All features
Microstructure Model

Price Prediction Model

Real-time LONG / SHORT signals derived from market microstructure data — tick by tick.

Helius Terminal's price prediction model computes intraday directional signals from live Level 1 quote data — without machine learning or historical training sets. The model is deterministic, fully transparent, and runs on every incoming tick from the Schwab streamer. Every signal is logged, evaluated 30 seconds later, and fed into a live accuracy tracker so you always know how the model is performing.

Key Capabilities

What this feature does for you

Real-Time LONG / SHORT / FLAT Signals

Every symbol in your watchlist receives a directional signal on each incoming tick — LONG, SHORT, or FLAT — computed from five market microstructure inputs.

Composite Weighted Score

Five signals are weighted into a composite: Order Book Imbalance (40%), Order Flow Imbalance (35%), 30-tick Momentum (15%), and VWAP Deviation (10%, mean-reverting bias).

Live Accuracy Dashboard

Track overall accuracy, confidence-stratified performance, Spearman Information Coefficient (IC), mean return, and cumulative P&L curve across all historical signals.

Per-Symbol Breakdown

Drill into accuracy, mean return, win rate, and long/short signal split for any individual symbol in your watchlist.

Confidence-Stratified Buckets

Signals are grouped by confidence level. Higher-confidence signals are expected to outperform — the accuracy dashboard validates this relationship in real time.

Realized Return Logging

Every emitted signal is logged to Firestore. 30 seconds after emission, the realized return is written back to the same record — giving you a full, auditable signal history.

Related Features

Ready to try it?

Sign up free and connect your brokerage account to get started in minutes.

Create Free Account