How accurate are our snow forecasts?
Every night we freeze that day's 7-day forecast for every resort we track. The next morning we collect what actually fell and score yesterday's predictions at every lead time — every resort, every day, automatically. No cherry-picked storms, no quiet deletions. This page is the result.
Currently scoring 3,273 resorts over a trailing 60-day window, last updated Oct 13, 2026 UTC.
Error by lead time
Average absolute error per forecast day, across resorts that saw snow in the window. Bias is signed: positive means we over-forecast, negative means we under-forecast.
| Forecast | Resorts | Avg. Error | Avg. Bias |
|---|---|---|---|
| Same day | 201 | 0.1in | 0.0in |
| 1 day ahead | 226 | 0.2in | 0.0in |
| 2 days ahead | 242 | 0.2in | 0.0in |
| 3 days ahead | 213 | 0.3in | -0.0in |
| 4 days ahead | 289 | 0.2in | 0.0in |
| 5 days ahead | 245 | 0.3in | 0.0in |
| 6 days ahead | 390 | 0.2in | 0.0in |
By resort
Short-range (1–3 days ahead) forecast error at the snowiest resorts in the window. Resorts appear once they have at least 10 scored days including real snowfall — a dry spell scores a perfect 0.0 that says nothing.
| Resort | Scored Days | Avg. Error | Overall |
|---|---|---|---|
| Centro de Ski Chapa Verde Del Libertador B Ohiggins, Chile | 60 | 1.1in | ran 14% high |
| Valle Hermoso Del Bio-Bio, Chile | 60 | 1.0in | ran 8% low |
| Antuco Del Bio-Bio, Chile | 60 | 1.0in | ran 14% low |
| Las Leñas Mendoza, Argentina | 60 | 0.7in | ran 13% high |
| Caviahue Neuquen, Argentina | 60 | 0.8in | ran 17% low |
| Farellones Metropolitana, Chile | 60 | 0.9in | ran 13% high |
| Coronet Peak Otago Region, New Zealand | 60 | 1.1in | ran 5% low |
| El Colorado Metropolitana, Chile | 60 | 0.8in | ran 24% high |
| Valle Nevado Metropolitana, Chile | 60 | 0.8in | ran 24% high |
| La Parva Metropolitana, Chile | 60 | 0.8in | ran 24% high |
| Los Penitentes Mendoza, Argentina | 60 | 0.6in | ran 11% high |
| Lagunillas Metropolitana, Chile | 60 | 0.8in | ran 7% high |
| Corralco De La Araucania, Chile | 60 | 0.8in | ran 27% low |
| Treble Cone Ski Area Otago Region, New Zealand | 60 | 0.7in | ran 21% low |
| Portillo De Valparaiso, Chile | 60 | 0.9in | ran 47% high |
| Cerro Perito Moreno Rio Negro, Argentina | 60 | 0.4in | ran 16% low |
| Roundhill Ski Field Canterbury Region, New Zealand | 60 | 1.0in | ran 24% high |
| Los Puquios Mendoza, Argentina | 60 | 0.4in | ran 11% high |
| Cardrona Alpine Resort Otago Region, New Zealand | 60 | 0.6in | ran 23% low |
| Ōhau Snow Fields Canterbury Region, New Zealand | 60 | 0.8in | ran 14% high |
| Cerro Bayo Neuquen, Argentina | 60 | 0.3in | ran 17% low |
| Chapelco Ski Resort Neuquen, Argentina | 60 | 0.3in | within 5% overall |
| Parque de Nieve Batea Mahuida Neuquen, Argentina | 60 | 0.4in | ran 47% low |
| Эльбрус, Elbrus Kabardino-Balkar, Russia | 60 | 0.5in | ran 14% low |
| Timberline Lodge Oregon, USA | 60 | 0.4in | ran 33% low |
Why we stop at 7 days
Some sites publish 14- or 15-day snow forecasts. We don't, because forecast error grows with lead time — you can watch it happen in our own table above. Beyond a week, a snowfall total is closer to a coin flip than a forecast, and publishing one would just be marketing dressed as meteorology. Seven days is where a forecast is still worth planning around.
How the scoring works
At the first forecast refresh after 00:00 UTC each day we snapshot the predicted snowfall for each of the next 7 days at every resort. Each night we collect measured snowfall at summit elevation and compute, per resort and per lead day, the average absolute error (how far off we were, regardless of direction) and the bias (whether we run high or low) over the trailing 60 days. Forecasts and measurements both come from the same elevation point, so the comparison is apples to apples.
See what the forecast says for your mountain: