Ocean swell doesn’t arrive as a uniform heartbeat — it travels in groups called wave trains. A wave train is 5–15 waves that share a period and direction, spaced closely in time. Between trains is a lull of smaller waves or near-flatness. That’s why the lineup feels like "nothing, nothing, SET, nothing, nothing, SET."
When two or more swells are running from different storms (say, a long-period groundswell from a distant North Pacific storm plus a short-period windswell from a nearer local low), their peaks and troughs occasionally align at the same beach at the same moment. When they align constructively, the resulting waves are larger than either swell alone — a clean-up set. When they align destructively, the result is a long lull.
LazySurfer doesn’t predict individual clean-up sets — nothing does, because they emerge from chance alignment in real time. But the app shows the maximum wave height from the buoy alongside the significant (average-of-top-third) height. When those two numbers diverge a lot, clean-up sets are more likely. And if you’ve logged sessions where outsized sets showed up, the ML model learns to flag forecasts with similar buoy signatures.