Games · Blackjack Switch
Blackjack Switch: they let you swap, then quietly take it back
Swapping the second card of two hands is a genuinely powerful privilege, and at 0.58% this is one of the cheapest games on the floor. The catch is that the casino takes its money back through three rule changes you can read past in a second: blackjacks pay even money, a dealer 22 pushes instead of busting, and a 21 you built by switching is only 21.
Switch the second card of each hand?
Educational simulation: fun credits only, no real money. Six decks, dealer hits soft 17, blackjack pays even money, and a dealer 22 pushes every hand except a natural. This felt allows one split per hand; the Simulate tab plays the full rules including resplits to four hands.
You post two equal bets every round, so the edge below is measured against both. A $5 selection means $10 at risk per round before any double or split.
The simulator switches by comparing the exact expected value of both arrangements, which is a shade better than any human method and a shade worse than the published optimal figure (its strategy is derived under an infinite-deck approximation). Runs therefore settle slightly above 0.58%, and that is the model talking, not a leak: with the deck model equalised, the simulator and our exact solver agree to well inside one standard error. Dealer-22 is the sharpest check on the felt here, since it is fixed by the dealer's rules alone and cannot be moved by strategy. The switch rate counts every round dealt, including the ones a dealer blackjack ends before you get to choose.
What the free switch actually costs you
| The rule | In ordinary blackjack | In Switch |
|---|---|---|
| Blackjack pays | 3:2 (or 6:5 on bad tables) | Even money |
| Dealer draws to 22 | Dealer busts: you win | Push, unless you hold a natural |
| 21 made by switching | n/a | Counts as 21 points, not a blackjack |
| Swapping cards between hands | Impossible | Allowed, once, on the second cards |
Those first three are the price of the fourth. Push 22 is the heaviest of them: the dealer reaches exactly 22 on roughly 7.4% of resolved rounds, and every one of those is a hand you would have won outright at a normal table.
Super Match, on your four starting cards
| Hand | Pays | Chance | Odds |
|---|---|---|---|
| Four of a kind | 40:1 | 0.0357% | 1 in 2,804 |
| Two pair | 8:1 | 1.5342% | 1 in 65 |
| Three of a kind | 5:1 | 1.9567% | 1 in 51 |
| Pair | 1:1 | 35.2205% | 1 in 3 |
| No pair: loses | −1 | 61.25% | n/a |
A pair alone shows up on 35.2% of deals, which is why this bet feels like it pays constantly. It still costs 2.55%, because a pair only returns your stake, and the 61.3% of deals with nothing take it away.
How to decide the switch
The exact answer compares the expected value of both arrangements, which is what this simulator does. Humans use a published shortcut instead: Cindy Liu's point count: score each way of arranging the four cards and take the higher. The source publishes two versions: the advanced count costs about 0.08% against perfect play and the simpler one about 0.17%: a fair price either way for something you can do at the table.
The Play tab shows both arrangements' values on every hand, so you can watch which one the math prefers and why. The how-to page walks through the point count itself.
How this simulation works
- Rules modeled
- Blackjack Switch: six decks, dealer hits soft 17 and peeks for blackjack on an ace or ten upcard. The player posts TWO equal bets and receives two hands, then may swap the SECOND card of each hand: once, before playing. A blackjack made by switching counts as 21 points, not a natural. Winning naturals pay EVEN MONEY rather than 3:2. PUSH 22: if the dealer's final total is a hard 22, every non-busted player hand pushes, with one exception; a player natural still wins. Double on any two cards, double after split, resplit to four hands, no surrender. The optional Super Match side bet resolves on the player's four initial cards before any switch: pair 1:1, three of a kind 5:1, two pair 8:1, four of a kind 40:1.
- Assumptions
- A freshly shuffled six-deck shoe per round, fair and independent, with no cut-card effect and no card counting. Fun credits only, no real money. The Play felt supports splitting one level per hand (up to four hands); the simulator plays the full resplit-to-four rules. The switch decision compares the exact expected value of both arrangements rather than using a human point count, so it is slightly better than any published method and slightly worse than published-optimal play, whose strategy is derived under a finer model than ours.
- Mathematical basis
- Super Match is ours: an exact closed-form enumeration of every four-card combination from a six-deck shoe, recomputed by the test suite on each run, which reproduces the published probabilities exactly, and is also the module's proof that six decks is the right assumption, since the counts are deck-count sensitive. The base-game house edge (0.58%), the dealer-22 frequency (7.3536%), the Super Match edge (2.55%) and the cost of the published human switching methods (0.08% advanced, 0.17% simple) are published figures (Wizard of Odds). Neither switching method is exercised by any code path here; they are cited, not derived.
- Validation
- Independently reviewed twice before publication (gate D-008). The first gate failed the module: it found a defect in the hand-value routine that made the dealt dealer stand on soft 17s the rules require it to hit, which moved the house edge by 0.061 percentage points and manufactured phantom dealer-22 pushes. A second reviewer then re-derived every fix from scratch, comparing the corrected routine against an independent reference across 1,111,110 ordered hands with zero mismatches, and reproducing our exact solver's 0.74032% to nine decimal places, and held the module a second time until its own documentation stopped claiming more than the tests proved. Both records, including what remains unverified, are published in full.