Slots
Nobody sits down and decides a game "feels like a seven". Each of the four stats ranks the title against similar games from the 13211 library we track. A crash game never gets lined up against a video slot.
What the colours mean
Three of the four stats are percentile ranks, so 5.0 is bang on the category median — half the pack sits below it. The bar is coloured off that midpoint, with a one-point dead zone either side so a single percentile doesn't flip the call:
| Colour | Means |
|---|---|
| green | More than a point above the median for its type. |
| amber | Within a point of the median, either way. |
| red | More than a point below it. |
| gold | We're not making that call. See below. |
Max win stays gold, even on a 50,000x. A bigger cap is paid for with a thinner hit rate — it's a different kind of slot, not a better one. Colouring that green would be selling you a story the numbers don't back.
RTP consistency is the other odd one out. There's no pack to rank against: it's how far the worst certified version sits below the best, scored as 10 minus the spread in points. Under a point is noise, so 9+ is green. More than two points and you're basically on a different game depending on the casino, so below 8 is red.
| Stat | Weight | What it looks at |
|---|---|---|
| RTP | 0.35 | Where the return sits against the median for that type of game. |
| Max win | 0.2 | Where the win cap sits against the same pack. |
| Hit rate | 0.2 | How often it pays, against other games in the same volatility band. |
| RTP consistency | 0.25 | How far the lowest certified version sits below the advertised RTP. A single-version game gets full marks. |
Why RTP consistency is in the score
Studios certify several returns and the operator picks one. The RTP you see on a review is usually the top file, not the one in that lobby. A game that can be hosted ten points light is, in practice, a different slot — so we score that risk instead of pretending it isn't there.
Missing data doesn't tank a game
If we can't score a stat, we drop it and stretch the remaining weights so they still add up. Unknown isn't treated as zero — that would just punish games we haven't finished documenting. Each score says how many stats went into it.
Same idea on the colours. We only draw a median once we've got at least twenty titles in that pack, and the meter tells you the size. Under twenty, the bar stays gold. A verdict you can't check is worse than no verdict.
Worked example: Le Clubber
Scored on 4 of 4 stats. Weights here are after we stretched them, so the bits add up to the total — last decimal might wobble from rounding.
| Stat | Score | Weight | Contribution | Basis |
|---|---|---|---|---|
| RTP | 7.7 | 0.35 | 2.69 | 96.34% against a slots median of 96.00% across 13211 titles. |
| Max win | 7 | 0.2 | 1.40 | 10,000x against 13209 other slots titles. |
| Hit rate | 7.2 | 0.2 | 1.44 | 32.47% against 2428 other medium volatility titles. |
| RTP consistency | 5.8 | 0.25 | 1.45 | 3 certified files from 96.34% down to 92.17%, a 4.17 point spread. |
| Total | 7 |
Open Le Clubber and check the inputs against the page.
Bonuses — the Value Meter
This is modelled EV, not a promise you'll cash out. Expected reward is the headline value minus the house edge on the playthrough, using either the tied game's certified RTP or the video-slot median. Break-even wagering is 1 ÷ house edge — at 96% RTP that's 25×.
| Stat | Weight | What it looks at |
|---|---|---|
| Wagering | 0.4 | How heavy the playthrough is versus break-even on the RTP we used. |
| Expected value | 0.3 | What's left after wagering cost, as a share of the headline value. |
| Cashout cap | 0.2 | How much of that headline value you can actually withdraw. |
| Restrictions | 0.1 | Sticky / non-cashable terms, conversion caps, and short wagering windows. |
Verdict bands
| Score | Verdict |
|---|---|
| 7.5+ | Strong |
| 5 – 7.4 | Fair |
| 2.5 – 4.9 | Weak |
| Below 2.5 | Avoid |
Market medians across scored offers: wagering 1.4, cashout 0.2330768504531722.
Casinos
Casinos are scored 0–10 on five things, then multiplied by the weights below. Homepage and slot-page lists are curated; inside the list, the score still sets the order.
| Criterion | Weight |
|---|---|
| Licence | 0.25 |
| Payouts | 0.25 |
| T&Cs | 0.2 |
| Support | 0.15 |
| Games | 0.15 |
Licence and payouts carry the most weight. A huge lobby doesn't make up for a thin licence or withdrawals that take a week.
Licence tiers
Used for sorting, and as context on the licence sub-score. Lower tier number is stronger.
| Tier | Regulators |
|---|---|
| 1 | UKGC, MGA, DGOJ, AGCO |
| 2 | Isle of Man, Gibraltar |
| 3 | Kahnawake, Curaçao GCB |
| 4 | Other / unverified |
What has to be on file
A licence score above zero needs at least one licence number. A payout score above zero needs a published withdrawal time and a min or max cashout with a currency. We won't publish those sub-scores without that.
Studios
Studio pages show library maths, not a reviewer's score. The figures come off the games we index and the RTP checks we've logged.
Certified RTP versions per game
Average number of distinct certified RTPs per title we index. More versions means more room for an operator to host a weaker file, so this stays gold — it's a trade-off, not a medal.
Low-RTP rate
On multi-RTP games, how often a casino check matches the worst version (within 0.005 points — same tolerance as the slot pages). We only print a percentage once a studio has 20 of those checks; under that we show the count and skip the rate. When we do print it, higher is worse for you.
Table games
Tables only use two stats, because the data has house edge and RTP spread — not hit rate, max win or volatility. On a thin library most titles won't score at all: house edge needs 20 peers of the same type (roulette vs roulette, not roulette vs blackjack).
| Stat | Weight | What it looks at |
|---|---|---|
| House edge | 0.6 | Rank within the same table type. Lower edge scores higher. |
| RTP consistency | 0.4 | 10 minus the certified RTP spread in points — same idea as slots. |
Dispute record
Public cases come off the casino score after the five weighted bits: 0.4 per upheld or partial case, 0.15 per still-open case, capped at 2.0. The summary on the casino page uses the same cases. We're not an ADR body.
What we won't do
- Put up a score without the numbers that produced it.
- Slap a placeholder on a game we haven't reviewed. It just won't have a score.
- Draw a bar at zero for a figure we never measured. That's an empty track.
- Colour a bar against a pack we can't tell you the size of.
- Call a bigger max win "better" when it's really just swingier.
- Let commission move a number. Placement is commercial. The score isn't.
- Quietly change a weight. The figures on this page are the ones the site uses.