Platform overview

Language: English · 简体中文

quant-platform provides reusable quantitative research and portfolio capabilities. It connects research signals to portfolio results through portfolio construction, backtesting, risk analysis, costs, execution simulation, and public artifact contracts.

What happens in a backtest

flowchart LR
    Data[Published data assets] --> Signal[Signals and scores]
    Signal --> Portfolio[Portfolio construction]
    Portfolio --> Backtest[Backtest and ledger]
    Backtest --> Results[Positions, returns, and research artifacts]
    Results --> Review[Review and handoff]

The minimal call path is:

DataFrame
  → StrategySpec
  → ExecutionModel
  → BacktestSpec
  → run_backtest
  → stats / returns / periods

Project responsibilities

Project	Responsibility
`quant-market-data-platform`	Data ingestion, normalization, quality checks, versioning, and publication
`quant-research`	Strategies, features, models, experiments, and research conclusions
`quant-platform`	Backtesting, portfolio construction, risk, costs, execution simulation, and public contracts
`quant-intel-platform`	Reports, dashboards, and research-result delivery
`quant-intel-deploy`	Research-result publication and deployment

Research projects collaborate with this repository through published, versioned data assets and artifacts. The platform remains strategy-agnostic and does not store real strategy inputs, credentials, or proprietary selection logic.

Three core objects

`StrategySpec`

Describes how securities are selected from scores and how target weights are assigned. Examples include top_k, weighting, holding buffers, and group limits.

`ExecutionModel`

Describes entry and exit, costs, slippage, trading calendars, and tradability constraints.

`BacktestSpec`

Combines the strategy, execution model, rebalance dates, holding period, and annualization convention into a serializable configuration.

Out of scope

Data-provider integration and credential management
Private strategies and proprietary features
Model training and research conclusions
Broker order placement and production trading runtime

Callers or other projects own these responsibilities. The platform provides reusable mechanisms and public interfaces.