What is Signal Fusion?
Signal Fusion is an analytical approach that combines multiple independent data sources and signal types to generate trading signals with higher confidence than any single source could provide alone. It's based on the principle that when multiple uncorrelated indicators align, the probability of a correct prediction increases.
The Problem Signal Fusion Solves
Individual trading signals have limitations:
- Insider buying: Informative but insiders can be wrong
- Dark pool activity: Shows institutional interest but not direction
- 13F filings: Delayed by 45 days
- Technical indicators: Prone to false signals
Each source has blind spots. Signal Fusion addresses this by requiring confluence—multiple independent signals pointing in the same direction.
How Signal Fusion Works
1. Data Aggregation
Collect signals from diverse sources:
- SEC filings (13F, Form 4, 8-K)
- Dark pool and block trade transactions
- Options flow
- Congressional trading
- Short interest changes
- Technical patterns
For deep dives on each source, see What Is Signal Fusion, How to Track Congressional Trades, Dark Pool Trading Explained, and How to Read 13F Filings.
2. Signal Normalization
Standardize signals to comparable scales:
- Convert raw values to percentile ranks
- Weight by historical predictive power
- Adjust for recency and reliability
3. Confluence Detection
Identify when multiple signals align:
- Count unique signal types pointing same direction
- Count unique data sources confirming
- Calculate combined confidence score
4. Confidence Scoring
Generate final confidence based on:
- Number of confirming signals
- Diversity of signal types
- Historical win rate of signal combinations
Signal Diversity Matters
Not all signal combinations are equal:
| Scenario | Confidence |
|---|---|
| 3 insider buys (same type) | Moderate |
| Insider buy + dark pool + 13F (diverse) | Higher |
| 5 signals from 4 different types | Highest |
Type diversity (different categories of data) matters more than raw signal count.
Example: High-Confidence Bullish Setup
A stock might show:
1. CEO purchase (Form 4) — insider confidence
2. Dark pool accumulation — institutional buying
3. 13F new position by respected fund — smart money interest
4. Unusual call buying — options market bullish
5. Short interest declining — bears covering
Five signals from five different types = high-confidence bullish signal.
Filtering and Thresholds
Signal Fusion applies filters to reduce noise:
- Minimum confidence: Typically 55%+ to surface a signal
- Edge requirement: Bullish score must exceed bearish by meaningful margin
- Recency: Recent signals weighted more heavily
- Source diversity: Requires multiple independent sources
Advantages of Signal Fusion
- Reduced false positives: Single-source noise filtered out
- Higher conviction: Multiple confirmations increase confidence
- Diversified risk: Not dependent on any single data source
- Adaptable: Can incorporate new data sources over time
Limitations
- Complexity: More moving parts than single-indicator systems
- Data costs: Requires access to multiple premium data sources
- Latency: Some sources (like 13F) have inherent delays
- Overfitting risk: Must validate that confluence actually predicts
Key Takeaways
- Signal Fusion combines multiple data sources for higher-confidence signals
- Diversity of signal types matters more than quantity
- Confluence across independent sources reduces false positives
- Crossbearing's platform is built on Signal Fusion methodology
- Smart Money Scanner · Pricing & Plans