Alpha Playground
For agents
- Lets you: Validate, simulate, submit and propose alphas as pods.
- Tools:
alpha_validate,alpha_simulate,alpha_submit,alpha_list,alpha_get,alpha_propose_pod,data_fields - Limit: Plus and above.
- Limit: Simulate and submit use the plan's daily checks: Plus 30, Pro and Ultra 200.
- Limit: More than 8 versions of one shape in 60 minutes stops simulation for the day.
The loop
alpha_validate: free, unlimited. Fix errors by column.alpha_simulate: one daily check. Read the grade, report and submission checks.- Revise once with a reason. Every version is a trial.
alpha_submit: saves to your library when every check passes.alpha_propose_pod: one alpha, or 2 to 8 blended at equal risk. The card lands in your chat.- Press Start in the app.
Example
{"alpha": {"family": "mom", "expression": "rank(ts_zscore(ret(20), 60) / vol(60))",
"universe": {"name": "majors", "instruments": ["BTC", "ETH", "SOL", "HYPE", "XRP"]},
"settings": {"delay": 1, "neutralization": "universe"}}}
Reference pages
These pages are generated from the code's registries.
- Connect over MCP: tokens, the URL, the tools, checks a day.
- Expression language: text and JSON forms.
- Operators: every function and infix operator.
- Data sheet: every field, unit, when it is known.
- Settings: delay, decay, neutralization, rebalance, truncation, weights, limits.
- Output: the standard output of a simulation.
- Submission checks: what submit requires.
- Example alphas: one pattern each.
- Claude Code skill: the loop as a skill.
What a curve shows
- At most 250 points, starting at 1.
- None unless the book averages at least 5 effective names.