Gahow Wang
722cade974
Share point-in-time signals and a cash-funded order ledger across backtests and trading. Correct eligibility, missing-price fills, quality factors, regime inputs, warmup windows, and cash-injection accounting. Add reproducible registered-strategy comparisons and paired forward candidates with a 20% resize band. Validation: 125 Python tests, 23 web tests, TypeScript/Vite build, 87 strategy causality checks, and real-data target/order/equity parity for 10 strategies.
quant-trading web
Web UI and API for the quant-trading repository: strategy backtests and live books (via ../bridge.py), plus a personal watchlist agent.
统一个人股票监控 MVP，支持 A股、港股、美股自选股，按日生成监控报告。数据接口固定使用 IPADS Responses API，密钥从仓库根目录 .env 的 IPADS_API_KEY 读取。
Run
npm install
npm run dev
Frontend: http://127.0.0.1:5173
Backend: http://127.0.0.1:8787
LAN access:
HOST=0.0.0.0 CLIENT_PORT=5174 PORT=8787 npm run dev
Open http://gahow-pc.ipads-lab.se.sjtu.edu.cn:5174 from another device on the same network.
IPADS API
配置文件：
IPADS_API_KEY=your_api_key
固定服务端配置：
base_url = "http://tianx.ipads-lab.se.sjtu.edu.cn:8319/v1"
wire_api = "responses"
env_key = "IPADS_API_KEY"
The backend calls:
POST http://tianx.ipads-lab.se.sjtu.edu.cn:8319/v1/responses
Authorization: Bearer ${IPADS_API_KEY}
The response output text must be JSON matching the stock snapshot schema used by the agent. If IPADS_API_KEY is blank, the app uses deterministic demo data so the dashboard remains usable.
Features
- Add/remove watchlist symbols such as
600519,000001.SZ,00700.HK,AAPL,MSFT.US. - Normalize symbols into A股、港股、美股 markets.
- Generate a daily report with earnings timing, peer stocks, valuation discount/premium, related news, and buy-watch/hold/avoid recommendations.
- Refreshes the cached daily report every 24 hours while the backend process is running.
- View each watchlist stock with locally cached D/W/M/Y K-line chart data, zoom controls, MA5/MA20/MA60, RSI14, MACD, volume, and cached financial statements. US K-line and quote/research fields use Nasdaq as the fixed data source and target at least 5 years of verified daily OHLCV history when available.
- Cache related market context such as sector, liquidity, volatility, drawdown, relative strength, capital flow, sentiment, and risk flags.
- Run an agent-backed quantitative analysis across technical, valuation, growth, quality, liquidity, sentiment, and risk factors to classify each stock as buy-watch/hold/avoid.
- Add tags to every watchlist stock and filter the watchlist by tag.
- In configured provider mode, prompts require verified data and do not permit approximate or fabricated values; missing data is surfaced instead of being filled with synthetic values.
- K-line cache updates incrementally: once 5 years of daily bars exist locally, refreshes request only the latest missing window and merge by date.
Local cache
Web state lives in the repo-level data/web/ (see src/server/paths.ts): watchlist.json, per-stock snapshot/K-line/financial data under cache/, and backtest results under strategy_cache/. The strategy page reads data/precomputed/backtest_results.json, written by precompute_backtest.py at the repo root.
Production runs as systemd services; see "Server Deployment" in the root CLAUDE.md.