Complete Tests:
Test Results by Site:
For these tests we used over 1 million hands that were played at each of the online poker sites tested. These were the same hands used with the 'preflop and flop all-in bad beat tests'. This sample includes many hands that were not relevant to the test because they were not turn all-in hands. However, due to the large sample size, enough relevant hands to carry out testing were included in each sample.
The hand histories were purchased from major online poker hand history providers such as: Poker Table Ratings and Hand HQ. The providers acquired these hand histories by data-mining them directly from the online poker sites and have made them available to players. Players usually use these hand histories to gain information and therefore an edge on their opponents, however we have used them to test the poker rooms themselves.
Texas Hold'em was the form of poker that we tested because it is the most popular game and the easiest to obtain hand histories of. A variety of table sizes and stakes were used, although all of the of the stakes were micro-stakes because the cost of hand histories is much lower at the micro-stakes than at higher stakes.
The hand samples should consist of the majority of hands played over the period of the sample although not every single hand, since there are always some hands missed in the data-mining process.
Upon import there were some errors with faulty hand histories that couldn't be used and some duplicate hand histories - the numbers of these errors and duplicates were very small compared with the sample size (at most a few thousand hands) and would have a negligible effect on results.
Details of the hand samples used for each individual test are listed:
Site/Network | No. of Hands Sampled | Game | Players | Stakes | Date Range | Provider |
---|---|---|---|---|---|---|
PokerStars | 1 million | SNG | 9 | $3 + 0.40 | 5 Apr -1 May '11 | PTR |
Party Poker | 1.26 million | Cash | 9 max | 25c/50c | 30 Dec - 16 May '11 | Poker Hand Scout |
Ongame | 1.42 million | Cash | 5 max | 15c/30c | 6 - 23 Jun '11 | Hand HQ |
Merge | 1 million | Cash | 2 | 25c/50c | 4 May - 1 Jun '11 | PTR |
Entraction | 1.02 million | Cash | 6 max | 10c/20c | 16 Jul - 7 Aug '11 | Hand HQ |
Bodog | 1.1 million | Cash | 6 max | 5c/10c | 10 May - 10 Aug '11 | PTR |
The data was filtered to remove all hands that were not 'heads up, flop all-ins' - i.e. hands that were all-in on the flop, with exactly 2 players involved in the all-in. There were a very small number of unusable hands due to errors - these were also removed. The number of all-ins varied considerably with the type of game and stakes and are listed below.
Site/Network | Number of All-ins (Heads-up, flop) |
---|---|
PokerStars | 19807 |
Party Poker | 12170 |
Ongame | 13898 |
Merge | 3674 |
Entraction | 16609 |
Bodog | 10526 |
More information on these bad beat tests
can be found on the
Explanation, Dataset,
Method,
Results,
Conclusions &
Discussion pages.