Slot RTP Drifts 1.4% Lower Once Autoplay Passes Spin 300
Independent test house MicroMetrics ran 4.2 billion spins across 38 UK-licensed slots between January and March 2026 and found that the average return to player falls from 96.4% in the first 300 spins to 95.0% once autoplay runs past spin 300 — a 1.4 percentage point drift. The effect held across volatility bands, providers and stake sizes, which rules out a handful of the usual explanations.
What the data actually shows
MicroMetrics split each session at spin 300 and compared the realised RTP on either side. Sessions that stopped at or before 300 spins returned 96.4%, within 0.2 points of the published figure on every game tested. Sessions that continued returned 95.0% on spins 301 and beyond.
The drift is not uniform. It is smallest on low-volatility games (0.6 points) and largest on high-volatility titles (2.1 points). That pattern matters, because it points away from a rigged reel strip and towards something structural about how long sessions are measured.
Three explanations that survive scrutiny
1. Survivorship in the sample
A player who is 300 spins deep is, by definition, a player who has not busted out. On a high-volatility slot, the players still spinning at that point are disproportionately those running above expectation — and mean reversion then drags their realised RTP down towards the theoretical figure. The 95.0% is not a worse game; it is a less flattering sample.
This explains the volatility gradient neatly. High-variance games produce a wider spread of early outcomes, so the selection effect at spin 300 is stronger.
2. Autoplay changes stake behaviour
MicroMetrics also logged stake changes. Players on autoplay were 3.4 times more likely to raise their stake after a win than manual spinners. Since the drift is measured in currency returned per currency wagered, and stakes are not held constant, part of the gap is a denominator effect rather than a payout effect. Strip out sessions with any mid-session stake change and the drift narrows to 0.7 points.
3. Session-length bias in published RTP
This is the uncomfortable one. RTP figures are certified over millions of spins, but operators and affiliates routinely quote them as if they describe a single session. They do not. A 300-spin session on a 96.4% game has a standard deviation of realised return wide enough that anything from 60% to 140% is unremarkable. The "drift" is largely the gap between a long-run average and a short-run distribution being mistaken for the same number.
What UK players should take from it
The Gambling Commission's remote gambling return for 2025 put average slot session length at 24 minutes, or roughly 340 spins at typical spin speeds. That places the average UK session almost exactly on the boundary where the drift appears — which is either a coincidence or a sign that session length is being tuned to the point where the player's experience diverges most from the headline RTP.
The practical read: autoplay past 300 spins does not change the game's mathematics, but it does change what you are likely to observe, and it removes the friction that might otherwise prompt you to stop. If you use autoplay, set a loss limit before you start it, not after. GamStop and the Commission's own research both point to autoplay as a feature that extends sessions beyond intended length, and the numbers above suggest those extra spins are, on average, the least generous ones you will play.
The open question is whether providers should be required to publish session-length-adjusted RTP alongside the headline figure. If a game returns 96.4% over 10 million spins but 95.0% over the 300-to-1,000 spin window where most players actually live, which number is the honest one to put on the help screen?