5.5 million blackjack decisions, graded.
Six months of play from 10,000+ players in 130 countries, every move checked against perfect strategy. Where people go wrong, why, and what fixes it.
- decisions
- 5.5M
- hands
- 3.6M
- players
- 10,000+
- index plays
- 98K
Key findings
- 01 · Surrender56%of correct surrenders are missed. No play is missed more.
- 02 · Hesitation1.95sto make a wrong call, against 1.65s for a right one.
- 03 · Tilt+17%more mistakes after three losses in a row.
- 04 · Counting64%of counters skip insurance at +3, right when the count says take it.
- 05 · Practice38%fewer mistakes after 5,000 decisions in the trainer: 10.4% down to 6.5%.
- 06 · Drills49%fewer mistakes on a hand once it's drilled: 20.5% down to 10.5%.
Where players go wrong, hand by hand.
Mistakes cluster in three places: stiff hands against a weak dealer, soft doubles, and surrender. Almost everything else is played well.
| You | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | A |
|---|---|---|---|---|---|---|---|---|---|---|
| 8 | 10 | 11 | 12 | 14 | 15 | 6 | 4 | 4 | 4 | 4 |
| 9 | 29 | 31 | 27 | 25 | 24 | 20 | 15 | 8 | 6 | 7 |
| 10 | 12 | 10 | 10 | 10 | 10 | 16 | 19 | 23 | 17 | 18 |
| 11 | 7 | 7 | 7 | 7 | 7 | 7 | 8 | 9 | 12 | 20 |
| 12 | 17 | 25 | 36 | 33 | 33 | 5 | 5 | 5 | 5 | 8 |
| 13 | 30 | 27 | 26 | 25 | 27 | 5 | 5 | 4 | 4 | 7 |
| 14 | 20 | 19 | 19 | 20 | 21 | 9 | 7 | 7 | 7 | 8 |
| 15 | 13 | 12 | 12 | 13 | 15 | 16 | 14 | 16 | 27 | 31 |
| 16 | 8 | 7 | 7 | 8 | 8 | 26 | 24 | 36 | 31 | 33 |
| 17 | 3 | 2 | 2 | 3 | 3 | 3 | 4 | 4 | 4 | 9 |
3,465,732 decisions since August 9, 2026, across every rule set players chose. Cells under 300 decisions are blank; face cards grouped with 10.
Players are timid, not reckless.
When the right play puts more money out, a double or a split, players miss it twice as often. Surrender is missed more often than it's made.
| Hand | Right play | Players did | Missed |
|---|---|---|---|
| Hard 15J♠5♥vK♣ | Surrender | Hit | 43.5% |
| Soft 18A♥7♣vK♦ | Hit | Stand | 37.4% |
| Hard 16J♣6♦vK♠ | Surrender | Hit | 33.2% |
| Hard 12J♦2♠v4♥ | Stand | Hit | 32.7% |
| Hard 12J♠2♥v5♣ | Stand | Hit | 29.2% |
| Hard 12J♥2♣v6♦ | Stand | Hit | 29.0% |
| Hard 13J♣3♦v2♠ | Stand | Hit | 28.4% |
| Hard 16J♦6♠v7♥ | Hit | Stand | 24.6% |
Spots with 2,000+ decisions. Surrender spots count only tables that offered late surrender.
The common thread is fear. Players hit 12 against a 4, though a 4 busts 40% of the time. They play out 16 against a 10 instead of surrendering. One in eight splits 10s against a 6. Every right play is on the basic strategy chart.
Try five of them yourself.
What's your play?
Pick a play to see how many of the 10,000+ players in this report missed it. Standard 6-deck rules, double allowed.
The longer players think, the more likely they're wrong.
Hard 20 takes 1.2 seconds and is almost never wrong. Soft 17 and 18 take nearly 3 seconds and are missed a third of the time. A pause usually means a player doesn't know.
Each dot is one spot with 3,000+ timed decisions. Dashed line: trend (r = 0.81).
Practice made players faster and more accurate.
The 110 players with the most trainer reps, from their first hand to decision 5,000.
- Time to decide −20%
- 1.9s1.5s
- Mistakes −38%
- 10.4%6.5%
Losing streaks make players worse. Even counters miss the basics.
Players miss 10.8% of decisions after a win and 13.6% after five straight losses. It isn't just weaker players losing more: measured against themselves, 61% of regulars play worse after a streak. And self-described counters, the sharpest group, still miss one basic play in eleven.
2,943,818 decisions. Losses in a row in the same shoe; Daily and Ranked excluded.
Players with 100+ graded decisions after choosing a level.
The table, by the numbers.
The biggest losers played the most: a median 990 hands. The edge grinds.
Play money only. Player totals exclude Daily and Ranked.
The count changes the play. Counters often don't.
Players with deviations on missed index plays 17.6% of the time, against 11.5% for ordinary plays. Most never bet the count at all.
Hi-Lo true count. Common index values shown.
Hi-Lo index is +3. Coral: decline. Green: take it.
1,542 players with 300+ hands. Daily, Ranked and auto-bet hands excluded.
At +3, insurance flips from a bad bet to a good one, yet only a third of offers were taken. Betting is worse: a count only pays if your bet rises with it, and just 23% of players double their bet at +4. Study the plays in the Illustrious 18, or see what a spread does to risk of ruin in the Bankroll Lab.
The players who played the most got 38% better.
So does it get better? We followed 110 players from their first hand to their 5,000th decision. Mistakes fell from 10.4% to 6.5%, and were still falling.
Each point is 27,500 decisions: 250 from each of the 110 players.
266 players with 2,500+ decisions. Their first 1,000 decisions vs their latest 1,000.
The biggest gains came on the plays beginners fear most: doubles, splits and surrenders.
Drilling a hand cuts its mistakes in half.
The trainer turns each miss into a drill and deals that spot until you get it right. Then we checked the same spot at the real table.
Your misses become your drills.
Every spot you miss in the trainer lands in your drill library, ready to deal again.
Open the drill library →Shuffling up...675 players. 65,193 decisions before the drill, 147,432 after.
Start your own curve.
Every player in Fig. 11 started with the same mistakes. Reps fixed them. Free, no sign-up, no real money.
Methodology
- Anonymous and aggregate. Every figure is a count or a rate across many players. No names, accounts, or individual players appear anywhere on this page.
- The window. Play from May 10, 2026 to November 20, 2026. Hand-level detail (every figure) starts August 9, 2026, when it was first recorded.
- What counts as a mistake. A decision that differs from the correct play for the exact rules the player chose (decks, soft 17, double after split, surrender), including count-based index plays when those were switched on.
- Following players. Figs. 11, 12 and 13 follow the same players over time, which removes the “only good players keep playing” effect.
- Left out. Ranked Challenge hands use fixed shoes and are excluded from accuracy. Ad blockers hide some players, so treat totals as a floor.
Quick answers
- What percentage of blackjack decisions do players get wrong?
- Across 3,465,732 graded decisions, players made the basic-strategy play 88.4% of the time, so about 1 decision in 9 was a mistake.
- Which blackjack play do players miss most often?
- Surrender. When surrendering was the correct play, players missed it 55.8% of the time. Hard 15 against a dealer 10 was the single most-missed spot.
- Do blackjack players get better with practice?
- Yes. The 110 players who made 5,000+ decisions missed 10.4% in their first 250 and 6.5% by decision 5,000, a 38% drop.
- Which card-counting deviations do players miss most?
- Insurance at a true count of +3 or higher was declined 64% of the time it should have been taken. Surrendering 15 against a 10 at counts of 0 and up was missed 42% of the time, and standing on 16 against a 10 at 0 and up was missed 28% of the time.
- Do blackjack drills work?
- On spots players drilled until they mastered them, their live-play miss rate fell from 20.5% before the drill to 10.5% after.
- Do blackjack players tilt after losing streaks?
- Yes. The same players missed 8.6% of decisions after a win and 10.1% after three or more losses in a row. 61% of 772 regulars played worse after a losing streak.