Data study

How far ahead can you trust a surf forecast? We checked a year of them

Every surf app forecasts out a week or more. Nobody publishes how often those calls come true. We took a year of forecasts made 1 to 7 days ahead at every break we cover, and checked them against the surf that showed up.

By · · 7 min read·

A giant wave at Nazaré, Portugal
91%of days called in the right size band 1 day outflat, small, fun, solid, big or huge
70%7 days outsame 114 breaks, same days
0.3 fttypical miss on the face 3 days outmedian, against the day's analysis
76%of 8 ft and bigger days were called big 7 days out93% the day before

The short answer

A surf forecast is close to right for 2 or 3 days. By a week out, treat the size as a rough guess.

We checked 39,444 break-days of forecasts at 114 breaks, from 1 December 2025 to 11 November 2026. Each was scored against the model's own picture of that day once it came. A forecast made 1 day ahead put the day in the right size band 9 in 10 times. At 3 days it was 8 in 10, and at 7 days 7 in 10. The typical miss on the face grew from 0.1 ft the day before to 0.6 ft a week out.

What we scored

A surfer does not care about a model's error in meters. The question is simpler: did the forecast call the right kind of day? So we sorted every day into a size band a surfer would name: flat, small, fun, solid, big or huge. A forecast was right when it put the day in the same band as the surf that came.

We also kept the miss in feet on the face, and the miss on the period and the swell direction. The exact bands are under how we measured.

Day by day of lead time

Forecast madeRight size bandOff by 2 bands or moreWithin 1 ftTypical missPeriod missDirection miss
1 day ahead91%0%96%0.1 ft0.2 s1°
2 days ahead85%0%89%0.3 ft0.3 s2°
3 days ahead81%1%85%0.3 ft0.4 s3°
4 days ahead79%1%81%0.4 ft0.5 s3°
5 days ahead76%1%78%0.4 ft0.5 s4°
6 days ahead74%2%74%0.5 ft0.6 s4°
7 days ahead70%2%70%0.6 ft0.7 s5°

Read the table as a slope, not a cliff. Each day of lead time costs a little. The first few days cost the least, because the storm that makes the swell is already on the chart. Further out, the storm itself is still a forecast.

The period and the direction hold up much better than the size. A forecast a week out usually knows which way the swell comes from and how long its period is. It is less sure how big it gets.

Big days: called, but called small

We counted 8,137 break-days with faces of 8 ft and up. The day before, the forecast called 93% of them big. A week out it called 76%, about as often as it got any day's band right. The miss in feet was bigger, though. On a typical big day it was 0.3 ft the day before, and 1.3 ft a week out. A week out, the forecast ran low on these days by 0.8 ft on average. Big swells often look smaller on the early runs.

If you plan around a big swell, watch it for a few days before you book. A big swell that holds on the forecast run after run is the one to fly for.

The buoy check

The size bands above score the forecast against the model's own analysis of each day. To check against measured waves, we used 24 NOAA buoys near our US breaks. We compared each buoy's wave height with the forecast for the same spot. On the day itself, the model's height was within 20% of the buoy on 71% of days. A week ahead, that fell to 51%. On average the model read 0.4 ft below the buoys.

Forecast madeWithin 20% of the buoyTypical missDays checked
Same day71%0.5 ft7,700
1 day ahead70%0.5 ft7,700
2 days ahead66%0.5 ft7,700
3 days ahead62%0.6 ft7,700
4 days ahead60%0.6 ft7,700
5 days ahead58%0.6 ft7,700
6 days ahead56%0.7 ft7,700
7 days ahead51%0.7 ft7,700

Buoy: the biggest hourly significant wave height between 6 am and 6 pm local. Model: the same forecast, read at the buoy's own position. Each buoy sits within 80 km of a break we cover. This is the deep-water height, not the face.

Some coasts are easier to call

We compared the coasts with 3 or more breaks. A week out, Peru and Ecuador held up best, with 82% of days in the right band. US Northeast held up worst, at 61%.

Coast1 day ahead3 days ahead7 days aheadBreaks
Peru and Ecuador94%87%82%4
Central America92%85%78%10
US West Coast92%85%76%10
Mexico's Pacific coast92%84%75%6
Caribbean91%82%74%3
Portugal94%84%72%4
Morocco and the Canary Islands94%84%72%4
Indonesia91%78%71%9
Brazil92%79%71%3
US Southeast91%81%69%7
Bay of Biscay91%80%68%6
US Mid-Atlantic91%79%67%3
South Africa and Namibia91%78%67%4
Australia's East Coast93%81%63%3
Ireland and the UK92%83%62%4
US Northeast88%78%61%16

Share of days in the right size band.

Coasts fed by long-period groundswell from distant storms tend to forecast best. The swell is on the chart for days before it lands. Coasts fed by storms close to shore tend to forecast worst. A nor'easter or a tropical storm can form, swing and fade inside the forecast window.

Across all breaks, a 5-day forecast did best in August, with 80% of days in the right band. It did worst in February, at 71%.

What it means for a surf trip

For how a forecast is built, and where each step slips, read how surf forecasts work. For why two apps print different numbers for the same swell, read why surf forecasts disagree.

Every break

Share of days the forecast put in the right size band, 1 December 2025 to 11 November 2026. Sorted by the 7-day column. A break that is often small is easier to call, so check its average face.

Break1 day ahead3 days ahead7 days aheadAverage face
Cacimba do Padre
Brazil
97%89%87%1.2 ft
Soup Bowl, Bathsheba
Barbados
93%90%85%4.9 ft
Lobitos
Peru
95%86%84%6.1 ft
Chicama
Peru
93%86%83%6.4 ft
Dominical
Costa Rica
92%86%83%6.2 ft
Blacks Beach
USA
93%86%83%5.5 ft
Playa Guiones
Costa Rica
93%87%83%6.2 ft
Lakey Peak
Indonesia
93%80%82%7.8 ft
Lower Trestles
USA
92%88%82%5.0 ft
Montañita
Ecuador
94%88%82%4.5 ft
Swami's
USA
93%88%82%5.4 ft
Santa Teresa
Costa Rica
91%86%81%6.7 ft
Tamarindo
Costa Rica
93%86%81%5.7 ft
Máncora
Peru
93%86%81%5.3 ft
Margaret River (Main Break)
Australia
96%89%81%13.9 ft
Scorpion Bay (San Juanico)
Mexico
95%85%80%4.4 ft
Playa Encuentro, Cabarete
Dominican Republic
93%86%80%5.2 ft
Pavones
Costa Rica
91%85%80%6.1 ft
Banzai Pipeline
USA
95%85%79%3.4 ft
Santa Catalina
Panama
92%82%78%4.7 ft
Steamer Lane
USA
91%87%78%7.5 ft
Popoyo
Nicaragua
90%84%78%6.2 ft
Cokes, Thulusdhoo
Maldives
89%84%78%3.9 ft
Malibu First Point
USA
91%87%78%4.9 ft
Huntington Beach Pier
USA
93%86%78%4.9 ft
Cocoa Beach Pier
USA
93%85%77%1.9 ft
Imsouane (The Bay)
Morocco
95%87%76%7.6 ft
Anchor Point, Taghazout
Morocco
95%84%76%5.9 ft
Playa Cerritos
Mexico
91%84%75%6.6 ft
Weligama
Sri Lanka
95%85%75%5.9 ft
La Saladita
Mexico
91%87%75%6.3 ft
Nazare (Praia do Norte)
Portugal
96%86%75%13.0 ft
Skeleton Bay
Namibia
95%82%75%7.4 ft
Punta Roca
El Salvador
93%85%74%5.8 ft
Lances Right (HTs)
Indonesia
90%80%74%8.5 ft
C Street
USA
93%82%74%4.1 ft
El Sunzal
El Salvador
92%83%73%5.4 ft
Hossegor (La Graviere)
France
93%83%73%9.2 ft
Jacksonville Beach Pier
USA
90%82%73%2.7 ft
Sayulita
Mexico
91%84%73%3.6 ft
Punta de Lobos
Chile
89%84%73%8.5 ft
Puerto Escondido (Zicatela)
Mexico
91%83%73%8.0 ft
Wrightsville Beach
USA
91%83%73%2.2 ft
Batu Bolong, Canggu
Indonesia
92%80%73%7.1 ft
Playa Venao
Panama
92%84%73%4.6 ft
Ribeira d'Ilhas
Portugal
94%85%73%9.9 ft
Honolua Bay
USA
90%81%72%3.8 ft
The Washout, Folly Beach
USA
91%81%72%2.3 ft
G-Land (Grajagan)
Indonesia
90%75%72%9.5 ft
Raglan (Manu Bay)
New Zealand
94%76%72%8.2 ft
Westport (The Jetty)
USA
92%84%72%7.3 ft
Barra de la Cruz
Mexico
91%84%71%6.1 ft
Supertubos, Peniche
Portugal
96%84%71%9.4 ft
Famara
Spain
94%81%71%7.6 ft
Padang Padang
Indonesia
92%80%71%10.0 ft
Piha
New Zealand
93%80%71%9.7 ft
T-Land (Nemberala)
Indonesia
91%81%71%6.0 ft
Cloudbreak
Fiji
91%81%71%8.1 ft
Côte des Basques
France
93%84%70%6.4 ft
1st Street Jetty, Virginia Beach
USA
92%78%69%2.2 ft
New Smyrna Beach (Ponce Inlet)
USA
88%79%69%3.2 ft
Pacific City (Cape Kiwanda)
USA
90%81%69%7.3 ft
Tonel, Sagres
Portugal
91%79%69%8.4 ft
Uluwatu
Indonesia
90%75%69%9.3 ft
Lagundri Bay, Nias
Indonesia
93%75%68%8.4 ft
New Pier, Durban
South Africa
89%78%68%7.0 ft
Ngor Right
Senegal
94%78%68%7.2 ft
First Point, Noosa Heads
Australia
91%81%68%4.3 ft
Teahupoo
French Polynesia
92%74%68%10.1 ft
Zarautz
Spain
92%78%68%5.5 ft
Arugam Bay (Main Point)
Sri Lanka
87%79%67%4.4 ft
Mundaka
Spain
92%77%67%8.7 ft
Itaúna, Saquarema
Brazil
90%73%66%6.6 ft
Jennette's Pier, Nags Head
USA
91%81%66%3.0 ft
Gilgo Beach
USA
90%81%66%3.0 ft
Sopelana (Arrietara)
Spain
89%79%65%6.5 ft
Casino Pier, Seaside Heights
USA
89%81%65%2.8 ft
Muizenberg
South Africa
90%76%65%6.1 ft
Cape Hatteras
USA
89%77%65%3.2 ft
Rockaway Beach
USA
91%81%65%2.8 ft
Bells Beach
Australia
89%77%65%6.0 ft
Mavericks
USA
92%82%64%8.2 ft
Desert Point (Bangko-Bangko)
Indonesia
92%76%64%9.2 ft
Monmouth Beach
USA
91%80%64%2.9 ft
The Pass, Byron Bay
Australia
94%78%64%5.5 ft
Lahinch
Ireland
92%84%63%8.9 ft
Somo (Santander)
Spain
88%81%63%6.8 ft
El Cotillo (Piedra Playa)
Spain
93%84%63%6.0 ft
Matunuck
USA
88%82%63%3.4 ft
Sebastian Inlet (First Peak)
USA
91%77%63%3.2 ft
Fistral Beach
United Kingdom
89%82%63%7.3 ft
Narragansett Town Beach
USA
90%82%63%2.9 ft
Long Beach
USA
90%79%62%3.2 ft
Bundoran (The Peak)
Ireland
92%85%62%7.6 ft
Holyoke, Long Beach Island
USA
86%75%62%3.6 ft
Robert Moses Field 5, Fire Island
USA
90%80%62%3.5 ft
Donghe
Taiwan
90%78%62%6.0 ft
One Eye, Tamarin
Mauritius
92%75%61%6.5 ft
Praia Mole, Florianopolis
Brazil
89%74%61%2.4 ft
Thurso East
United Kingdom
93%81%61%8.6 ft
Point Judith (Lighthouse)
USA
89%80%61%3.3 ft
Manasquan Inlet
USA
84%74%60%3.6 ft
Praia de Santa Bárbara
Portugal
93%80%60%5.2 ft
Tsurigasaki (Shidashita)
Japan
90%77%60%5.0 ft
Belmar 16th Avenue
USA
87%79%60%3.4 ft
Snapper Rocks (Superbank)
Australia
94%82%59%4.3 ft
Cloud 9
Philippines
84%72%58%5.9 ft
Reef Road, Palm Beach
USA
91%81%58%3.8 ft
Jeffreys Bay (Supertubes)
South Africa
92%77%58%4.9 ft
Rincon (Tres Palmas)
USA
87%70%58%3.6 ft
Ditch Plains, Montauk
USA
87%76%57%3.8 ft
Montauk Point (Turtles)
USA
88%75%56%4.2 ft
The Bowl, Shinnecock Inlet
USA
85%73%55%4.1 ft
Ruggles, Newport
USA
86%74%52%3.7 ft

How we measured

The forecasts are Open-Meteo's marine forecast, which blends Météo-France's MFWAM and ECMWF's WAM. Open-Meteo keeps what its forecast said 1 to 7 days before each hour. We took every day from 1 December 2025 to 11 November 2026, 346 days, at 114 breaks.

For each day we took the biggest face between 6 am and 6 pm local time. We used the same physics as the BarrelFare app, with each break's own swell window and focus. "What showed up" is the model's own analysis of that day, which takes in buoy and satellite data. The buoy check below it uses only measured waves. The archive keeps the total sea, not each swell, so these are total-sea faces.

The size bands, on the face, are:

A day counts as right when the forecast and the day fall in the same band.

Every break's numbers are in one CSV: download the table. You may reuse it under CC BY 4.0 if you credit BarrelFare and link to this page.

How to cite

BarrelFare, "Surf forecast accuracy study", 13 November 2026, https://barrelfare.com/articles/surf-forecast-accuracy. Data: https://barrelfare.com/articles/surf-forecast-accuracy.csv. License: CC BY 4.0.

Fly when the swell shows up.

The app watches the forecast and the fares at every break in this study, and tells you when they line up.

See where it's firing Get the app