API: The "FICO Score" for Data Stability

I noticed that in almost every field—from fantasy sports to industrial IoT—people were making decisions based on 'Average' performance. But an average of 50 can mean 50, 50, 50 (stable) or 0, 100, 50 (chaos).

I wanted a way to quantify that chaos.

I built a scoring engine using Python and Numba that calculates a normalized 'Predictability Score' (0-100) for any time-series dataset. It uses the Coefficient of Variation and exponential decay to flag when a system is becoming unstable before it fails.

I launched the MVP on Hacker News and hit #3, confirming that engineers are hungry for better metrics than just 'Mean' and 'Median'. Now, I'm turning it into the standard API for data reliability