Skip to content

README.md

Latest commit

 

History

History
210 lines (157 loc) · 6.19 KB

README.md

File metadata and controls

210 lines (157 loc) · 6.19 KB

QuantPlatformKit

Shared broker adapters, domain models, execution ports, and notification utilities for QuantStrategyLab strategies.

English | 中文 | 中文详版

English

Scope

This repository is the shared platform layer for QuantStrategyLab strategy services.

It is intended to contain:

  • common domain models
  • narrow ports for market data, portfolio snapshots, execution, notifications, and state
  • broker-specific adapters
  • small reusable notification utilities

It is not intended to contain:

  • strategy rules
  • target allocation logic
  • Cloud Run entrypoints
  • scheduler or workflow orchestration specific to one strategy

Strategy contract boundary

The current mainline split is:

  • platform repositories assemble StrategyContext
  • platform repositories load a strategy entrypoint through load_strategy_entrypoint(...)
  • strategy repositories return a unified StrategyDecision
  • platform-local decision mappers turn that decision into broker orders, notifications, and runtime state updates

Strategy repositories should expose manifest + evaluate(ctx) and keep any migration-window runtime metadata behind StrategyRuntimeAdapter. Broker-specific order sequencing and UI layout should stay out of strategy outputs.

Migration details and follow-up guidance live in docs/strategy_contract_migration.md.

For the stricter end-state rules for new US equity strategies, see:

Package layout

src/quant_platform_kit/
  common/
    models.py
    ports.py
    strategies.py
  ibkr/
    connection.py
    market_data.py
    portfolio.py
    execution.py
  binance/
    client.py
    account.py
    market_data.py
    execution.py
  schwab/
    auth.py
    market_data.py
    portfolio.py
    execution.py
  longbridge/
    auth.py
    market_data.py
    portfolio.py
    execution.py
  notifications/
    telegram.py
tests/

Development

Run tests with:

PYTHONPATH=src python3 -m unittest discover -s tests

Release and deployment model

QuantPlatformKit is a shared dependency, not a runtime service. Strategy repos should pin a fixed Git tag such as:

quant-platform-kit @ git+https://github.com/QuantStrategyLab/QuantPlatformKit.git@v0.7.1

Cloud Run and self-hosted runner deployments should continue to deploy the strategy repositories only. See docs/deployment_model.md for:

  • service naming suggestions
  • fixed-tag dependency rules
  • Google Cloud trigger rebind steps after repo rename
  • HK / SG multi-service guidance for LongBridgePlatform
  • US equity cross-platform strategy contract rules

中文

QuantPlatformKit 是 QuantStrategyLab 的共享平台层仓库。

它负责放这些内容：

  • 统一领域模型
  • 市场数据、持仓、执行、通知相关的公共接口
  • IBKR / Schwab / LongBridge / Binance 的平台适配层
  • 少量可复用的通知和运行时工具

它不负责放这些内容：

  • 具体策略规则
  • 目标仓位和调仓计算
  • Cloud Run 或 VPS 入口
  • 某一个平台仓库自己的调度和部署编排

范围

这个仓库是各平台仓库共享的公共依赖。

策略契约边界

当前主线边界已经固定为：

  • 平台仓库负责组装 StrategyContext
  • 平台仓库通过 load_strategy_entrypoint(...) 加载策略入口
  • 策略仓库只返回统一的 StrategyDecision
  • 平台自己的 decision mapper 再把决策映射成券商订单、通知和运行时状态更新

策略仓库应该暴露 manifest + evaluate(ctx)；如果迁移窗口里还需要少量运行时元数据，就放在 StrategyRuntimeAdapter 里，不要把券商专属下单顺序或展示布局塞回策略输出。

迁移说明和后续约束见 docs/strategy_contract_migration.md。

以后新增美股策略要遵守的更严格跨平台规范见：

目录结构

src/quant_platform_kit/
  common/
    models.py
    ports.py
    strategies.py
  ibkr/
    connection.py
    market_data.py
    portfolio.py
    execution.py
  binance/
    client.py
    account.py
    market_data.py
    execution.py
  schwab/
    auth.py
    market_data.py
    portfolio.py
    execution.py
  longbridge/
    auth.py
    market_data.py
    portfolio.py
    execution.py
  notifications/
    telegram.py
tests/

开发

运行测试：

PYTHONPATH=src python3 -m unittest discover -s tests

发布和部署

QuantPlatformKit 只作为共享依赖，不单独部署。策略仓库应该固定依赖某个 Git tag，例如：

quant-platform-kit @ git+https://github.com/QuantStrategyLab/QuantPlatformKit.git@v0.7.1

部署说明见：