6:5 naturals, the single zero, dealer hitting soft 17, La Partage — each rule is a measurable shift in the edge. Here is the ledger.
I. Why the edge is a stack, not a number
Each rule in a table game changes the expected value of a bet by a small, computable amount. To a good approximation these adjustments are additive: start from the base game's edge, then add or subtract the effect of each rule the table actually uses. This is why the phrase "the house edge in blackjack" is meaningless without a rule sheet — the real edge is the sum of the levers in play at that specific table. The figures below are the standard published adjustments; where a lever's exact value depends on other rules present, that is noted.
II. Blackjack — the levers, largest first
| Rule change | Effect on house edge | Direction |
|---|---|---|
| Naturals pay 6:5 instead of 3:2 | ≈ +1.4 pp | Much worse for player |
| Single deck instead of six (same other rules) | ≈ −0.5 pp | Better — if the payout stays 3:2 |
| Dealer hits soft 17 (H17) | ≈ +0.2 pp | Worse |
| No double after split (no-DAS) | ≈ +0.14 pp | Worse |
| European no-hole-card rule | ≈ +0.11 pp | Worse |
| No surrender offered | ≈ +0.08 pp | Worse |
| Re-split aces allowed | ≈ −0.07 pp | Better |
The lesson buried in that table: a single-deck game is not automatically the better bet. Today's single-deck tables almost universally pay 6:5, so the −0.5 point saving from the deck count is swamped by the +1.4 points from the payout. A 3:2 six-deck game beats a 6:5 single-deck game handily. Check the payout on the felt before the deck count.
III. Roulette — the zeros are everything
| Wheel / rule | House edge | Note |
|---|---|---|
| European (single zero) | ≈ 2.70% | = 1/37 on every bet type |
| American (double zero) | ≈ 5.26% | = 2/38 — the second zero nearly doubles it |
| French, La Partage (even-money bets) | ≈ 1.35% | Half the stake returned on zero |
| French, En Prison (even-money bets) | ≈ 1.35% | Stake "imprisoned" for one spin instead |
| American five-number bet (0-00-1-2-3) | ≈ 7.89% | The one bet with a worse edge than the wheel |
Roulette has essentially one lever: how many green pockets the wheel carries, and whether a partial-refund rule softens the even-money bets. Everything else about the layout is cosmetic — the payout structure is deliberately arranged so that straight-ups, splits, columns and dozens all carry the identical base edge. The full derivations of 2.70% and 5.26% are in what the house edge actually is.
IV. Baccarat — commission and payout
| Rule / bet | House edge | Note |
|---|---|---|
| Banker, 5% commission | ≈ 1.06% | The standard, and the cheapest |
| Player | ≈ 1.24% | No commission, wins less often |
| "Commission-free" Banker (Super 6 / half-win on 6) | ≈ 1.46% | Usually worse — read the small print |
| Tie at 8:1 | ≈ 14.4% | Avoid |
| Tie at 9:1 | ≈ 4.8% | Still poor |
Beware the marketing lever: "commission-free baccarat" removes the 5% charge but typically pays a Banker win of 6 only half (1:2) when it totals six, which raises the edge above the standard commissioned game. The label sounds player-friendly; the arithmetic is not. The commission derivation is in the baccarat probability study.
V. A worked stack
Take a low-limit Strip blackjack table advertising a friendly minimum. Read the felt and it turns out to run 6:5 naturals, dealer hits soft 17, and no surrender:
0.4% (base) + 1.4 (6:5) + 0.2 (H17) + 0.08 (no surrender) ≈ 2.1% house edge
Roughly five times the cost of the 0.4% game two pits over — with no change to the posted minimum bet. The price is in the rules, not the placard.- House edge is a base figure plus a stack of near-additive rule adjustments.
- Blackjack's biggest lever by far is the 6:5 vs 3:2 naturals payout (≈ 1.4 pp).
- A 3:2 six-deck game beats a 6:5 single-deck game — payout outweighs deck count.
- Roulette's only real lever is the number of zeros (and La Partage on even-money bets).
- "Commission-free" baccarat is usually worse than the standard 5% Banker bet.