trading-learn
Skill1 starsChanged 8 days agoMITView on GitHub
The self-learning loop. Analyze the trade journal and out-of-sample trades by context (hour, weekday, volatility, trend, volume, signal age), propose filters, and test each proposal on unseen data before it can be used. Use when the user says "learn from the trades", "what's not working", "improve the strategy", "self-improve", or on a weekly schedule.
In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- llms.txtllms.txt
- trading-backtestSkill
- trading-deskSkill
- trading-go-liveSkill
- trading-paper-deskSkill
- trading-risk-rulesSkill
- trading-rotationSkill
- trading-strategy-labSkill
- trading-venuesSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Run python -m tradedesk learn, or narrow it with --strategies donchian_1h --symbols ETH.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Learning loop
Run python -m tradedesk learn, or narrow it with --strategies donchian_1h --symbols ETH. Results go to reports/learn-latest.json.
How it decides
- It groups trades by each context field.
- It flags a group when the group has 30 or more trades and averages -0.2R or worse, while the strategy as a whole is positive. If the strategy is losing overall, its worst pockets are shown as diagnostics only.
- Each flag becomes a proposal: the same strategy, skipping that group.
- Walk-forward re-runs each proposal. It is marked PROMOTE only if its average out-of-sample result improves by 0.05R or more and it keeps at least 70% of the trades.
- A promoted variant goes back onto the ladder at forward. Nothing edits live settings.
Explain to the user
Learning from a small sample mostly learns noise, which is why every idea must win on data it never saw. Report the rejected ideas too: they are the evidence that the loop is honest.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

