Twelve myths,
checked.

Martingale doubles after every loss. The math: an N-loss streak requires 2^N units; the table maximum caps you well before bankroll does. EV per round is unchanged at −0.5%; the difference is purely variance. Over a $10k bankroll at $25 starting bet, you go bust on ~10 consecutive losses (~0.1% probability per session), losing the equivalent of ~400 winning sessions.

Dealer bust rate is fixed by the upcard: ~28% on average across upcards, ~35-42% when showing 4-6, ~17-21% when showing 7+. Every dealer everywhere produces the same bust rates over a sufficient sample. The dealer "running hot" for an hour is variance, not skill.

Long-run EV is independent of seat position because the cards are dealt sequentially from a randomized shoe. The outcome of any single hand depends on which specific card you receive, but over a thousand hands every seat sees the same EV distribution. The "bad third base" memory is a cognitive bias — losses caused by third-base decisions are vivid; wins where third base "saved" the table are forgotten.

Federal and state law in the US, UK gambling law, and most jurisdictions worldwide consider card counting a form of skilled play, not cheating. Counting only becomes illegal when assisted by a device (concealed computer, marked cards, etc.). Mental math: legal.

P(dealer has blackjack | dealer shows ace, fresh shoe) = 16/49 ≈ 32.7%. Insurance pays 2:1, requiring 33.3% to break even. House edge: (33.3% − 32.7%) × 1 unit + (67.3% × −0.5 units) ÷ insurance bet ≈ 7.4% on the insurance stake. Only flips +EV at counted true count ≥ +3.

In live studios, the hole card is dealt face-down with the camera positioned away from the player view; dealers are trained on a "lift then place" motion that gives no visible angle. The studios use dedicated hole-card review processes for any incident where a dealer accidentally exposes the card. RNG has no hole card — there's a deterministic seed-derived outcome and nothing to read.

Player BJ vs dealer A: dealer also has BJ ~30.8% of the time (push, $0). Otherwise (~69.2%), player wins 3:2 ($1.50). Expected payout: 0.692 × $1.50 + 0.308 × $0 ≈ $1.038 per $1 bet. Even money pays $1.00 — leaves ~3.8c per dollar on the table. Skip even money for the same reason as insurance: same bet by another name.

A 0.5% house edge means players win ~49.75% of $1-equivalent decisions over volume; over 100 hands the standard deviation of return is ~$11.50. So sessions ending −$30 to +$30 on a $100 buy-in are completely normal variance — within the first standard deviation, just from variance not "rigging." The "rigged" feeling is loss aversion, not statistics.

In a shuffled shoe, sequential outcomes are not statistically independent (cards already drawn aren't in the remaining deck) — but the residual signal is small unless you're actively counting and the shoe has deep penetration. Without active counting, every hand is effectively independent. The "feel" of a turn is your brain finding patterns in noise.

Natural blackjack frequency ≈ 4.83%. 3:2 pays $1.50/$1 on a natural; 6:5 pays $1.20/$1. Difference per BJ: $0.30. Edge added: 0.0483 × $0.30 ≈ $0.01393 per $1 wagered = +1.39% house edge. Independent of every other rule. The "next pit pays 3:2" rule of thumb pays for itself in ~30 hands.

1-deck S17 DAS 3:2: ~0.18% house edge. 6-deck S17 DAS 3:2: ~0.41% house edge. 1-deck 6:5: ~1.45%+ house edge. The deck count delta is real (~+0.23-0.48%) but smaller than the payout delta (~+1.39%). Always check the payout before sitting at any blackjack table.

House edges by side bet (typical): Perfect Pairs ~6.0%; 21+3 ~3-8% depending on paytable; Bust It ~6.9%; Lucky Ladies ~17%; Insurance ~7.4%. Compare to ~0.41% main game. The single rule: never take any side bet, ever. Every dollar staked there is paying ~10× the rate of your main game.