The Predictability Score is not just a random number; it's a robust statistical measure designed to quantify the consistency and reliability of any data series. This page provides a high-level overview of the principles behind our calculation.
At its heart, every predictive system or data set contains two components:
A system is "predictable" when the signal is strong and the noise is low. Our score measures this relationship by calculating the size of the "noise" relative to the "signal."
Different industries have different tolerances for volatility. A "noisy" stock portfolio might be acceptable, but a "noisy" pharmaceutical batch is not. Our Pro and API tiers allow you to adjust the sensitivity of the calculation using the "k-factor" to match your specific domain, from the leniency needed for sports analytics to the strict precision required for medical manufacturing.
Our algorithm is based on a standard statistical model (the Coefficient of Variation combined with an exponential decay function) to ensure the score is both reliable and defensible. We believe in transparent math, not "black box" solutions. You can see the full formula and an explanation of the "k-factor" directly on our calculator page.
