The Surf Exaggeration Myth

So the meme goes like this: at Surfline, we have some perverse incentive to exaggerate upcoming surf events because that increases web traffic, premium subscribers and advertiser revenue. I hear it here at Surfline, and heard it plenty in my years leading Magicseaweed.

Never-mind that our advertisers are largely long-term partners who agree to work with us months in advance, or that our web traffic is consistently high anyway. Or even that it’s a trick you could only reasonably pull a few times before you started to see a negative payoff — which doesn’t make much sense for a company with a decades-long reputation to protect. It’s a myth that just won’t die.

We get as excited as any surfer when we see a blob like this on a chart, but our job is to give calm, clear headed guidance and that’s what we seek to do.

The good news is, it’s both untrue and provably so. We just need to look at the bias in our forecasts — that is, any error we make in our surf predictions that’s consistently low or high. Here’s the data for the last year for Orange County:

Bias for Orange County is almost non-existent, but where it exists it’s actually negative in the longer range. That is to say we have a tendency to fractionally under call long range events.

It’s a pretty boring graph. While our error increases as you look further ahead (which is bog standard for a weather forecast) our bias remains really low — and actually trends negative over time. That’s to say we have a very slight tendency to under-call long-range events.

But this is the average of all events, which is not a completely fair test. The myth we’re trying to debunk here is over-cooking the more interesting events. We have data for that, too.

We can look at bias just for events that arrive in the ‘fair to good’ + range. It’s immediately noticeable that the negative bias here is actually higher. That means we’re actually a little more cautious on these above-average events than normal. Or that these events are larger, so our caution — although proportionate — is larger in absolute terms.

The tendency towards caution actually increases for days rated ‘fair to good’ or better.

Of course having a low bias doesn’t mean we’re perfectly accurate in the long range. We’re not. Neither does it mean we always under-call future events. We will sometimes over and underestimate distant swells — but we make equal error in both directions. We’re as accurate as we know how to be.

That this myth exists likely evolves from the way we recall those errors when they do occur. Scoring something better-than-expected feels like a win; you own the positive outcome. Less-than-expected feels like you’ve been cheated and someone else is to blame. Of course the ‘someone’ here is your friendly local surf forecaster.

This article is part of a series of attempts to expose the truth about surf forecasting, warts and all. That means taking on pernicious myths that the data disproves — but also being honest about the issues with modern surf forecasting and talking about the steps we’re taking to fix them. If there’s a theory you have that you want explored or questions about this or any other surf forecast topic please fire away in the comments below.

Just to be brutally honest there are some locations where we show some positive bias in the long term. Here’s the data for the Outer Banks where our forecasts are consistently overcalling the long-range swell by an average of slightly under two inches. For anyone making critical go/don’t go decisions at that level of precision — we’re sorry.