Know where you’re going, while you’re getting there.
Live demo · Project It
Nine odd things, tracked from free public data and B1G+A’s own play-by-play, projected right in your browser. No backend, no keys, no magic. Why these? Because we could.
Projections are just that: projections. Model estimates built from the data, not guarantees.
Loading…
Gage height, last 7 days, plus our next 24 hours. Why the river? Because we could.
Loading live data…
Data: U.S. Geological Survey (gauge 06610000). Provisional data, subject to revision.
Counted since midnight Central time, projected to midnight from today’s pace and the last 7 days’ hourly pattern.
Loading live data…
Data: U.S. Geological Survey (earthquake catalog).
English Wikipedia, sampled hourly for 12 hours, projected one hour out.
Loading live data…
Data: Wikimedia (English Wikipedia recent changes API).
Our simple model against the National Weather Service. We’re not the weather service. We’re just curious.
Loading live data…
Data: National Weather Service (api.weather.gov), station KOMA.
Every Big Ten punt this season, laid end to end. 1 mile = 1,760 yards.
Loading live data…
Data: B1G+A play-by-play. Distances are straight-line (great-circle).
Muffed or fumbled punt returns in Big Ten games. Catching a spinning ball with eleven people sprinting at you is hard. These are the times it didn’t work.
Loading live data…
Data: B1G+A play-by-play. Muffs and fumbles after the catch are counted together, both teams.
Big Ten teams keeping the offense on the field on 4th-and-1.
Loading live data…
Data: B1G+A play-by-play.
One number a month, projected one month ahead.
Loading live data…
Data: U.S. Bureau of Labor Statistics, CPI Average Price series APU0000709112.
Throw it back.
The share of away-team home runs that fans throw back onto the field. One number we can’t download.
Data: MLB Stats API (count of away-team home runs only). No affiliation with MLB or any club.
Projections are just that: projections. Model estimates built from the data, not guarantees.
How the projections work
Every model here fits in a sentence, and every one shows how it did. That’s the point: if you can’t explain a projection, you shouldn’t trust it.
Hourly averages of USGS gage height. Next 24 hours = current level + the slope of the last 12 hours, damped 8% per hour (a damped trend). The band is 90%: how wrong the same model was at each horizon over the past week.
Today’s count (since midnight Central) + the remaining hours of an average day from the last 7 days, scaled by today’s pace (trusted more as the day goes). Band: the larger of Poisson noise and the model’s misses at this time of day over the past 6 days.
Edits per minute in a 3-minute window, sampled hourly for 12 hours. Next hour = now + half the slope of the last 6 samples. Band: 1.645 × the model’s past one-step misses.
“Yesterday’s shape at today’s level”: projection = now + (yesterday at that hour − yesterday at this hour). Band from the same model’s misses over the last 24 hours. Shown next to the NWS hourly forecast.
Next month = last price + half the usual change for that calendar month (prior 3 years) + half the average monthly change over the past year. Band = 1.645 × RMSE of the same model over 24 months. Monthly data, fetched at build time.
Season total so far + the trailing 4-week rate per team-game × the team-games left (12 games per team; regular season ends Nov 28, 2026). Band from week-to-week spread.
Live data: U.S. Geological Survey, Wikimedia, National Weather Service. Build-time data: U.S. Bureau of Labor Statistics, MLB Stats API, B1G+A play-by-play. All times Central (CT). Live trackers refresh every 15 minutes.
Start here
Send us the data you already have. We’ll show you where you’re moving, where you’re stuck, and where things are broken, in reports that are easy to read. Starting at $2,500. No build required.