How casino games work

There's no strategy. There is arithmetic.

Nothing you do changes a slot's expected outcome — not stake size, not timing, not stopping the reels early, not switching machines after a win. The only two levers are which machine (its RTP) and how long you play (total wagered).

At 5 chips a spin, 600 spins an hour, on the classic 5.76% machine: 5 × 600 × 0.0576 = 173 chips an hour, on average. That number is the product. Everything else — the sound design, the near misses, the free spin round — is packaging around it.

Why the swing feels so much bigger than 5%

Because hour to hour, it is. High-volatility games hand back most of their RTP through rare events, so a typical session looks nothing like the average. Run Simulate 500 spins a few times and watch the actual return land anywhere between "great night" and "wiped out" while the underlying number never moves.

Choosing between the three machines

* Only if you count the jackpot money. Ignore the jackpots and it's 93.29%.

Look at those two middle rows together, because that's the trick. The classic machine pays something on more than a third of spins — but most of those are a single cherry handing your stake straight back. Nothing happened, and the machine still made a noise. That gap between "a win" and "more money than you started the spin with" is the most useful thing on this page.

Myths worth being able to kill in one line

• "It's due." Every spin is independent. 400 dry spins doesn't change spin 401 at all.
• "It went cold after my win." There's no state carried between spins on a standard RNG slot.
• "Bigger stakes hit more often." Stake scales prizes and losses equally. Some games do gate features behind max bet — that's a pay table difference, and it's disclosed in the rules screen.
• "The casino flips a switch when I'm winning." In licensed markets the RTP is certified by an independent test house, and changing it means recertification and usually regulator notice.
• "The jackpot's overdue." True only for must-drop jackpots, which are contractually guaranteed to pay by a certain point. For everything else, no.

Basic strategy, and why it isn't a "system"

Basic strategy is simply the mathematically best action for every combination of your hand and the dealer's up card, worked out by simulating every possible outcome. It doesn't win — it loses least, taking the house edge from around 2% down to about 0.5%. Learning it is the whole skill of blackjack, short of counting.

Hard totals — no flexible ace

The logic for that 12–16 row: those hands probably bust if you draw. A dealer showing 2–6 will probably bust too, so you stand and let them do it. A dealer showing 7 or higher will probably make 17+, so a losing 15 has to be improved even though drawing is risky.

Soft totals — you're holding a flexible ace

You can't bust a soft hand with one card, so hitting is free. Soft 18 losing to a dealer 9 or 10 is the one people get wrong most often — it looks like a good hand and it isn't.

Pairs

Card counting, honestly

Used cards aren't put back, so the remaining shoe shifts as it depletes. A deck rich in tens and aces favours the player — more blackjacks, more dealer busts. Counters keep a rough running total and raise their bet when it's high. It's legal, it isn't cheating, and it's basically dead online: RNG blackjack reshuffles every hand, and live tables use continuous shufflers plus betting-pattern detection.

Every bet is secretly the same bet

Take a straight-up number. You win once in 37 spins and get paid 35 plus your chip back. Over 37 spins staking 1 each time, you stake 37 and receive 36. You lose 1 in 37 — 2.70%.

Now take red. You win 18 times in 37 and get paid 1:1. Over 37 spins you stake 37 and receive 36. Also 2.70%. Dozens, corners, splits, columns — all 2.70%. The wheel doesn't care what shape you drew on the layout. Choosing a bet only chooses your volatility: straight-up is rare and huge, red is frequent and small, and the expected cost is identical.

Why betting systems don't work

The Martingale — double your bet after every loss so one win recovers everything — is the one everybody invents in their first hour. It fails for two reasons that have nothing to do with luck: table limits, and the fact that your money runs out. Starting at 1 chip, eight losses in a row (about 1 in 250, so a fairly normal evening) needs a 256-chip bet to win back 1 chip. It works right up until it catastrophically doesn't.

No arrangement of bets on a negative-expectation game creates a positive expectation. That's provable rather than merely observed: the expected value of a sum is the sum of the expected values, however you order or size them. Any system being sold to you is either ignorance or fraud.

The gambler's fallacy

Red has landed nine times. Black is not "due". The wheel has no memory and no obligation to even things out. The famous 1913 Monte Carlo run went black 26 times in a row, and the house made a fortune from people who were certain red had become inevitable.

The correct play, in full

Bet Banker. Every hand. It's the lowest-edge bet on the table at 1.06% even after commission, and there's nothing else to decide. Player at 1.24% is perfectly fine too. That is the entirety of baccarat strategy.

Never bet the Tie

It pays 8:1 on something that happens about 9.5% of the time — the fair price would be roughly 9.5:1. That gap is a 14.36% house edge: more than five times worse than roulette, and worse than most slots. Some tables pay 9:1, which improves it to about 4.8% and still leaves it the worst bet available. Side bets like Perfect Pair or Dragon Bonus run 6–13%.

Roads, streaks and pattern betting

Baccarat has the most developed superstition culture of any casino game — big road, bead plate, cockroach pig. Players track banker/player streaks and bet the pattern. Since shoes are near-independent, none of it changes expectation. It changes engagement, which is exactly why operators supply the displays for free.

The one genuine exception is edge sorting, which exploits tiny asymmetries in card back printing — the technique behind the Phil Ivey court cases. Courts in the UK and US ruled against him. It isn't a strategy, and casinos treat it as a dispute, not a skill.

Where 1.41% comes from

Two dice give 36 combinations. On the come-out, pass wins on 7 (6 ways) or 11 (2 ways) = 8. It loses on 2 (1 way), 3 (2), 12 (1) = 4. The remaining 24 set a point — and from there, repeating the point before a 7 is always against you. A point of 4 has 3 ways versus a 7's 6 ways, so you're a 2:1 underdog being paid 1:1.

Work it all the way out and the pass line wins 244 times in 495, or 49.29%. That shortfall is the 1.41%.

How to actually play craps

1. Bet the pass line (or don't pass) — the smallest amount you're comfortable with.
2. Once a point is set, take the maximum odds you can. That bet has no edge, so every chip you move into it lowers your blended house edge.
3. Ignore everything in the centre of the table. All of it.

With 5× odds, a 5-unit line bet backed by 25 in odds gives a blended edge near 0.33% — as good as anything in a casino outside advantage play. The catch is variance: your exposure per roll is six times bigger, so the swings are severe.

Bet the boring number

Look at the table on the Try it tab and you'll see something people find genuinely surprising: the 1 is the best bet on the wheel and the 40 is the worst. Backing 1 costs about 2.1%, which is competitive with roulette. Backing 40 costs more than 12%, which is worse than most slots.

That's not a coincidence, it's the standard shape of casino pricing. The exciting bet always costs more, because excitement is what's being sold.

Spreading bets doesn't help

Covering 1, 2 and 5 at once feels safer and it is — your results get smoother. But your expected cost is just the weighted average of the three edges, so you've bought lower variance at a higher average price than backing 1 alone. Nothing you do with bet placement changes the direction of the maths.

What to notice as someone working in this

• Game shows sit in the casino tab but behave like entertainment products. Their acquisition funnels and their harm profile both look different from slots.
• The presenter is a live employee, so these are the most expensive products in the building to run — which is part of why the margin is higher.
• Bonus rounds are separate bets with their own, usually worse, edges. In Crazy Time the bonus segment bets carry the highest house edge on the layout.
• Several regulators now treat game show advertising as high-risk for appealing to under-18s. Expect that to keep tightening.

The number that matters isn't the RTP

A 1% house edge is the best published price in online gambling. It's better than blackjack. It's better than baccarat. And it is, in practice, one of the most expensive ways to play — because of speed.

Crash costs eight times what blackjack costs per hour despite having twice the... no, half the edge. Read that table again if it feels wrong — it's the single most useful thing on this page, and almost nobody who plays these games has done the sum.

Cash-out targets are a volatility dial, not a strategy

Set 1.20× and you'll win about 82% of rounds in small amounts. Set 50× and you'll win about 2% of rounds in large ones. Both return 99%. There is no clever target. If someone is selling you one, they're selling you the Martingale in a hoodie.

Plinko's real lesson

The 110× buckets are printed at both edges in big friendly numbers, and each one turns up once in 65,536 drops. The 0.3× bucket in the middle turns up once every five drops. Nearly everything you'll see in a session is the middle of the curve, quietly taking 70% of your stake. It's the clearest picture of expected value versus advertised prize anywhere in a casino.

Being able to explain provably fair

If you interview anywhere near crypto casinos or originals, you will be asked this. The short version: a hash published before the round is a commitment the operator can't wriggle out of afterwards. It guarantees the outcome wasn't tampered with. It guarantees nothing about the odds, which are still whatever the pay table says.

The three things that actually win at poker

Poker isn't about the cards you're dealt — over time everyone gets the same cards. It's about making better decisions than your opponents with the same information. Three ideas do most of the heavy lifting.

1. Position

Acting last is a huge advantage, because you've watched everyone else act before you decide. The player on the button acts last every round after the flop, so play more hands there and fewer from early position. A hand you'd fold under the gun is often a raise on the button. This one idea separates winning players from losing ones more than any other.

2. Starting hands

3. Pot odds

This is the maths that turns poker from guessing into a calculable decision. When someone bets, you're being offered a price to keep playing. Compare the price to your chance of winning:

Say the pot is 100 and someone bets 50. You must call 50 to win a pot that would then be 200, so you're paying 50 to win 200 — a quarter, or 25%. If your hand wins more than 25% of the time, calling makes money. Try it below:

The rake, and why it matters

The house takes about 5% of each pot (usually capped at a few big blinds). It sounds tiny, but it's the reason most players are long-term losers even in a game of pure skill — you're not just beating the other players, you're beating them by enough to cover the rake too. In a game where the money just moves between players, the rake is the only guaranteed loser's seat, and everyone pays into it. Rake is how a poker room makes money without ever having a stake in who wins, which is exactly why it's regulated as a service rather than as gambling in many places.