The Best Blackjack Strategy, Tested on 30 Billion Hands - Dice Lab
November 13th, 2026 at 8:22:34 AM
permalink
The Best Blackjack Strategy, Tested on 30 Billion Hands - Dice Lab - 252 subscribers
https://www.youtube.com/watch?v=f70jGzeAzdc
Never bust. Play like the dealer. Always take insurance. Split every pair. Blame the guy at third base. Every blackjack player has a system, so I built a six-deck blackjack table in code and sat ten players down at it, each with a different strategy. Then I let the dealer deal more than 30 billion hands.
The players: Random, Mimic the Dealer, Never Bust, Split Everything, Average Joe, Insurance, Basic (the perfect strategy chart), Martingale, Counter (Hi-Lo) and Wonger (back-counting).
What came out:
Perfect basic strategy still loses, but only 0.43% of what it bets. It wins 43.3% of hands and loses 47.9%. Doubles and splits make up the difference: without them, it loses 2.36%.
Never Bust (never hits a hard 12+, never doubles or splits) lost 6.33%, 15× worse than basic. Mimic the Dealer lost 5.68%, because you bust first. Split Everything lost 7.02%. Random lost 42.7%.
Average Joe (stands on 15+, doubles only on 10/11, splits only aces and 8s, takes even money) lost 2.46%, about $24.60 per 100 hands of $10.
16 vs a dealer 10 is the worst hand in the game: you lose about 54 cents per dollar whether you hit or stand. 11 vs 6 (double) and A-A vs 6 (split) are the best spots, +67 cents per dollar.
Dealer busts 42.2% with a 6 up, 23.0% with a 10, 16.7% with an ace.
Insurance loses 7.5% of every dollar put on it. Declining even money is worth about $1.03 per $1, instead of exactly $1.
The bad player at third base changed our result on 14% of hands: 7.04% for us, 7.01% against. Our win rate didn't move.
A 6:5 table quadruples the house edge, from 0.43% to 1.79%. Dealer hits soft 17 costs another 0.21%. A single-deck 3:2 table gives the perfect player a tiny edge (+0.02%), but single-deck 6:5 H17 is −1.56%.
Hi-Lo card counting ($10 to $80 spread) beats the house by 0.49%: about $7 per 100 hands, with ~2,900 hours needed to be 95% sure you're ahead. With $1,000, 58% of counters went broke within 10,000 hands. Back-counting (playing only at true count +1 or higher) reaches +1.86%. On a 6:5 table the counter loses 0.98%; on a shuffling machine, 0.40%.
Martingale finished ahead on 36% of nights and went broke on 62%. Its median night ended with $5.
How it was tested: 6 decks, blackjack pays 3:2, dealer stands on soft 17 and peeks, double on any two, double after split, split to 4 hands, no surrender, 75% penetration, $10 base bet, $1,000 bankroll and 300-hand nights. 30,666,968,639 hands in total.
Chapters:
0:00 30 billion hands
1:04 The game in 30 seconds
2:19 The 10 players
2:52 Random, Mimic the Dealer, Never Bust
4:39 Average Joe
5:09 Perfect basic strategy
6:40 The worst hand in blackjack
7:12 Insurance and even money
8:02 The guy at third base
8:51 Streaks and the Martingale
9:43 The table rules (3:2 vs 6:5)
11:14 Card counting
14:12 The final ranking
#blackjack #casino #strategy
https://www.youtube.com/watch?v=f70jGzeAzdc
Never bust. Play like the dealer. Always take insurance. Split every pair. Blame the guy at third base. Every blackjack player has a system, so I built a six-deck blackjack table in code and sat ten players down at it, each with a different strategy. Then I let the dealer deal more than 30 billion hands.
The players: Random, Mimic the Dealer, Never Bust, Split Everything, Average Joe, Insurance, Basic (the perfect strategy chart), Martingale, Counter (Hi-Lo) and Wonger (back-counting).
What came out:
Perfect basic strategy still loses, but only 0.43% of what it bets. It wins 43.3% of hands and loses 47.9%. Doubles and splits make up the difference: without them, it loses 2.36%.
Never Bust (never hits a hard 12+, never doubles or splits) lost 6.33%, 15× worse than basic. Mimic the Dealer lost 5.68%, because you bust first. Split Everything lost 7.02%. Random lost 42.7%.
Average Joe (stands on 15+, doubles only on 10/11, splits only aces and 8s, takes even money) lost 2.46%, about $24.60 per 100 hands of $10.
16 vs a dealer 10 is the worst hand in the game: you lose about 54 cents per dollar whether you hit or stand. 11 vs 6 (double) and A-A vs 6 (split) are the best spots, +67 cents per dollar.
Dealer busts 42.2% with a 6 up, 23.0% with a 10, 16.7% with an ace.
Insurance loses 7.5% of every dollar put on it. Declining even money is worth about $1.03 per $1, instead of exactly $1.
The bad player at third base changed our result on 14% of hands: 7.04% for us, 7.01% against. Our win rate didn't move.
A 6:5 table quadruples the house edge, from 0.43% to 1.79%. Dealer hits soft 17 costs another 0.21%. A single-deck 3:2 table gives the perfect player a tiny edge (+0.02%), but single-deck 6:5 H17 is −1.56%.
Hi-Lo card counting ($10 to $80 spread) beats the house by 0.49%: about $7 per 100 hands, with ~2,900 hours needed to be 95% sure you're ahead. With $1,000, 58% of counters went broke within 10,000 hands. Back-counting (playing only at true count +1 or higher) reaches +1.86%. On a 6:5 table the counter loses 0.98%; on a shuffling machine, 0.40%.
Martingale finished ahead on 36% of nights and went broke on 62%. Its median night ended with $5.
How it was tested: 6 decks, blackjack pays 3:2, dealer stands on soft 17 and peeks, double on any two, double after split, split to 4 hands, no surrender, 75% penetration, $10 base bet, $1,000 bankroll and 300-hand nights. 30,666,968,639 hands in total.
Chapters:
0:00 30 billion hands
1:04 The game in 30 seconds
2:19 The 10 players
2:52 Random, Mimic the Dealer, Never Bust
4:39 Average Joe
5:09 Perfect basic strategy
6:40 The worst hand in blackjack
7:12 Insurance and even money
8:02 The guy at third base
8:51 Streaks and the Martingale
9:43 The table rules (3:2 vs 6:5)
11:14 Card counting
14:12 The final ranking
#blackjack #casino #strategy
