|
1,360,000 hands. That is how many hands we have tracked on PokerStars.com. We have been monitoring all-in situations at PokerStars.com for the past year and we have come to the conclusion that the games at this site are unfair and biased toward poor play. Read on to see our methodology.
How we tracked it:
By using 15 different computers running PokerStars software and each monitoring five tournament tables at once, we were able to isolate nearly 1.4 million hands where two or more players were all in against each other preflop. These hands were then grouped into three categories: 80/20 (i.e. pair vs. pair), 70/30 (i.e. AK vs. AQ), and 60/40 (i.e. JT vs. 45). We broke these matchups down and analyzed each. After reviewing all hands, we would expect the higher of two pocket pairs to hold up 81% of the time and tie about 0.5%. We would expect a high card hand matching neither of the opponent's cards to hold up roughly 59% of the time, with about 1.1% ending in a tie. Finally, we would expect dominating hands to win 71% of the time with 1.1% ending in a tie. The margin of error for win percentages is +- 1.7%.
The results:
- In coinflip situations PokerStars worked out as expected, with a tiny advantage to the pocket pair of roughly 1.5%.
- In 60/40 situations PokerStars worked out roughly as expected, favoring the weaker hand only .8% more often than expected, well within our margin of error.
- In 80/20 situations we saw a little bit of deviation. The smaller pocket pair actually won 26.3% of the time, and the higher pair winning only 72.7% of the time. This is a well outside of our margin of error by nearly triple!
- In 70/30 situations things get even worse. The lesser of the two hands wins a whopping 41% of the time, a full 10% more than it should! With more than 475,000 hands to analyze, this is more than a simple statistical anomaly. This is downright fradulent.
Other Findings:
We became curious about what was going on so we looked closer at many of the 70/30 hands (a three-outer). We found that hands such as Kx vs. Kx were within the standard deviation, as was Qx vs. Qx. We did not have enough hands of Jx and lower in 3-outer situations to analyze. Upon looking closer at Ax hands we found that the weaker Ax beat the better Ax (i.e. A4 beating AK) about 4% more than it should. We also found than Ax vs a pocket pair beat the pocket pair more than 12% more often than it should! So clearly aces flop too often, right? Wrong. Simply looking at flops shows an even distribution of aces. However, when players are all in and need an ace, it tends to appear. When no ace is needed it does not appear, balancing out the discrepancy. Furthermore, we found that when two players held an ace, one or both of the remaining aces would appear on the board over 30% of the time. This is a huge discrepancy, nearly double the total of about 19% expected. We can only guess that this is to induce action, as this statistical anomaly does not occur with other cards.
Conclusion:
PokerStars is NOT a fair site! We do not recommend that you play there until they address these statistical anomalies. If they contact us about these things, we will confront them with the hand histories we recorded. Regardless of their response, our study is statistically significant enough to warrant extreme caution when dealing with PokerStars. We are currently performing the same tests with PartyPoker, ParadisePoker, and PokerRoom.com. Early results indicate that ParadisePoker is completely random and in line with expectations, PokerRoom looks good as well. We are investigating PartyPoker closer and will withhold judgement until we feel we have enough hands to make a sound analysis. If you must play Poker online, we recommend Paradise Poker or PokerRoom for the time being. Good luck at the tables, and may they not be at PokerStars!
|
|
