Home / Reality Guides / Slots & Casino Games: How They Really Work
Reality Guide
How slot machines actually work — RNG, RTP, and the tricks that can't beat them
Full guide · Last checked November 2026 · The Betting Site
Slot machines are surrounded by more false belief than any other gambling product: hot machines, due wins, timing tricks, doubling systems. This page replaces the folklore with the licence conditions, the published mathematics and the psychology research — how the random number generator is regulated, where to find a game's real return figure, what the house edge is on the standard casino games, and why every famous “trick” fails on arithmetic, not luck.
The RNG — and the licence condition that bans “due” machines
A licensed slot's outcomes are produced by a random number generator (RNG) governed, in Great Britain, by the Gambling Commission's Remote Technical Standards, RTS 7. The standard requires results to be “acceptably random” — uniformly distributed, unpredictable, non-cycling — and states the sentence that kills a hundred myths: <strong>“Adaptive behaviour (that is, a compensated game) is not permitted.”</strong> A licensed game cannot loosen because you are losing, tighten because you are winning, or store up a jackpot it “owes” anyone. RTS 7C separately bans designs that mislead about likelihood, including substituting losing events with near-miss losing events. Games are tested by Commission-approved test houses before release, against standards whose PAR sheets record the game's theoretical RTP, hit frequency and reel strips.
So “every spin is an independent trial” is not a slogan on a casino's wall — in the UK market it is a licence condition, tested before the game reaches you. The honest criticism of slots is not secret adjustment; it is the published edge and the design psychology, both covered below.
RTP and house edge — find the number before you play
RTP (return to player) is the share of stakes a game is designed to return over its life: a 96% RTP slot has a 4% house edge. UK rules (RTS 3) require game rules and RTP or house-edge information to be easily available before play — so “read the stated RTP first” is actionable advice, not a platitude. RTP is per game, never universal: two slots side by side can differ by several points.
For a real-world sense of scale, Nevada's Gaming Control Board publishes actual results: statewide, slots kept about 7.1% of money played over the twelve months to mid/late 2026 (about 92.9% returned), with the Las Vegas Strip at about 7.7–7.8% and penny slots at about 8.8–9.1%. Those are market averages across thousands of machines, not any single machine's RTP — and Nevada law sets only a 75% minimum theoretical payout, far below typical online figures. The average slot player, in the world's most measured market, loses roughly seven cents of every dollar cycled through the machine.
The house-edge table — the games ranked honestly
House edges are not opinions; they are derivations from the games' rules (the canonical textbook treatment is Ethier, <em>The Doctrine of Chances</em>, 2010). Ranked best to worst for the player:
| Game / bet | House edge | Note |
|---|---|---|
| Blackjack, basic strategy | ≈0.5% | Roughly 0.28% with liberal rules, much worse with poor ones. Tables paying 6:5 on blackjack are the single biggest rule trap — walk past them. |
| Baccarat — Banker | 1.06% | The best standard bet in the game. |
| Baccarat — Player | 1.24% | |
| French roulette, even-money, la partage | 1.35% | Half the standard European edge on even-money bets. |
| Craps — pass line | 1.41% | |
| European roulette (single zero) | 2.70% | Exactly 1/37. |
| American roulette (double zero) | 5.26% | 2/38 — nearly double the European edge for the same-looking wheel. |
| Slots | typically 3–12%+ by design | Nevada's realised average is ≈7.1% (NGCB). Always per game — read the stated RTP. |
| Baccarat — Tie (pays 8:1) | 14.36% | One of the worst standard bets in a casino, sitting one click from the best one. |
Losses disguised as wins, and near-misses — what your brain is being shown
Losses disguised as wins (LDWs). On multiline slots, an outcome can pay back less than your total stake and still be presented with win sounds and animation. Dixon and colleagues (2010, Addiction) measured players' skin-conductance responses: LDWs produced arousal equal to real wins and greater than plain losses. Harrigan, Dixon & Brown (2015) found that with 20 lines bet, the majority of “hits” on the simulated commercial game were LDWs — most of what feels like winning is getting part of your stake back, celebrated. Dixon et al. (2018) found players systematically overestimate how often they won. UK rules now ban celebrating a return at or below the stake (RTS 14F) — the regulation exists because the trick worked. Near-misses. Two jackpot symbols with the third just off the line feel like closeness. Clark et al. (2009, Neuron) found near-misses were rated less pleasant than full misses yet increased the desire to keep playing, recruiting win-related brain circuitry — despite paying nothing. The effect was strongest when the player had personal control over the spin. Clark's 2010 review states the bottom line: near-misses and personal control have no objective influence on winning chances. A near-miss is a loss wearing a costume.The tricks, one by one
Martingale and progression systems. Doubling after every loss feels like engineering; it is arithmetic with the signs unchanged. Every individual bet keeps the same negative expected value, so any sequence of them has negative total expectation — a system rearranges when you lose, not whether. Martingale converts many small wins into rare catastrophic losses, and table maximums plus a finite bankroll guarantee the catastrophe is unrecoverable when it arrives: ten straight losses from a £10 base — unremarkable on an even-money bet — require a £10,240 next stake, with £20,470 already lost, to win back £10. Hot and cold machines; “due” wins. Impossible under RTS 7: independent trials, no memory, adaptive behaviour prohibited. A machine cannot be due, because it does not know what it paid before. Believing otherwise is the gambler's fallacy with a licence condition written against it. Timing and button tricks. Stop buttons, spin timing, “playing a machine someone just left” — no mechanism exists by which any of them can affect an RNG outcome drawn when the spin starts. Slam-stop was banned in the UK in 2021 precisely because it fed an illusion of control, not because it worked. Autoplay. Banned in the UK since 2021 because it dissociated play from spending. Where it still exists, it changes exactly one thing: bets per hour. Expected loss per hour rises in proportion — autoplay is a trick that works, on you.What actually changes your expected loss
Expected loss per hour = house edge × stake × bets per hour. Only three levers exist, and the edge is never one of them — but game choice decides <em>which</em> edge you accept: European over American roulette (2.70% vs 5.26%), Banker over Tie (1.06% vs 14.36%), basic-strategy blackjack (≈0.5%), and reading a slot's stated RTP before playing it. Stake and speed are wholly yours: a slower game at smaller stakes loses less per hour at the same edge, with the same entertainment. That is the complete list of real control. Everything else is a story told about randomness — and if the stories stop working and the losses stop being entertainment, our <a href="/guides/self-control-and-problem-gambling/">self-control guide</a> and <a href="/guides/how-to-stop-gambling/">recovery guide</a> are the pages that matter.
The UK's design rules double as a museum of what used to be legal: from 31 October 2021, slots lost autoplay, turbo and slam-stop, gained a 2.5-second minimum game cycle and the ban on celebrating losses as wins; from 17 January 2025 the package extended to other casino games, with a 5-second minimum cycle (RTS 14). Regulators do not ban features that do nothing.
Sources
Every figure and finding on this page traces to the research sources below — regulators, official schemes, courts and peer-reviewed work. Where a claim could not be verified against a primary source, it was left out.
- UK Gambling Commission — RTS 7, generation of random outcomes
- UK Gambling Commission — RTS 14, responsible product design
- Dixon et al. (2010) — losses disguised as wins, Addiction (PubMed)
- Clark et al. (2009) — near-misses and gambling motivation, Neuron (PMC)
Keep reading
- Value betting, explained honestly — EV, overround and CLV
- Self-control & problem gambling — what actually helps
- Odds explained — formats, implied probability and the margin
- Getting out: gambling addiction recovery, step by step
Slots & Casino Games: How They Really Work FAQ
Are slot machines rigged?
Not in the adaptive sense, at a UK-licensed operator: RTS 7 requires acceptably random results and expressly prohibits compensated or adaptive games — a machine cannot tighten or loosen based on your play, and games are tested by approved test houses before release. The legitimate criticism is different: the edge is designed in, published, and compounded by speed and design psychology.
What is RTP and where do I find it?
RTP (return to player) is the designed long-run share of stakes returned: 96% RTP means a 4% house edge. It is per game, never universal. UK rules require RTP or house-edge information to be easily available before play — check the game's rules or information panel. If a game will not tell you its RTP, that silence is information too.
Does the martingale system work?
No. Each bet keeps the same negative expected value, so the sequence does too; doubling only changes the shape of your losses — many small wins, then one loss too large to chase. Ten straight losses from a £10 base need a £10,240 stake to recover £10, with £20,470 already gone, and table limits cut the ladder off before the bankroll does.
Is a machine ever “due” a win?
No. Every spin is an independent random draw — under UK licence conditions the game has no memory and adaptive behaviour is prohibited. A machine that has not paid out is not storing anything up, and one that just paid is not 'hot'. 'Due' is the gambler's fallacy.
Which casino game gives me the best chance?
The lowest published edges: blackjack played with basic strategy (about 0.5%), baccarat Banker (1.06%), French roulette with la partage on even-money bets (1.35%), and the craps pass line (1.41%). The worst standard bets include the baccarat Tie at 14.36% and American roulette at 5.26%. No game offers a positive edge to the player — 'best' means slowest loss, not profit.
Reader comments
Comments show on this page. No account needed — please keep it about your own experience; we remove spam and abuse, and never edit real comments.
Loading comments…