AI column

Data@donghyeok99Staff

Claude Trading Skills: 74 skills that check risk before a trade

Give it two market-breadth scores and this Claude skill set returns a one-page equity exposure ceiling

Breadth and uptrend scores of 60 draw in, a 36% ceiling settles on the exposure column, then REDUCE_ONLY is stamped. Fictional README numbers.
Breadth and uptrend scores of 60 draw in, a 36% ceiling settles on the exposure column, then REDUCE_ONLY is stamped. Fictional README numbers.
Who
Built by individual investor tradermonty to improve his own trading process, then opened for time-constrained individual investors.
When
The repository was created on October 19, 2025 and had commits on November 18, 2026. It has no GitHub Release yet.
Where
Runs in the Claude web app and Claude Code. The scripts support Python 3.11 to 3.13 on Linux, Windows and macOS.

What was hard

After work it is easy to go straight to picking stocks and skip two questions: is today a day to add risk at all, and how much can one trade lose. The trade journal lasts a few days, and why last month lost money stays in memory only.

What it does about it

This repository turns each skipped step into a skill and puts them in order. Market breadth and uptrend participation are scored first, that score sets an equity exposure ceiling, and only then come candidates and position size. The key idea is that the AI never buys or sells. Every workflow has a decision step a person answers, and the result is a document to review, not an order.

tradermonty/claude-trading-skills

What people made with it

GitHub@benli122026-05-04

Tried to journal three real trades from an Israeli broker in trader-memory-core and wrote up where fractional shares and partial exits did not fit. Today's schema has both.

I tried to log three real trades from IBI Smart (Israeli broker, fractional USD shares) into
likes 0 · comments 4 (2026-11-19 as of)GitHub ↗

How it runs inside

  1. 1

    You → Claude Code

    Ask for today's market checkThe market-regime-daily workflow names which skills run and in what order.workflows/market-regime-daily.yaml
  2. 2

    Claude Code → market-breadth-analyzer

    Run market_breadth_analyzer.pyIt fetches public CSVs, scores six components and writes a 0-100 composite as JSON and Markdown.skills/market-breadth-analyzer/scripts/market_breadth_analyzer.py
  3. 3

    Claude Code → uptrend-analyzer

    Run uptrend_analyzer.pyIt diagnoses how many stocks take part in the uptrend from public CSV data and writes a report.skills/uptrend-analyzer/scripts/uptrend_analyzer.py
  4. 4

    Claude Code → exposure-coach

    Pass the --breadth and --uptrend JSONBoth reports go to calculate_exposure.py. Missing inputs lower confidence instead of stopping the run.skills/exposure-coach/scripts/calculate_exposure.py
  5. 5

    exposure-coach → Market Posture Summary

    exposure_posture_*.json and .mdOne page with the exposure ceiling, growth or value bias, participation, recommendation and confidence.skills/exposure-coach/SKILL.md
  6. 6

    Market Posture Summary → You

    allow / restrict / cash-priorityThe last step is a decision gate a person answers. The output is a posture, not a trade signal.workflows/market-regime-daily.yaml

What goes in, what comes out

breadth 60, uptrend 60 (fictional README data)36% exposure ceiling, REDUCE_ONLY, LOW confidence
Account 100,000, entry 155, stop 148.50, risk 1%153 shares, 994.50 at risk (the position-sizer document's example)
Four fictional closed tradesA weekly digest with a 50% win rate and 75 dollars realized profit

Is it for you

Good for

  • Individual investors with long-term, ETF and dividend holdings at the core and some swing trading on the side
  • People who want the market state and a risk limit settled before looking for stocks
  • People who want to keep a trade journal and a monthly review going
  • People who want to learn the market-check routine before paying for data

Not for

  • It places no orders and gives no signals. It is not for automated trading or scalping, and it is educational, not financial advice.
  • Many screeners need an FMP key (free tier: 250 requests a day). FINVIZ Elite costs $39.50 a month.
  • The Claude.ai Free plan does not include Claude Code, per the README.
  • The preview numbers in the README are fictional data, not live market readings.

Reviewed by Devalley’s editor before publishing. How the column is made

Sign in to post, comment and like.Sign in

Report

Reason

Read the community rules