STOP PRESS NEWS: TrueGambler.com released a statistical report for online blackjack on 1st September which casts doubt on the fairness of three of four well known software brands including Microgaming. There are doubts about the accuracy of many of the calculations at the moment because the report doesn't explain them. If it is found to be correct, there seems little doubt that subtle cheating happens as long as the data is accurate. Millions of hands of data was recorded and analysed. Although the payout rate would not be altered very much, True Gambler would strike a huge blow for the online casino industry if the analysis is correct. As explained earlier, there is always the possibility that the payout rate is tweaked slighty by the casino software companies to increase profits. If the report is shown to be accurate, many more online casinos will become blacklisted here.
TrueGambler used a software client to record the hands from over a hundred players using the software. The data was sent to a central server owned by True Gambler and collated and later analysed. There are current doubts about the way the software handled split pairs which could explain why two online casino software companies, Microgaming and Random Logic, had similarly skewed distributions of splittable player cards. If this wasn't a problem with the True Gambler software, then both look to have cheated in a similar way but this seems unlikely. For these two mentioned software companies, the results show significantly less pairs of splittable initial player cards such as AA than expected in a random deal.
TRUEGAMBLER UPDATE (October)... An updated report was released a couple of weeks after the first. It identified a problem with the analysis and corrected other mistakes. There was indeed a problem with the software reporting splittable hands in the collection process which was later corrected. The first report only removed the problem splittable hands but this obviously biased the results for the players' initial hand. The new report rightly only includes the data since the problem was resolved but I don't understand why it wasn't done in the first place when the initial report was written.
The new results exonerate Random Logic but Microgaming still has suspicious results in all three tests although one seems to indicate a higher than normal payout considering the win/loss/push ratio but doubles and splits are not really considered completely and also it only takes into account around 83% of the hands (only the ones that the players didn't bust). Maybe there is a problem calculating the theoretical probabilities or some other problem with the data that can't be ruled out yet. The calculations are not verifiable until all the data is released.
I will comment on the results of the player initial cards (the first two dealt), which is another of the three tests in the report. It seems normal apart from hands that include aces which make the results virtually impossible in a random deal. It looks as if there is an undetected bug with the TrueGambler recognition software. In particular, there seems to be far too few Jacks with Aces as a player's initial hand and far too many tens with Aces. Since it makes no difference in blackjack whether a card is a jack or a ten, it seems likely that it's a bug in the recognition that went undetected but why it only involves hands with aces is a mystery. Many Microgaming casinos have different card graphics and some may not be accurately recognised. There are also too few AA and too many A4's although to me this looks likely to be another bug in the recognition. Other than that the results look quite normal for the player initial hands.
The only other software with suspicious results (of the four tested) is Odds On software. It seems the dealer has too many blackjacks. This is a similar result to Microgaming so maybe this is caused by another bug in TrueGambler's OCA (Online Casino Analyser) client software or the way the data was analysed. Counting blackjacks seems simple enough but it's not that easy for these two softwares because they are the only ones that don't check for blackjack (european no hole card rule). For reasons I won't go into here, the results and theoretical probabilites for this test can only be calculated when the player doesn't bust. There is room for confusion because a dealer blackack is shown for OddsOn if the player busts but not with Microgaming (i.e with Microgaming the dealer only shows one card if the player busts so the dealer could have blackjack without the player knowing but Odds On turns over the hole card which could show a dealer blackjack).
TRUEGAMBLER NOVEMBER UPDATE...Microgaming Blackjack results corrected again. It has become clear that the initial report was released too early. More mistakes have been discovered in the analysis and in the data collection process. Now Microgaming blackjack is shown as highly likely to have been completely fair and randomly dealt. Any slight manipulation of the cards would have been obvious considering the millions of hands in the test so if rigged, the resulting probabilities would be astronomically low, but the Win/Tie/Loss ratios and 'dealer outcome' figures are within very normal range. The player initial hands are still a bit unlikely but quite possible (much more possible than last time). Some recognition errors of the ten-value cards were found which I suggested before was probably the case. The payout isn't altered much by the initial player hand results so there may be more errors in the data collection or it could just be a normal variation. Winneronline discussion on the subject
| 
Calendar 
LINKED: 
