Skip to main content

QuantFlow Trading

Real-time execution path for live feature computation

DataInfra → FeatureDAG → Trading (Real-Time Execution)

Trading is the real-time execution layer of QuantFlow.

It runs FeatureDAG pipelines on live market data to compute features and generate trading signals.

The same feature definitions used in Research are executed in real time without modification.

Real-time feature computation
Deterministic execution (same as batch)
Stateful processing (order book, rolling windows)
Signal generation and aggregation
Monitoring and observability

DolphinDB — native streaming engine
WebSocket — real-time data ingestion
Delta-based Order Book — efficient state reconstruction
FeatureDAG IR — shared computation graph
Consolidated Deployment — optimized engine grouping

Ingest — Connect to WebSocket market data feeds
Order Book — Reconstruct full depth state
State Engine — Maintain rolling windows and triggers
Feature Compute — Execute FeatureDAG in real time
Feature Output — Consolidate features into output stream tables
Signal — Generate trading outputs
Monitoring — Track latency and execution health

Handles high-frequency streaming data
Processes real-time market events efficiently
Supports low-latency execution

Trading operationalizes what Research validates.

← Back to Home