Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
npx claudepluginhub zxx264547/worldquant-brain-research --plugin alpha-idea-generatorAI主导的量化Alpha研究系统 — 首个已提交Alpha:EUR Sharpe 1.84
首个 Alpha 成功提交: omnKPLX5 (A18)
zscore(-ts_max(vec_max(mean_loan_rate), 66)) * (group_rank(ts_mean(volume, 20), sector) > 0.03)EUR 策略验证: mean_loan_rate w66 在 EUR 达到 1.81-1.84,高于 USA 同公式。EUR 新区竞争少,prod_corr 更容易通过。
# 1. 创建 Python 环境
python3.12 -m venv wq_env && source wq_env/bin/activate
pip install requests pandas
# 2. 配置 BRAIN 账号
cp worldquant_brain/config/user_config.json.example worldquant_brain/config/user_config.json
# 编辑填入邮箱密码
# 3. 验证
wq_env/bin/python -c "
from worldquant_brain.scripts.core.api_client import RetryableBrainClient
import asyncio
asyncio.run(RetryableBrainClient().ensure_authenticated())
print('OK')
"
# Python环境
/home/zxx/wq_env/bin/python
# Alpha提交(必须用curl -L跟踪303)
python worldquant_brain/scripts/submit_alpha.py <alpha_id>
# 离线数据分析
python worldquant_brain/scripts/brain_data_scope.py star <dataset>
python worldquant_brain/scripts/brain_data_scope.py badge <dataset>
# 信号灯系统
python worldquant_brain/scripts/direction_radar.py <results.json>
# 论文驱动Alpha生成
python worldquant_brain/scripts/seed_alpha_generator.py
PERCEIVE → PLAN → DISPATCH → REFLECT → REMEMBER → EVOLVE
| 阶段 | 工具 | 作用 |
|---|---|---|
| PERCEIVE | brain_data_scope.py | 离线查历史数据 |
| PERCEIVE | CURRENT_STATE.md | 当前进展 |
| DECIDE | direction_radar.py | 信号灯判断方向 |
| EXECUTE | backtest_runner.py | 批量回测 |
| SUBMIT | submit_alpha.py | curl -L 提交 |
| REMEMBER | ingest-all | 导入结果到数据库 |
| 工具 | 来源 | 作用 |
|---|---|---|
brain_data_scope.py | 帖子 #40091088861591 | 离线 SQLite 数据分析 |
direction_radar.py | 帖子 #2009 | 信号灯系统 + 算子多样性 |
seed_alpha_generator.py | 帖子 #39870945020183 | arXiv 论文驱动 Alpha 生成 |
submit_alpha.py | 项目自研 | curl -L 正确提交 Alpha |
| 数据集 | 区域 | 最佳 Sharpe | 状态 |
|---|---|---|---|
| shortinterest3 | USA | 2.45 | ❌ prod_corr 0.95 |
| shortinterest3 | EUR | 1.84 | ✅ 已提交 |
| shortinterest3 | ASI | 1.20 | 未达标 |
| risk60 | USA | 2.37 | ❌ prod_corr > 0.9 |
| analyst4 | USA | 1.17 | 未达标 |
| 15+ 其他数据集 | USA | < 1.0 | 信号太弱 |
| 编号 | Alpha ID | 区域 | Sharpe | 状态 |
|---|---|---|---|---|
| A18 | omnKPLX5 | EUR | 1.84 | ACTIVE/OS |
| A19 | 0mAEmWL1 | EUR | 1.84 | ❌ SELF_CORR |
| A20 | kqn3AA6O | EUR | 1.64 | ❌ SELF_CORR |
| A21 | YPNpRl2w | EUR | 1.62 | ❌ SELF_CORR |
worldquant_brain/
├── engine/ # 核心引擎
│ ├── backtest_runner.py # 回测执行器
│ └── settings_manager.py # 设置管理器
├── scripts/
│ ├── core/api_client.py # API客户端
│ ├── submit_alpha.py # Alpha提交 (curl -L)
│ ├── brain_data_scope.py # 离线数据分析
│ ├── direction_radar.py # 信号灯系统
│ └── seed_alpha_generator.py # 论文驱动生成
├── knowledge_base/
│ ├── memory/ # AI记忆
│ └── skills/ # 技能文档
│ ├── onboarding.md # 首次启动指南
│ ├── alpha-submit.md # 提交方法
│ └── skills-index.md # 技能索引
├── data/
│ ├── field_analysis.db # 离线分析数据库
│ └── forum.sqlite3 # 论坛知识库 (2173篇)
└── config/
└── user_config.json # BRAIN账号 (不提交Git)
| 指标 | 要求 |
|---|---|
| Sharpe | >= 1.58 |
| Fitness | > 0.5 |
| PPC | < 0.5 |
| Margin | > Turnover |
| Turnover | > 0.01 且 < 0.7 |
| ProdCorr | < 0.7 |
| Weight Conc. | < 0.1 |
项目已集成三篇核心论坛帖子:
最后更新:2026-05-16
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Comprehensive UI/UX design plugin for mobile (iOS, Android, React Native) and web applications with design systems, accessibility, and modern patterns
Agent that simplifies and refines code for clarity, consistency, and maintainability while preserving functionality
AI image generation Creative Director powered by Google Gemini Nano Banana models. Claude interprets intent, selects domain expertise, constructs optimized prompts, and orchestrates Gemini for best results.
Qiushi Skill: methodology skills for AI agents guided by seeking truth from facts, with Claude Code, Cursor, OpenClaw, Codex, OpenCode, and Hermes guidance.
深度优化 — 将Sharpe>1.0的Alpha推向>=1.58的提交标准
AI研究主管 — 认知循环驱动、研究协调、决策下发
Alpha探索Worker — 系统测试想法,按复杂度递增执行回测
研究助理 — 搜索论坛和邮件,提炼知识到知识库
Alpha生成/回测/优化 — 表达式构建、批量回测、信号灯系统、算子纠错
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claim