Tuesday, September 26, 2006

Log analysis

I was playing an STT tonight. I was in the money, lying 3rd, short-stacked and high blinds. I went all-in with AQ, got called and beat by A6. This really miffed me. A6 is about 25% or a 3/1 shot to beat me. I couldn't help think how often this happened. Not just in hands I was invloved in, but others too.

This got me thinking. I wrote a little application to analyse my logs. My logs had about 6,000 hand histories. My application found about 150 instances of Ax in a showdown with Ax. The results are listed below. You can right-click on the table and export to Excel (in Internet Explorer) to analyse the results.

Overall it's quite worrying. 50 of the 150 results were unexpected. ie. the underdog won. This equates to a proability of 33% instead of the expected 25%. This is quite a significant difference. There were 50 instances of AK v Ax. Again, in theory, the probability of Ax beating AK is around 25%. But from the results below, the actual result is 33%.

If you want a copy of the application, send me an email at jules.anon@googlemail.com - the application is actually an ASPX webpage written in VB.NET. So, you'll need to run it on IIS locally.

Logs Query
Hand1 Hand2 WinningHand Expected
AK AQ AK -1
AK AT AK -1
AK A5 AK -1
AK A5 A5 0
AK A2 A2 0
AK A6 AK -1
AK AT AK -1
AK A7 AK -1
AK A7 AK -1
AK A8 A8 0
AK A8 AK -1
AK A7 AK -1
AK A7 A7 0
AK A9 AK -1
AK AQ AK -1
AK AQ AK -1
AK A4 AK -1
AK A9 AK -1
AK AT AT 0
AK A6 A6 0
AK A7 AK -1
AK AT AT 0
AK A6 A6 0
AK A7 AK -1
AK A4 AK -1
AK AJ AK -1
AK A2 AK -1
AK A2 AK -1
AK AJ AJ 0
AK AJ AK -1
AK A8 AK -1
AK AQ AK -1
AK A3 AK -1
AK A5 AK -1
AK A8 A8 0
AK A7 A7 0
AK A9 A9 0
AK A3 A3 0
AK AQ AK -1
AK AJ AJ 0
AK AJ AJ 0
AK A3 AK -1
AK AT AK -1
AK A3 AK -1
AK AQ AK -1
AK AJ AJ 0
AK A9 A9 0
AK A7 AK -1
AQ A8 A8 0
AQ A6 AQ -1
AQ A9 AQ -1
AQ A6 AQ -1
AQ AJ AJ 0
AQ AT AT 0
AQ A7 AQ -1
AQ AJ AQ -1
AQ A6 A6 0
AQ A4 A4 0
AQ A7 AQ -1
AQ A7 A7 0
AQ AJ AJ 0
AQ A3 AQ -1
AQ A2 A2 0
AQ A6 AQ -1
AQ A9 AQ -1
AQ AJ AQ -1
AQ A9 AQ -1
AQ A6 AQ -1
AQ A9 AQ -1
AQ A7 AQ -1
AQ A3 AQ -1
AQ AT AQ -1
AQ A3 AQ -1
AQ A4 AQ -1
AQ A4 A4 0
AQ AJ AQ -1
AQ A9 AQ -1
AQ A9 AQ -1
AQ A6 A6 0
AJ A3 AJ -1
AJ A9 AJ -1
AJ A8 AJ -1
AJ A9 AJ -1
AJ A7 A7 0
AJ A6 AJ -1
AJ A8 AJ -1
AJ A6 AJ -1
AJ A5 A5 0
AJ A3 A3 0
AJ A7 A7 0
AJ A8 AJ -1
AJ A6 AJ -1
AJ A9 AJ -1
AJ AT AT 0
AJ A5 A5 0
AJ AT AJ -1
AJ A9 A9 0
AJ A4 A4 0
AJ A7 A7 0
AJ A5 AJ -1
AJ A5 AJ -1
AJ A5 A5 0
AJ A5 A5 0
AJ A5 AJ -1
AJ A7 AJ -1
AT A9 AT -1
AT A8 AT -1
AT A8 A8 0
AT A2 AT -1
AT A4 AT -1
AT A4 AT -1
AT A8 AT -1
AT A8 AT -1
AT A3 A3 0
AT A5 AT -1
AT A8 AT -1
AT A3 A3 0
AT A8 AT -1
AT A4 A4 0
AT A2 AT -1
AT A6 AT -1
AT A8 AT -1
A9 A7 A7 0
A9 A4 A9 -1
A9 A4 A9 -1
A9 A4 A9 -1
A9 A6 A9 -1
A9 A4 A4 0
A9 A3 A9 -1
A9 A4 A4 0
A8 A3 A8 -1
A8 A4 A8 -1
A8 A2 A2 0
A8 A5 A8 -1
A8 A6 A8 -1
A7 A6 A7 -1
A7 A4 A4 0
A7 A2 A7 -1
A7 A4 A7 -1
A5 A2 A5 -1
A5 A2 A5 -1
A5 A4 A5 -1
A5 A4 A5 -1
A5 A4 A5 -1
A5 A4 A5 -1

1 comment:

Posh said...

Rum

according to the hendonmob site, AKs v A6s (about as bad as it gets for the rag ace while removing the flush argument as both are equally likely) is 70-30 so your 33% result is only 10% out and over a 150 sample is well within standard variance.

nothing to worry about I'd say.
dibs