Forecast check · last 7 days
Every score comes from the same simple idea: compare what the forecast said seven days ago with what actually happened, then grade the difference. We split the last seven days into fourteen periods — seven daytimes (9am–9pm) and seven overnights (9pm–9am) — and score two things.
Temperature counts for 40%. We average how far off the forecast high was across all fourteen periods, then start at 100 and subtract 6 points for every degree of miss. A forecast that ran 2° off on average scores an 88.
Precipitation counts for 60%. For each period we ask one yes-or-no question: did the forecast and the actuals agree on whether at least a hundredth of an inch of rain or snow fell? The score is the percentage of periods that agreed. Getting the rain call right matters more than nailing the exact temperature, so it carries the heavier weight.
The overall score blends the two, 40/60, out of 100. Sky conditions — sunny, cloudy, rainy — are shown on each day card for context but don't affect the score. The state-capital leaderboard is ranked with this exact same formula.
What the forecast said seven days before each period vs. actual — daytime highs on top, overnight lows below.
Day · 9am–9pm highs
Night · 9pm–9am lows
The week-old forecast vs. actual, split into daytime and overnight. Oldest to newest — scroll to see the whole week.