What you need to do is not look only at AA vs other cards. You won't find anything.
You have to take at least 2 parameters into account.
1. Compare a shorter chip stack, who is constantly set-up against a larger chip stack for elimination with a dominant hand, for example short gets KK, large gets AA. Sometimes it's as subtle as K5 for shorty UTG vs K6 for big boy in the big blind. I know you have to steal the blind to survive, I know Mr. Big will call you with his K6 because he has you 10x on chips and is in his interest to eliminate you on the bubble. Add up the number of instances you notice this in all-ins.
If I am programming an algorithm for
equity distribution, or for elimination, I don't need to crack your AA to cheat you, I just need to give another player AA when you have KK, K6 when you are forced to go all in for a blind steal with K5, when you are in an elimination scenario, and re-balance it when you are not, and the
odds look perfectly normal. I also don't need a huge statistical anomaly. 2-3% in favor of the large stack is more than enough over 1000 player fields and thousands of tournaments. Not much different than how a casino only has a tiny percentage edge over the player. Heck if I target your account 1% is more than enough if you play every day to make sure you go broke in a timely fashion.
2. Compare the number of times the higher chip stick cracks your dominant hand, again in elimination scenarios.
Add those two together and that's where the discrepancy will reveal itself.
Those two stats added together should actually be 50% over time. In other words chip count should have zero bearing on who wins a hand at show down in an elimination scenario.
But this is how the algorithms eliminate players in a timely fashion to keep the tournaments moving. It's as simple as that. Without this, 1000+ player tournaments would take 24 hours or more to complete not 4-6. You would have tons more people making come-backs.
If I'm programming the algorithm I'm not going to be so stupid to skew the % of times AA hold up or gets dished out which is so easy to track, when I can do it perfectly fine without ever touching those statistics. I'm going to do it based on
poker game rules, position, player tendencies, not card statistics. I leave those alone.
In a fair game, over time, I should have the dominant hand when I am short just as much when I am big in elimination scenarios.
In a fair game, my hand should crack the dominant hand when I am short just as much as when I am big in elimination scenarios.
In a fair game, my dominant hand should hold up just as much when I am short as when I am big in elimination scenarios.
Now that you break it down like this, can you see how you can adjust it without ever touching card stats? Take these rules, and add another step, apply it to the rules of poker, such as where players are forced to steal blinds from to survive, and I need to adjust it even less than a catch-all code to push players out of the tournament and it's going to be pretty damn difficult for you to figure out where my algo set-you up. Hand history is going to look 100% legit, that's for sure.
It's situational. Think of football. Red zone offense and defense, vs other situations. On the bubble, not on the bubble. Etc. I can increase the number of action flops on the bubble vs when you are not on the bubble. I can submit my card dealing RNG for verification, and get certified, but I don't have to submit the entirety of the algo, not the overlayers. Not the user accounts overlayers, "that's proprietary, that's a security risk".
I can adjust my algorithm through table rotation to adjust it for faster elimination, or adjust it for longer duration. Since my company is interested in as many tournaments as possible, as many rebuys as possible, it's going to be adjusted for elimination and to shorten duration. Shuffle and deck RNG can be 100% untouched.
As long as
online poker is programmed, don't ever trust it. Online poker has a house advantage beyond rake, whereas Live poker does not, beyond rake.