Forecast Accuracy
We re-ran the exact same model at past cutoffs and compared its predictions to what actually happened. Anyone can output a number — Shadow proves its confidence bands are calibrated.
Band Coverage
77%
Actuals inside p10–p90 (target 80%)
Mean Error
1.5pp
Avg |actual − predicted p50|
Backtests Run
35
Rolling cutoffs across segments
Predictions Scored
604
Predicted-vs-actual data points
Calibrated. Reality landed inside Shadow's stated confidence band 77% of the time — within tolerance of the 80% target. The bands mean what they say.
Measured on demo fixtures, not live telemetry. These segments are generated series, so this page proves the backtest harness and the band arithmetic are sound — not that the model predicts your product well. That claim only holds once it runs on your data.
New Users · iOS · UK · Standard — 14-Day Activation Rate
Predicted band vs. actual. Shown because it has the most scored points (72) — the most evidence, not the best result. Drawn from the earliest cutoff, the longest horizon.
92% covered1.7pp error
rendering forecast…
Per-Segment Calibration
Coverage and error for every backtested segment, furthest from the 80% target first
New Users · iOS · UK · Standard — Card Provisioning Rate72 pts2.6pp err63% covered
New Users · Android · EEA · Standard72 pts2.4pp err65% covered
New Users · iOS · UK · Standard — 14-Day Activation Rate72 pts1.7pp err92% covered
New Users · iOS · UK · Standard — 30-Day Premium Upsell Rate72 pts1.2pp err71% covered
New Users · iOS · UK · Standard — Salary Deposit Set-up Rate7 pts0.7pp err71% covered
New Users · iOS · UK · Standard — First Transaction Rate72 pts1.1pp err88% covered
Activated · iOS · US · Metal · Investors72 pts0.9pp err88% covered
New Users · iOS · UK · Standard — Identity Verification Rate72 pts1.1pp err74% covered
Activated · iOS · UK · Premium72 pts0.9pp err78% covered
New Users · Android · India · Prepaid/UPI21 pts1.8pp err81% covered