Poker Software and Analysis

Using Poker Office 5, I performed a search for players who are suitable blind stealing candidates.

The results of my search can be seen to the right.

The most immediate characteristic that jumps out at me is that most of the "Folded SB to Steal" and "Folded BB to Steal" values are either 0.00 or 100.00. This strongly suggests that my sample sizes are probably one. That is obviously too small.

Most of the Fold statistics are 0% or very low. This means that blind steals will mostly fail against most players. However, the Folded to Continuation Bet values are almost all above 20%. While the individual samples are small, the fact all the Folded to Continution Bet values are so high suggests that if players follow up their steal attempts with continuation bets, then they should succeed often enough.

How To Improve These Results

This analysis has been hindered by a lack of data. I could get more data by playing more Poker, but that would requie a lot of time thta I don't have right now. Other alternatives for acquiring more data are:

• Buy hand histories. There are a number of online services that will sell hand histories for a reasonable price.
• Purchase a datamining tool such as Idleminer. These tools will automatically open online Poker tables and record hand histories. These can be imported into Poker Office.

Unfortunately Poker Office 5 does not provide sample sizes or confidence intervals for these statistics. If we know our sample sizes are very large then we should be able to be fairly confident of their accuracy. I shall have to make a feature request to Poker Office in the near future.