Anti-Gambling Trader skill
A skill and CLI that test a trade log with statistics, scan investment group chats for scam signs and scaffold a trading bot
- 1,162 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- Works with trade logs and screenshots that hold amounts, accounts and names, and when passed to a cloud agent the content goes to its provider
- `scaffold` generates code with templates for 13 real exchanges and brokers: after your changes and key entry it can send real orders and spend money
- The graphical interface accepts read-only broker API keys and can optionally store them in Windows Credential Manager
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/anti-gambling-traderDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Works with trade logs and screenshots that hold amounts, accounts and names, and when passed to a cloud agent the content goes to its provider
- `scaffold` generates code with templates for 13 real exchanges and brokers: after your changes and key entry it can send real orders and spend money
- The graphical interface accepts read-only broker API keys and can optionally store them in Windows Credential Manager
This is third-party code. Review the repository files before installing.
What it does
Anti-Gambling Trader judges whether your trades show a repeatable edge or just luck. You pass a trade log in CSV, JSON or Excel for stocks, futures, crypto or forex. `anti-gambling-trader analyze` computes expectancy, profit factor, drawdown and the Sharpe and Sortino ratios. It then tests significance with two tests, a t-test and a bootstrap, and checks the result on a separate part of the history. The verdict has five levels from gambling to statistical edge, and the tool does not soften a weak result. The second part works against scammers. `scan-text` finds common tricks in group and LINE chats, `guru-check` estimates how likely a guru's claimed numbers are to appear by chance, `forensics` looks for signs of faked returns and `survivorship` shows how win streaks arise among random participants. `risk-sim` models the share of scenarios where the account is wiped out. Reports save as HTML and as a card you can show family. The third part scaffolds a trading program with `scaffold`: strategy, risk control, a broker interface, a data source and a chart. A paper broker is the default. Templates exist for 13 real exchanges and brokers, but no orders reach a market until you fill in keys and lift the safeguards yourself. The graphical interface `anti-gambling-trader ui` opens in a browser on 127.0.0.1.
Who it is for. For private traders and investors who want to test their trades or a guru's promises with Claude Code or Codex.
Good fit when
- You keep a trade log and want to know whether it confirms an edge: `anti-gambling-trader analyze trades.csv`
- An investment group chat offers guaranteed profit: `scan-text` parses the messages and shows a danger level
- A channel author claims 90% winning trades over ten bets: `guru-check --win-rate 0.9 --trades 10` shows how easily that happens by chance
- You need to estimate the risk of wiping out an account at a real capital size: `risk-sim <file> --equity <amount>`
- You want a trading bot skeleton with a paper mode and safeguards before touching a real account
Not a fit when
- You want trading signals or a price forecast: the tool gives none and openly discourages trading when the statistics are poor
- The log has only a few trades: a small sample gives an insufficient data verdict
- You want a ready bot that trades a real account right away: the scaffold runs on a paper broker by default
- You expect a legal finding of fraud: the tool gives only a statistical estimate
Example request
Analyze my trades.csv with Anti-Gambling Trader and tell me whether my strategy has a statistical edgeLimitations
- The skill and documentation are written in Chinese, and part of the command output is Chinese and aimed at Taiwan: Taiwanese exchanges and brokers, the 165 hotline and LINE.
- Python 3.10 or newer is required. Excel reading, screenshot OCR and the desktop window install as separate extras.
- Screenshot analysis relies on text recognition and can confuse a sign, a decimal point or the trade direction, so check every field against the original.
- Real trading orders are not supported in the graphical interface. The exchange and broker connections in templates need your own keys and your own verification.
- The output does not replace investment advice.
How to disable. The skill sits in the repository folder `.claude/skills/anti-gambling-trader` (`.agents/skills/anti-gambling-trader` for Codex), so removing that folder from the project or from your skills folder is enough. `pip uninstall` with the package name shown in `pip list` removes the program itself. Saved reports and CSV files stay on disk and are deleted by hand.
FAQ
How do I check whether results are luck or strategy?
Prepare a log of closed trades in CSV or Excel and run `anti-gambling-trader analyze trades.csv`. The tool computes expectancy, tests significance with a t-test and a bootstrap, and checks the result on a separate part of the history. The outcome is one of five verdicts.
Can I check an investment guru's claims?
Yes. `guru-check --win-rate 0.9 --trades 10 --monthly-return 0.2` shows how often such numbers arise by chance. `forensics` checks monthly return data for signs of faking. The tool estimates probabilities and calls nobody a fraudster.
How do I install the skill in Claude Code?
Clone the repository, run `pip install -e .` in its folder and start Claude Code from there. The skill sits in `.claude/skills/anti-gambling-trader`, and Codex uses the copy in `.agents/skills/anti-gambling-trader`. There is no separate plugin install command.
Does the program send my trades over the internet?
The program analyzes the log locally and does not upload it. If you give files, screenshots or tool output to Claude Code or another cloud AI, the content may be transmitted under that provider's rules. The author advises hiding names, account numbers and LINE IDs first.
Can the tool trade a real account?
The graphical interface places no real orders and accepts read-only keys only. `scaffold` creates a project with a paper broker, and running on a real account is possible only after your own work: connecting the broker, adding keys and lifting the safeguards.
Discussion
No questions or reviews yet.
Related
SkillsA set of skills for individual investors and traders: market review, stock screening, technical analysis, trade planning, journaling and post-trade review
Skills and an MCP server for options traders: quotes, technical analysis, Greeks, spreads, scanners and Interactive Brokers portfolio
The Trade It plugin connects brokerage accounts to Claude Code, Codex and Cursor: the agent reads positions and prepares orders for review
Day Trader skill
The Day Trader skill
11 skills for day trading: chart scanner, news, entry and exit, risk checks and a history replay before any order
Claude plugin for reading investor filings, sizing portfolio positions and preparing orders for a separate brokerage connector
A skill that finds bottlenecks in an industry's supply chain and shortlists A-share, Hong Kong and US stocks with data checks
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Anti-Gambling Trader (this card) | no | 1,162 | |
| Claude Trading Skills | no | 3,656 | a skill set for market review, screening and trade planning; Anti-Gambling Trader tests completed trades and gurus' promises with statistics |
| Day Trader skill | no | 5,553 | 11 day-trading skills with a chart scanner and pre-order backtest; Anti-Gambling Trader evaluates a log and looks for scam signs |
README in short
The README starts with launching the graphical interface in a browser or desktop window and a statement that real orders are not supported. It then explains the tool's purpose, the commands for log analysis and anti-scam checks and the method of testing an edge with five verdicts. A separate section covers building trading programs and the four safety layers, and the end gives the project layout, test commands and a disclaimer.
README badge
Original SKILL.md
--- name: anti-gambling-trader description: >- 分析台股 / 台股ETF / 台指期選擇權 / 美股 / 加密貨幣 / 外匯的交易紀錄 (CSV / JSON / Excel),用統計學判斷使用者的獲利是「可重複的優勢」還是 「運氣 + 倖存者偏差(賭博)」,不適合長期投資會明確勸退。內建反詐工具: 掃描群組對話的詐騙話術(scan-text)、檢驗老師宣稱的績效(guru-check)、 鑑識假對帳單(forensics)、倖存者偏差示範(survivorship)。可模擬爆倉 風險(risk-sim)、偵測優勢衰退(trend)、輸出可傳給家人的 HTML 報告與 分享圖卡,並產生 14 種券商選項的交易程式腳架(預設紙上模擬)。 當使用者提到:分析我的交易、對帳單、我是不是在賭博、勝率盈虧比、 這老師可信嗎、這是詐騙嗎、我會不會賠光爆倉、我最近是不是退步了、 把報告傳給家人、建立我的交易程式、接券商 API 時,使用此技能。 --- # 反詐投資王 — Anti-Gambling Trader 立場:**不討好使用者,只說統計上的實話。** 多數人虧錢是把「運氣好」誤當 「有本事」;詐騙集團正是靠這個認知弱點收割。勸退結論不得軟化。 ## 意圖路由表(第一查找點) | 使用者會怎麼說 | 指令 | 轉述紀律(一句話) | |---|---|---| | 第一次用 / 我該從哪開始 | `start` | 依手上素材分流，不要求使用者先理解全部指令 | | 幫我快速判斷適不適合交易 | `fit-check <交易檔>` | 只分「停手/紙上/極小額驗證」；沒有完整紀錄最多只能紙上模擬 | | 我不會填表格 / 幫我記一筆 | `record [--out my_trades.csv]` | 一筆=一筆已平倉交易；策略填當時理由，不准事後美化 | | 幫我看對帳單 / 我這套賺不賺 | `analyze <檔案> [--market ...]` | 一句話裁決 → 期望值/雙 p 值/樣本外 → 警訊;🛡 反詐提醒必轉達 | | 我是不是在賭博 / 只是運氣好? | `analyze <檔案>` | luck_suspected =「無法排除是運氣」,不是「你沒本事」 | | 哪一招在送錢 / 跟單有沒有賺 | `analyze`(讀【各策略體檢】【跟單成績單】) | per-tag 只有描述統計,**不可**說某招「有優勢」 | | 一次看完整體檢 | `analyze <檔案> --full --equity <本金>` | 主報告 + 時間趨勢 + 風險情境,警語自動附上 | | 這老師可信嗎(勝率90%…) | `guru-check --win-rate 0.9 --trades N ...` | 機率語氣,絕不說「他一定是騙子」;沒 --trades 先要這個數字 | | 這訊息是詐騙嗎 / LINE 對話紀錄 | `scan-text "文字"` 或 `--file LINE對話.txt` | 序數等級,不轉百分比;「低」≠安全；轉述發言者/時間/原句/理由 | | 老師連續猜對 N 次好神 | `survivorship --traders 1000 --streak N` | 是對「現象」的數學解釋,非對特定人的指控 | | 這是他的月報酬,是不是假的 | `forensics 0.02,0.03,... [--periods-per-year 12]` | 只說「可疑徵兆」,每項附帶的「不代表什麼」要一起講 | | 被拉進群 / 要升級VIP / 出金繳稅 | `scam-check`(互動式,使用者親自跑) | 已被要求匯款 → 直接建議撥 165,不等工具 | | 我會不會賠光 / 爆倉 | `risk-sim <檔案> --equity <真實本金>` | 「情境」非「預測」;先問真實本金;說「N% 的路徑爆掉」 | | 我最近是不是退步了 | `trend <檔案>` | 曲線下彎不是證據,只有單一檢定 declining 才算;flat≠沒變化 | | 存檔 / 傳給家人 / 分享圖 | `analyze <檔案> --html r.html --card c.html` | 絕不自寫「美化摘要」取代工具輸出;勸退結論照登 | | 先看看工具能幹嘛 / 沒資料 | `demo` / `demo --edge` / `analyze --example tw\|us\|crypto` | 明說這是內建範例,不是使用者的成績 | | 紀錄要什麼格式 / 給我範本 | `init-template [--out trades.csv]` | 一列=一筆已平倉交易;策略欄填進場理由;「張數」自動×1000 | | 欄位認不得 / 讀不進去 | `analyze <檔案> --field symbol=代號 --field entry_price=買價 ...` | 先讓自動辨識試一次;.xlsx 需 openpyxl | | 存成 JSON / 程式化處理 | `analyze <檔案> --json out.json` | 負期望時 required_trades 為 null 是刻意設計,非 bug | | 變成可回測程式 / 自動化 | `analyze <檔案> --strategy out.py [--framework ...]` | 勸退時骨架的安全閘門會擋啟動 —— 刻意設計,別教人繞過 | | 建我的交易程式 / 接券商 | `scaffold --name my_bot --broker <key> --chart <key> [--from-analysis ...]` | 產出是待填框架,預設紙上模擬;建議帶 --from-analysis |