股票解套分析 (Stock Break-Even Analyzer)

Overview

Analyze one or more A-share holdings to answer two questions: (1) what price is needed to
break even, and how much must it rise; (2) how likely is that, given recent technicals.
Deliver either a self-contained web app (best for beginners — open link, fill cost, get report)
or a Python script (best for users who want code).

When to use

• User mentions being "套牢 / 被套 / 回本 / 解套" or asks "还要涨多少才能回本".
• User provides holdings as name + cost + quantity and wants a break-even target.
• User asks for a reusable stock break-even / unstuck analysis tool.

Workflow

1. Collect holdings. For each stock get: 名称, 6位代码, 市场(sh/sz), 持仓成本价, 持仓数量.
   If the user only gave names, ask for cost and quantity (required for break-even math).
2. Choose deliverable by audience.
   - Non-technical user → build the web app (Step A). This is the default for beginners.
   - User explicitly wants code / a script → also provide the Python script (Step B).
3. Run / deploy, then present the report (target price, required gain %, probability, strategy).
4. Append disclaimer: "本报告仅供研究参考，不构成个人投资建议。"

Step A — Web app (preferred for beginners)

• Copy assets/index.html to a fresh directory (e.g. ./web_app/).
• Pre-fill the user's holdings into the DEFAULT_STOCKS array near the top of the <script>
  (fields: name, code, market, cost, qty, cur:""). Leave cur:"" so price auto-fetches.
• Deploy that directory with workbuddy_cloudstudio_deploy (action: deploy, entry: index.html).
  Present the returned 分享链接 to the user. Tell them they just open the link, optionally
  edit holdings in the table, and click "开始分析".
• Note: the page fetches data client-side via <script> tags (no backend, no CORS issues):
  realtime from 腾讯 qt.gtimg.cn, history K-line from 东方财富 JSONP. Works by double-clicking
  the local index.html too.

Step B — Python script

• Copy scripts/stock_break_even_analyzer.py to the user's workspace.
• Edit the STOCKS list at the top with the user's holdings.
• Run with the managed Python:
  C:\Users\Administrator\.workbuddy\binaries\python\versions\3.13.12\python.exe stock_break_even_analyzer.py
• The script prints a per-stock report to stdout.

Data sources & field mapping (do not re-derive)

• Realtime (web): 腾讯 https://qt.gtimg.cn/q=<market><code> returns
  v_sh600172="1~名称~代码~当前~昨收~今开~...". Field indices: 1=名称,
  3=当前价, 4=昨收, 5=今开, 33=最高, 34=最低, 30=时间(YYYYMMDDHHMMSS).
• Realtime (python): 新浪 https://hq.sinajs.cn/list=<market><code> returns 34 fields;
  date/time are at indices 30 and 31 (NOT the last two). Requires Referer: https://finance.sina.com.cn.
• History K-line (both): 东方财富 JSONP
  https://push2his.eastmoney.com/api/qt/stock/kline/get?secid=<1|0>.<code>&klt=101&fqt=1&lmt=120&end=20500101&fields1=f1,f2,f3&fields2=f51,f52,f53,f54,f55,f56&cb=<callback>.
  secid prefix: sh→1, sz→0. klines entries: 日期,开,收,高,低,量.

Probability model (transparent, 0–100)

Four dimensions, summed and clamped:
- Distance (0–50): target vs historical window high. a = (1 - need%/headroom%)*50
when target ≤ window high; else decays. headroom% = (histHigh - current)/current*100.
- MA / trend (0–25): +15 if price above all of MA5/20/60; +8 if above some; +0 if below all.
Plus clamp(MA20_slope%, 0, 10).
- Volume (0–15): clamp((vol5/vol20 - 0.8)*15, 0, 15).
- Volatility (0–10): clamp((histHigh-histLow)/histLow% / 5, 0, 10).
Already-profitable (current ≥ target) → 100. Strategy thresholds: ≥60 hold/wait breakout,
40–60 watch + stop-loss, <40 hard / consider stop or averaging-down (caution).

Pitfalls

• The agent's sandbox often blocks 东方财富 (push2his "Remote end closed"). This does NOT
  affect the deployed web page (runs in the user's real browser). If the Python script's K-line
  fetch fails in-sandbox, report it as [MISSING] and fall back to realtime-only; the user's
  browser/web version still works.
• Do not invent prices. If a fetch fails, mark it [MISSING]; never fabricate.
• 嘉林杰 is commonly a typo for 嘉麟杰 (002486.SZ) — confirm with the user if ambiguous.

Bundled resources

• assets/index.html — self-contained web app (realtime + K-line + full report, no backend).
• scripts/stock_break_even_analyzer.py — equivalent Python version with comments.