xquant-core 0.1.0

Institutional-grade quantitative trading framework for alpha research and production.

XQuant

Institutional-Grade Quantitative Trading Framework for Alpha Research and Production

XQuant is a high-performance, modular framework designed for the development, backtesting, and deployment of quantitative trading strategies. Built with an emphasis on mathematical rigor and architectural flexibility, XQuant provides researchers and engineers with the tools necessary to bridge the gap between abstract alpha discovery and systematic production execution.

---

Technical Foundations

Core Architecture

XQuant utilizes a strictly decoupled, layered architecture to ensure modularity and extensibility:

• Data Orchestration: Standardized OHLCV ingestion with support for Equities, Digital Assets, and FX via optimized adapters (yfinance, CCXT, OANDA). Features include multi-asset alignment, timezone normalization, and adaptive trading calendars.
• Feature Engineering: A comprehensive library of 50+ technical indicators and microstructure metrics (OFI, VWAP Deviation, Bid-Ask Spread). Supports custom factor development via an abstract Factor API.
• Signal Discovery: Institutional-grade signal evaluation using Information Coefficient (IC), ICIR, and Rank IC. Includes advanced signal decay and half-life analysis.
• Dual-Engine Backtesting:
  • Vectorized: Ultra-fast NumPy/Pandas engine for high-throughput parameter optimization.
  • Event-Driven: High-fidelity simulation utilizing an asynchronous event loop for realistic execution modeling, slippage, and market impact.
• Statistical Validation: Robustness suite featuring Walk-Forward Optimization (WFO), Monte Carlo simulations, and Overfit Detection using the Deflated Sharpe Ratio (DSR).
• Risk & Portfolio Management: Advanced metrics suite (Sharpe, Sortino, VaR, CVaR) coupled with volatility-targeted position sizing and Kelly Criterion allocation.
• Live Execution Infrastructure: Low-latency broker adapters with support for paper trading and live production hooks (CCXT).

---

System Components

Layer	Functional Scope
Ingestion	Multi-source adapters, OHLCV normalization, resampling.
Features	Technical indicators, microstructure, cross-asset correlations, regime detection.
Signals	Alpha discovery, scoring, ensemble combination, decay analysis.
Backtest	Vectorized research engine, Event-driven execution engine.
Validation	OOS testing, WFO, Monte Carlo, Bias/Overfit detection.
Risk	Institutional performance metrics, exposure tracking, drawdown analysis.
Execution	Paper trading simulation, Live broker adapters.
Analytics	Performance tearsheets, high-fidelity visualization, data export.

---

Installation and Deployment

Prerequisites

• Python 3.9+
• NumPy, Pandas, Polars
• CCXT (for digital asset execution)

Setup

git clone https://github.com/ak495867/xquant.git
cd xquant
pip install -e .

---

Usage Overview

Automated Strategy Execution

XQuant provides comprehensive examples to demonstrate full-pipeline execution:

# Execute Equity Momentum Strategy
python examples/equity_momentum.py

# Execute Crypto Mean Reversion Strategy
python examples/crypto_mean_reversion.py

Strategic Research Workflow

Researchers can utilize the modular API to build custom pipelines:

from xquant.data.loader import DataLoader
from xquant.features.technical import RSI
from xquant.backtest.engine.vectorized import VectorizedEngine

# Ingestion
data = DataLoader().load("AAPL", start_date="2022-01-01")

# Engineering
data['rsi'] = RSI(window=14).compute(data)

# Signal and Backtest
signals = (data['rsi'] < 30).astype(int)
results = VectorizedEngine().run(data, signals)

---

Development and Contributions

XQuant is built for the community. We maintain rigorous standards for code quality and mathematical accuracy. Please refer to CONTRIBUTING.md for technical specifications on adding new features or adapters.

License

XQuant is distributed under the MIT License.