Many years ago, I took a C++ class at the local college, and our professor was a brilliant programmer, but an awful teacher. So a few of us who already knew what we were doing taught the class for the professor. For a final project, we had to write a computer program to simulate Five Card Stud Poker given predetermined rules for the AI.
One classmate implemented the rules with a twist – the AI selected its best hand from both the cards in hand and those it discarded for the turn.
We ran side by side simulations and his AI won about twenty percent more often. For comparison, the house edge in Blackjack is about five percent.
That’s the power level of the London Mulligan.