 hours.

My sense going into this data analysis was that I would see no significant correlation between length of time played and result. And that basically turns out to be correct. To be precise, the correlation coefficient for the data set shown in the first chart above is 0.16. A correlation coefficient is a way of expressing how related two sets of numbers are. It can range between 1.0 (perfect positive correlation; the data points would line up exactly on a line slanting up and to the right) and -1.0 (the opposite). So I have only a slightly positive tendency for longer sessions to be more profitable.

Next I ran the analysis you see in the second chart. It's the same data set, but with the absolute dollar values converted to dollars-per-hour rates. The correlation here is even weaker: 0.12, which means that I make only marginally more per hour during longer sessions.

You can also see that I don't go for the marathon sessions that some other players make famous. It's not my thing. On those rare occasions that I've gone past the 12-hour mark, I feel crummy and I don't play well. It's just not worth it.

My conclusions:

(1) As I suspected, the hard data confirm that I'm reasonably well disciplined about quitting when things are going badly for me that day. I definitely need to improve in this, because I can't count the number of times I've spewed off one extra buy-in (usually $100) even after I recognize that I'm playing poorly, or the competition has the best of me, or I'm just impossibly unlucky that session, or whatever. I'd love to have back all of the Benjamins that I've lost that way. But at least I can honestly say that I don't let the situation turn from "problem" to "disaster" before getting out of Dodge on the days that the donkeys are having their way with me.

(2) Between roughly two and eight hours, there is basically no correlation between how long I'm sitting at the table and how much I can expect to have made. Some days I get on a huge rush early, then stagnate for a few hours, neither making nor losing anything, until I get up and leave. Other days it's the reverse, and I toil and sweat with no headway, until four or five hours in I get smacked in the face by the deck and run my stack up ridicuously fast. Other times, it's just slow, steady progress. Other times, it's up and down, up and down, and whether I book a win or a loss is just a matter of where the pendulum is when I clock out. It's absolutely impossible to predict, and it appears not to matter.