How Accurate Is a Snow Day Predictor?

Short answer: a snow day predictor is a probability estimate, not a forecast of a decision. It can tell you how a given weather setup usually plays out for schools. It cannot read a superintendent’s mind. Here’s how to think about accuracy honestly.

What the predictor actually knows

Our predictor pulls the live hourly forecast for your exact location — snowfall, temperature, freezing rain, wind, and existing snow depth — and scores the overnight-into-morning window with a transparent model. Everything it knows comes from that forecast. When the forecast is good (a clear, well-modeled storm a day out), the estimate is solid. When the forecast itself is uncertain (a borderline rain/snow line, a storm five days away), the estimate inherits that uncertainty.

What it can’t know

The specific district’s tendencies. Two towns five miles apart can make opposite calls. We approximate this with region and district-type adjustments, but we don’t model every district individually.
Human and political factors. Days already missed, an upcoming exam schedule, a cautious vs. hardy administration — none of that is in a weather feed.
Last-minute forecast busts. A storm that tracks 50 miles north, or warms two degrees, can turn a snow day into a normal one overnight.

Why we show the factors

A black-box percentage is easy to distrust. That’s why every result lists the top contributing factors — “6″ overnight snow,” “ice,” “timed for the 4 a.m. commute.” If the drivers match what you see out the window, the number is more believable. If they don’t, you know to discount it.

How to read the number

Above ~80%: a closure is likely; still confirm with your district.
35–65%: genuinely uncertain — this is where a 2-hour delay often happens instead.
Below ~15%: plan on school as normal.
On a weekend or holiday: the odds are zero because there’s nothing to cancel.

A good rule: check the predictor the night before for a sense of the odds, then confirm with your actual district in the morning. We’re a fun, fast second opinion — not the official announcement. See exactly how the model is built on the methodology page.