Probabilistic ratings for Japanese and Korean Netflix titles
Before a title streams, Moriarno produces a pre-release rating from Aaa to B — the probability it will reach the Netflix Global Top 10. Built for producers, investors, and platforms making decisions before the data exists.
How we rate
Ratings are generated from four signal categories: cast track record, studio track record, content characteristics, and release timing. The v1.6 model is a 7-feature ensemble (logistic regression + gradient boosting) trained on 132 historical Japanese Netflix releases. Validated across 80,000+ configurations. LOOCV AUC 0.9176, with a TR_A cap on small-sample 100% studio rates (Session 36 spec, May 2026).
What you get
Every title receives a grade from Aaa to B based on its probability of not reaching the Netflix Global Top 10 — failure probability.
| Grade | Failure Probability | Category |
|---|---|---|
| Aaa | ≤20% | Investment Grade |
| Aa | 20–40% | Investment Grade |
| A | 40–55% | Investment Grade |
| Baa | 55–70% | Investment Grade |
| Ba | 70–90% | Speculative |
| B | >90% | Speculative |
Each title also receives a Magnitude Grade predicting performance conditional on G10 entry: Strong (≥20M viewing hours), Moderate (10–20M), Brief (<10M).
Published forecasts are never retroactively changed. Each rating is tied to a specific model version and issue date — when models improve, new forecasts are issued under the new version. Previous ratings remain on record.
Who this is for
Production companies greenlighting series. Investors evaluating IP. Platforms acquiring titles for global audiences. If you make outcome decisions before the data exists, Moriarno provides the missing layer.
Contact: contact@moriarno.com