Studios
How often a studio's games return nothing, counted from casinos' public bet feeds. This is not a return to player figure and we do not publish one: see why below.
| # | Studio | Games | Rounds seen | Returned nothing |
|---|---|---|---|---|
| 1 | Hacksaw Gaming | 270 | 69,425 | 90.0% ±0.22 pts |
| 2 | Pragmatic Play | 949 | 56,584 | 83.9% ±0.30 pts |
| 3 | Play'n GO | 145 | 15,510 | 87.8% ±0.52 pts |
| 4 | Nolimit City | 68 | 7,486 | 88.2% ±0.73 pts |
| 5 | ELK Studios | 94 | 6,734 | 88.3% ±0.77 pts |
| 6 | BGaming | 76 | 4,562 | 78.1% ±1.20 pts |
| 7 | 1spin4win | 116 | 3,971 | 88.3% ±1.00 pts |
| 8 | Evolution | 346 | 2,513 | 52.5% ±1.95 pts |
| 9 | Gaming Corps | 38 | 2,511 | 90.9% ±1.12 pts |
| 10 | Clawbuster | 31 | 2,110 | 85.4% ±1.51 pts |
| 11 | Relax Gaming | 78 | 1,914 | 86.4% ±1.54 pts |
| 12 | PG Soft | 32 | 1,707 | 79.1% ±1.93 pts |
| 13 | Penguin King | 17 | 1,276 | 81.0% ±2.15 pts |
| 14 | Spinomenal | 116 | 1,176 | 83.9% ±2.10 pts |
| 15 | Novomatic | 30 | 1,033 | 82.0% ±2.34 pts |
| 16 | Push Gaming | 48 | 916 | 90.0% ±1.95 pts |
| 17 | 3 Oaks Gaming | 47 | 890 | 80.4% ±2.61 pts |
| 18 | BetSoft | 46 | 858 | 88.2% ±2.16 pts |
| 19 | Quickspin | 32 | 838 | 86.4% ±2.32 pts |
| 20 | Endorphina | 27 | 735 | 81.4% ±2.82 pts |
| 21 | AvatarUX | 44 | 717 | 78.8% ±2.99 pts |
| 22 | Mascot | 13 | 648 | 84.6% ±2.78 pts |
| 23 | Voltent | 54 | 618 | 90.8% ±2.28 pts |
Why there is no RTP column
The obvious version of this page is a studio RTP table. We cannot publish one honestly, and it is worth being exact about the reason.
Pooling a studio's games averages together distributions with completely different scales, a 5,000x slot beside a 2x table game, so the variance grows faster than the sample. Hacksaw across 53,000 observed rounds gives a mean of 124% with an interval of plus or minus 12 points. No casino game returns more than it takes, so that figure is impossible, and every studio we track fails the same way. More data makes it worse, not better.
The share of rounds that returned nothing is a proportion. Its error is sqrt(p(1-p)/n), which settles quickly, so the same 53,000 rounds give 89.9% to within a quarter of a point. Same data, 45 times the precision, and a figure that is actually true.
Three things are excluded before any of this is counted. A casino's own games, because they are not a studio's work. Distribution platforms such as hub88 and softswiss, because they publish other studios' games and crediting them names the wrong company. And any game on any feed whose observed return is provably above break even, because that proves the feed is not showing every round of it: leaving those in moved one studio's figure by 13 points against an interval of one. Our methodology sets out the rest.