This is for entertainment purposes only. It shows a system I've developed for betting on blackjack. The results were initially really bad, with a 25% loss ratio. Over time, and with lots of small improvements, the latest revision has a positive profit ratio of about 1.3%. I tested it with several million hands per test. The house win ratio is about 52% now, but the profit ratio has steadily improved with changes to betting on splits and doubledowns.
The latest revision has support for:
- a betting formula to determine how much to bet on the next hand.
- a formula to determine when to doubledown -- it currently tests at a 3-2 win ratio
- a formula to determine when to split -- it currently tests at a 8-7 win ratio
Some other assumptions in this testing framework: I'm using a pseudo-random generator to shuffle the cards, so the card distribution is very randomized. When there are less than 52 cards remaining in the shoe, the cards are re-shuffled.
A few interesting observations:
- The likelihood of a losing or winning streak of only 1 win/loss is lower than 2+ wins/losses
- The likelihood of a win goes up dramatically if there are more than 66% 10s or face cards showing, prior to betting
- The likelihood of the dealer going bust is about 22%