PROJECT 01 · QUANTITATIVE RESEARCH
Quantitative Research Platform
An institutional-grade, in-house quantitative research and trading platform spanning data, research, trading, operations and platform.
01 · Overview
Overview
Quantitative Research Platform is Xuanheng’s in-house quantitative research and trading platform built to institutional standards. It covers the full stack of data governance, factor research, portfolio construction, backtesting and simulation, trade execution and risk control, with investment operations and platform engineering supporting daily research.
Research proceeds on point-in-time (PIT) data: from a research question through the investable universe and factor definitions, into factor research and mining and factor testing and screening, then to portfolio construction and backtest validation, and finally to paper and live execution under risk controls. Data quality, research reproducibility and human oversight come first throughout — a research conclusion is never automatically a trading decision.
The platform separates system responsibility from delivered progress: data governance, the investable universe, operators and formal factors, backtesting and the execution system are in place; the formula representation has completed its first phase, while factor mining, the risk model and strategy assembly are being advanced along the established architecture, and live trading is admitted stage by stage.
02 · Capabilities
Capabilities
- 01
Data Governance
Data contracts, multi-source ingestion and publication checks, and read-only access with point-in-time (PIT) data assembly.
- 02
Investable Universe
The reviewed ETF universe with point-in-time membership and eligibility, with human review kept in the loop.
- 03
Factor Research & Mining
Typed operators, a formal factor registry with immutable baselines, formula representation and search-based mining.
- 04
Factor Testing & Portfolio
Factor evaluation and screening, weight optimization and portfolio construction, backtesting, simulation and performance analysis.
- 05
Execution & Operations
Paper and live execution runtime covering orders, fills and reconciliation, with live trading under staged admission; investment operations track holdings, benchmarks and performance.
03 · Research Lifecycle
Research Lifecycle
- 01
Research question
- 02
Data & investable universe
- 03
Factor research & mining
- 04
Factor testing & screening
- 05
Portfolio construction & backtesting
- 06
Paper/live execution & risk control
04 · System View
System View
- 01
Data Platform Layer
- 02
Research & Modeling Layer
- 03
Portfolio & Trading Layer
- 04
Platform & Engineering Layer
05 · Current State
Current State
Implemented
Data governance (contracts, ingestion, publication, PIT access), the investable universe, operators and the formal factor registry, the backtesting framework, the execution runtime with investment operations, and orchestration with the business console are in place. The first phase of the formula representation — text parsing, an intermediate representation and registry linkage — is merged as formal code.
Current Focus
Building out formula compilation and execution, and factor search mining along the Factor Miner research protocol, completing the staged admission of live trading, and extending PIT data coverage and backtest consistency.
Next Direction
The risk model, strategy assembly and automated optimization workflows will be delivered as research validates them; the platform does not claim a capability before it is proven.
06 · Principles
Principles
- 01
Data first: provenance, time alignment and visibility are verified before any factor performance.
- 02
Research must be reproducible: definitions, runs and baselines stay registered and traceable.
- 03
Research, simulation and live trading advance in stages; trading decisions remain human.
- 04
Platform responsibility is distinct from delivery status: nothing is presented as complete before it is proven.