Luma — Signal Aggregator
Luma aggregates, scores, and normalizes signals from technical, on-chain, and sentiment sources.
Overview
Luma is the Signal Aggregation Agent. While Vanta tells you what kind of market you're in, Luma tells you what the signals are saying. It ingests data from multiple source categories, normalizes them into a common framework, and produces confidence-weighted signal scores.
The core insight behind Luma is that no single signal source is reliable on its own. Technical indicators lag. On-chain data can be noisy. Sentiment is fickle. But when multiple independent sources converge on the same conclusion, the probability of that conclusion being correct increases dramatically.
Signal Sources
Luma processes signals across three primary categories:
Technical Signals
- Trend indicators (moving averages, MACD, ADX)
- Momentum oscillators (RSI, stochastic, rate of change)
- Volume-price analysis (OBV, VWAP, volume profile)
- Volatility measures (Bollinger Bands, Keltner Channels, ATR)
- Pattern recognition (support/resistance, chart patterns)
On-Chain Signals
- Exchange flow analysis (inflows, outflows, net flow)
- Whale wallet tracking (large holder behavior)
- Funding rate analysis (sentiment implied by funding)
- Open interest changes (positioning shifts)
- Liquidation clusters (areas of forced selling/buying)
Sentiment Signals
- Social media sentiment scoring
- News sentiment analysis
- Fear and greed index components
- Derivatives positioning sentiment
- Institutional flow indicators
Signal Processing
Luma doesn't just aggregate — it processes signals through several layers:
- Normalization — All signals are mapped to a common -1 to +1 scale
- Confidence weighting — Each signal is weighted by its historical reliability in the current regime
- Correlation filtering — Highly correlated signals are de-weighted to avoid double-counting
- Freshness decay — Older signals receive progressively less weight
- Conflict detection — Contradictory signals are flagged for human review
The output is a composite signal score with a confidence interval, which Riven uses to architect strategies.
Configuration
- Source selection — Enable or disable specific signal categories
- Custom indicators — Add your own technical indicators to Luma's processing
- Confidence thresholds — Minimum confidence for signals to be actionable
- Update frequency — How often Luma recalculates signal scores
- Alert conditions — Notifications for signal convergence events