Earnings rates in $6/$12 and the perpetual question of when to leave

One of the questions I struggle with as a poker player is when to leave.  Sometimes it’s obvious when to leave… and sometimes it’s not.  One of the grandfathers of poker theory, David Sklansky, says that you shouldn’t "quit winners", or at least I believe he said that.  For years I have been obsessed with that concept.  I always intend to bail on a game when I’m at a ‘high water mark’, but I always think, "Hey, if I was good enough to win this much, then I must be able to win more, yes?"

You’d think there’s nothing new under the sun in poker analysis, but today I tried a new data gathering technique.  I went to my local card room and tracked my stack size in a $6/$12 limit hold’em game every hour on the hour.  I started out with $240 and played for 8 hours.  Here are the results:

  • 10am: $240 (start)
  • 11am: $199
  • 12pm: $129
  • 01pm: $329
  • 02pm: $327 (during this hour my stack hit $480, and then I took a horrible beat at 2:55pm)
  • 03pm: $343
  • 04pm: $248
  • 05pm: $345
  • 06pm: $248 (leave)

At 6pm I realized the table had almost entirely turned over, I was looking at a number of talented faces, and they were all less tired than me.  So I left.  Here’s a little graph showing my profit and then a running profit average over my entire session.

The large blue bars are my absolute profit, and the red bars are a running average profit.

Here’s another graph.  This is my total earnings per hour, from the start of the session till the moment of measurement.  What I get out of this second graph is really just a hypothesis: that I am playing best somewhere in my third or fourth hour of play, and I’m too tired to recover after that.  There are factors besides simple session duration that would affect this, for example, table turnover, but it’s something worth considering.

For years I have repeatedly said that I really should be leaving a table if I’m "up" and at least switching tables and banking the money, or switching games altogether.  Perhaps if I continue to gather this sort of data I will be able to come up with a more accurate picture of my best session period, and provide some structure about when to stop.

One structure that is applied in the stock market is to sell when your purchase hits a predetermined profit.  I know a stockbroker that calls his clients and tells them to sell a particular fund whenever they make 20%.  The funds are heavily optioned so that the principal is safe, but he figures, "Hey you made 20%, get the heck out of dodge, what, are you going for 25%?"

So while I was writing this, it occurred to me, "Hey, I’ve actually got a lot of data like this already on file.  I’ve kept every session of live poker I’ve ever played stored on www.pokercharts.com."  So I pulled out about 110 sessions of limit hold’em I’ve recorded since mid-2004 when I became very serious about poker.  I then plotted a scatter chart of the total session profit against the length of the session.  The longer sessions (>18 hours) were really long weekends in AC or Vegas.  One could argue that they were multiple sessions.  One could also argue that I didn’t sleep much and wasn’t truly rested.  Without trying to bias the data, here’s what I came up with.

I’m really on a roll now, and will be doing an analysis that normalizes across stakes and should remove any extra weighting for higher limit games.