About Auroxyl

Systematic Trading, Fully Automated

Auroxyl Capital is a private quantitative trading platform that deploys machine-learning-driven strategies across U.S. equities and ETFs, executing autonomously via Interactive Brokers.

Platform

What Is Auroxyl

Auroxyl is a fully automated quantitative trading system that operates without manual intervention. From data ingestion and signal generation to order routing and portfolio reconciliation — every step is systematic.

The platform integrates directly with Interactive Brokers (IBKR) for live execution, managing the complete trade lifecycle: position sizing, order submission, partial fill handling, stop-loss management, and daily reconciliation.

Capital is allocated across two complementary strategies that adapt to market regimes in real time. A suite of circuit breakers, margin checks, and anomaly detectors ensure the system operates safely around the clock.

ML Signal Engine
Multi-seed gradient boosting ensemble, proprietary alpha factors
↓
Risk & Regime Layer
Bayesian regime detection, circuit breakers, margin guard
↓
Execution Engine
Smart routing, limit orders, partial fill management
↓
Interactive Brokers
Live execution, portfolio sync, exit order management
Strategies

Dual-Strategy Architecture

Capital is allocated 60/40 across two complementary strategies that target different sources of return.

60%
ETF Rotation

Regime-aware allocation across broad market and sector ETFs. A Bayesian risk model dynamically shifts between full equity exposure and capital preservation based on macro conditions. When the model detects bear regime, the portfolio moves to 100% cash.

Universe VOO / SMH / NLR
Regime Signal SPY MA175/MA200 + Bayesian risk
Bull Allocation 70% / 15% / 15%
Rebalance 21-day periodic + regime change
Vol Target 22% annualised
40%
ML Equity Prediction

Machine-learning ensemble ranks thousands of U.S. equities daily on proprietary alpha factors, selecting top-conviction names for 5-day overlapping cohort rotation. Walk-forward validated out-of-sample across all market regimes.

Model 15-seed LambdaRank max ensemble
Universe ~3,000+ U.S. equities
Holding Period 5 trading days
Selection Top 4 per cohort, 5 overlapping
Stop Loss 5% on final day (tail protection)
Founder

About the Founder

Taocheng Qian
Taocheng Qian
Founder & Developer

Taocheng designed and built the entire Auroxyl platform from the ground up — from the machine learning pipeline and alpha factor research to the C#/.NET execution engine and IBKR integration. Every component, from data ingestion to order routing, is his work.

His background in data science provides the quantitative foundation for the system's statistical approach: walk-forward validation, bootstrap significance testing, and rigorous overfitting diagnostics are core to the research process.

Bachelor of Science
Major in Data Science
University of Melbourne
Infrastructure

Technology Stack

Built with institutional-grade tools for reliability, speed, and transparency.

Python / LightGBM
ML training, prediction, alpha factors
C# / .NET 10 WPF
Execution engine, UI, scheduling
Interactive Brokers
Live execution, portfolio management
SQLite / WAL
Position tracking, audit trail, state
Polygon.io
Real-time market data, OHLCV
Serilog / Telegram
Structured logging, real-time alerts
Get in Touch

Collaboration Enquiries

This fund does not accept investors. Open to collaboration and partnership enquiries only.

erlian3230@gmail.com