Every fantasy tool on the internet claims its projections are sharp. Almost none of them show you the receipts.
Here's how we do it: every projection Ribby's daily engine makes gets logged the moment it's made. The next morning, we pull what actually happened and grade ourselves against it — automatically, every night, no cherry-picking. The results live on a scoreboard inside the product we call the Model Report Card: every projection plotted against reality, with the honesty line drawn right through it.
When the model runs hot, you'll see it. When it runs cold, you'll see that too — same chart, same math. No screenshot-the-good-week marketing. The grading is the feature.
This isn't just a trust exercise (though it is that). Grading ourselves nightly is how the product gets better. When the data showed our projections were too optimistic in a specific range, we recalibrated on the graded evidence — not on a hunch. New signals we're testing ride along silently, get graded for weeks, and only earn a place in the model when the backtest proves they help. If a shiny idea doesn't beat the scoreboard, it doesn't ship.
We think this should be table stakes for anyone selling you predictions. Ask your favorite tout where their misses are published.
Ours are in the app. 📊