What Are Predictive Scores?
Predictive Scores estimate the probability of future market movements for narratives, tokens, and sectors. They answer: "Is this narrative going to moon?" or "Will Bitcoin break $100k?"
Key Metrics
- Probability Upside (0-100) โ % chance of positive price movement in next 30 days
- Confidence Interval โ Lower and upper bounds (e.g., 55% to 75% upside probability) reflecting model uncertainty
- Probability Breakout โ Chance of +50% move in favorable direction
- Probability Breakdown โ Chance of -30% move in adverse direction
- Momentum Score โ Current short-term momentum (0-100)
Model Inputs (What We Feed It)
- Alpha Signals โ Top 20 tokens from daily alpha scan
- Whale Tracker โ Large wallet movements + conviction scores
- Narrative Heat โ Current narrative dominance % and velocity
- Smart Money Moves โ Institutional wallet activity
- On-Chain Metrics โ Active addresses, whale transactions, exchange inflow/outflow
- Historical Signal Outcomes โ How accurate have past signals been?
How Confidence Intervals Work
A score of "65% upside (55-75% CI)" means: the model's best estimate is 65% probability of upside, but it could reasonably be as low as 55% or as high as 75% depending on data richness and model variance.
When confidence is wide (55-75%, 20% range), the model is uncertain. When confidence is narrow (63-67%, 4% range), the model is confident.
Update Schedule
Predictive Scores are updated daily at 08:00 UTC using fresh signal data from overnight runs.
Accuracy & Disclaimers
- Backtested, not perfect. See Signal Credibility for historical hit rates.
- Black swan events. Market crashes, regulatory shocks, and hacks are unpredictable.
- Not financial advice. Do your own research before investing.
- Model is only as good as its inputs. If whale signal is stale, prediction is stale.
Key Drivers
Each Predictive Score includes a list of Key Drivers โ the top 3-5 signals that contributed most to the probability. Use this to understand why the score is high or low.