I've been following this thread and I have to admit I'm struggling to understand what the OP's issue is with dealer bust hands.
I would have thought the most important stat is the number of hands you won against the dealers hands in a session. As a player you can win by having better cards than the dealer or by the dealer busting out. That would seem a better metric to analyse and give you a better indication of how close you are to RTP for that session. Remember though that RTP is calculated over 1000's if not 10.0000's hands, so playing less you will see fluctuations against RTP, bot positive and negative.
Varying your initial bet during a session will also have an impact on your bankroll and therefore whether you win or lose in a session. Keeping your stake constant will give a better idea against RTP.
He initially complained about the
bust-percentage being far below the expected 28.2%. His problem is that his analysis does also include a lot of game-rounds where the dealer will have a bust-probability of 0 (Player BJ and Player bust) and he would need to remove these hands from his analysis.
You would need to do the following:
1) Observe a proper amount of game-rounds (say 1,000, better 5,000 or 10,000), also possible in fun-mode
2)
Remove all hands with Player-BJ and Player-bust from your observation
3) Count the remaining game-rounds
4) for every dealer upcard separately: divide the dealer-busts by total hands-played
For example:
Dealer Upcard 8
-500 hands played
-amount of player-BJ 25, amount of player busts 150,
remaining hands observed 325, dealer busted 75 times
75/(500-25-150) = 0.2308 = 23.08% dealer-busts against upcard 8
5) repeat this procedure with all dealer-upcards
6) weight your numbers. This is necessary because you will remove proportionally more hands where you play against 7-A (because you need to hit till 16) and to avert distortion.
example (simplified), remaining hands after step 2):
dealer upcard 2: 480
dealer upcard 3: 480
dealer upcard 4: 500
dealer upcard 5: 500
dealer upcard 6: 500
dealer upcard 7: 325
dealer upcard 8: 325
dealer upcard 9: 325
dealer upcard 10/J/Q/K: 325 each
dealer upcard A: 325
Total hands remaining for analysis: 5060
5060/13 = 389.23
example: weighting for upcard 8: 325/389.23 = 0.8350 x 23.08% = 19.27%
This needs to be done for all other upcards.
7) Sum up all weightings and divide it by 13.
8) Your
final result should be between 27.2%-29.2% (confidence level 95%) with a sample size of 6,500 (assumption).
I think OP's problem is mainly his
playing-strategy. He should consider learning proper basic strategy. I'm sure his RTP-results will improve a lot. And no guarantee that my calculation is correct.
I also assumed basic strategy with the underlying numbers. However the playing-strategy has no impact on dealer bust%.