EPIC IDENT™ is the proprietary order-flow perception module at the core of the EPIC Agentic AI trading platform. It continuously processes live market microstructure data to provide the agent swarm with high-resolution situational awareness of institutional positioning, liquidity dynamics, and algorithmic participant behavior.Primary Functions

• Liquidity absorption & iceberg detection
• Machine-liquidity fingerprinting (recurring order-flow signatures from dominant automated participants)
• Aggressive/passive imbalance quantification at the inside and deeper book levels
• Sub-second regime-change alerts (spoofing, exhaustion, acceleration)

How It Works

1. Data Ingestion – Direct order flow feeds + select venue-specific streams.
2. Feature Extraction – Real-time calculation of order-book delta, reject rates, latency fingerprints, time-of-day seasonality, volatility scaling.
3. Pattern Classification – Hybrid rule-based + reinforcement-learning classifiers trained on nine years of labeled microstructure sequences.
4. Fingerprint Matching – Maintains a dynamic library of participant-specific behavioral signatures that evolve intraday.
5. Signal Propagation – Outputs probabilistic imbalance scores and directional bias to the broader agent swarm for decision making.

Integration with the Agentic Swarm

Key Advantages Over Traditional Tools

• No reliance on static volume profiles or historical back-testing for signal logic
• Continuous intraday retraining on live microstructure (data that cannot be recreated historically)
• Latency: < 8 ms from exchange receipt to swarm propagation (NYC/Chicago co-lo)
• Fully deterministic output for auditability while remaining adaptive

Use-Case Sketch (genericized)
In strongly trending conditions, IDENT™ frequently flags exhaustion when aggressive order flow at the offer begins to be absorbed with dwindling replenishment. This context allows the swarm to position counter-directionally or dynamically scale exposure ahead of visible price reversals.