Transparent by design
Track record
Every fusion read (Bullish Setup, Watch Zone, or Risk Weakening) is logged with a timestamp and fusion probability. Every user Agree / Disagree vote is stored. The model has no memory, so it cannot be optimised post-hoc. What you see is what was called, when it was called.
How the fusion score works
Each indicator scores from −1 (fully bearish) to +1 (fully bullish). Scores are weighted and summed into a single probability: P = 0.5 + (weighted_sum / 2). P ≥ 0.55 → Bullish Setup, P ≤ 0.45 → Risk Weakening, otherwise Watch Zone.
| Indicator | Weight | Logic |
|---|---|---|
| Relative Strength | 18% | IBD-style RS vs Nifty 500 (1y / 3m blend). Underperformers score lower. |
| NSE Delivery % | 20% | Delivery vs intraday turnover. Accumulation signature on Indian rails. |
| EMA Stack 20/50/200 | 16% | 20 > 50 > 200 reads bullish; inverted stacks bearish; mixed reads neutral. |
| Momentum 20D | 13% | 20-session price change. Normalised momentum participation. |
| Volume vs VWAP | 13% | Price vs 20-day rolling VWAP plus volume confirmation. |
| News Sentiment | 10% | Claude Haiku scores recent headlines from −1 to +1. Deliberately light weight. |
| Mean Reversion (BB) | 10% | 20-day Bollinger 2σ contrarian tilt. Trims overextended trend reads. |
Community accuracy stats
Aggregate Agree / Disagree accuracy will appear here once enough votes have accumulated across multiple trading sessions. Every vote you cast on the canvas contributes to the public record.
Open a signal canvas to cast your first vote →Free sign-in
Educational research tool · not investment advice · past performance is not indicative of future results