How accurate is Frontside?

Across 324 live calls by surf instructors at 49 spots in Bali, Portugal, California and Hawaii between 20 September and 20 October 2026, Frontside's forecast matched the instructor's call 81% of the time.Surfline, checked against the same calls at the same spot and hour, matched 64%. One month, four coasts, people standing on the beach. The limits are below, and the per-spot table follows once the full set is exported.

What was measured, and how

The question is narrow on purpose: when a surf instructor on the beach called the conditions, which forecast had said the same thing. Every row is one spot at one hour, with three entries against it: the instructor's live call of what the water was doing, what Frontside's forecast said for that spot and hour, and what Surfline published for the same spot and hour.

The instructor's call is the truth in this comparison. It is a person who teaches at that spot, looking at the water as the session happens, not a camera and not another model's output. A forecast counts as right when it matched that call, and wrong when it did not; there is no partial credit.

324 calls at 49 spots in Bali, Portugal, California and Hawaii between 20 September and 20 October 2026, logged as they happened.

The result

Frontside matched the instructor's call on 81% of the 324 calls. Surfline matched on 64% of the same calls. Both numbers are on the same rule and the same rows, so the gap is the comparison, not a difference in method.

The per-spot table, with the spots where Frontside loses, is not on this page yet. The tagging set lives on one laptop and will be published here as soon as it is exported, losses included, because small samples at single spots are how a forecast's weak points are found.

What this does not measure

Two limits, before anyone quotes the headline. The instructors tagged at our request, so they knew whose forecast was being checked - an outside check would remove that, and we would welcome one. And a matched call is coarser than a measurement in centimetres: it says the forecast read the session right, not how close the numbers were.

The method and the raw comparison are ours to share. If you want to pull the numbers apart, or run the same check at your spot, get in touch and we will hand over what we have.

Common questions

How accurate is Frontside's surf forecast?

On 324 live calls by surf instructors at 49 spots in Bali, Portugal, California and Hawaii between 20 September and 20 October 2026, Frontside's forecast matched the instructor's call 81% of the time.

Is Frontside more accurate than Surfline?

On that set, yes: Surfline matched the same instructors' calls 64% of the time against Frontside's 81%, at the same spots and hours. It is one month on four coasts, and the per-spot table is still to be published.

How is surf forecast accuracy measured here?

One spot at one hour, three entries: a surf instructor's live call of the conditions, Frontside's forecast, and Surfline's forecast for the same spot and hour. A forecast is right when it matched the instructor's call. 324 calls over one month.

Compare the spots near youSee which one is actually worth the drive