Good output example
A good output says the agent skipped because invalidation was missing, the setup was late, and adding risk would break the paper exposure limit. It creates a useful journal entry even though no simulated trade happened.
Template
AI trading agent output format template
A paper agent is easier to review when every entry, exit, skip, and missed-trade note uses the same output fields. Use this template to turn agent responses into audit-ready journal evidence.
Output format is not an order format
Trading Boy does not execute live trades, hold funds, or provide financial advice. This output format is for simulated paper-trading review. It is not a live order schema, a signal feed, or an instruction to place trades.
Required output fields
Use the same fields across entries, skips, exits, and missed trades. Consistent output makes the review process stronger because missing evidence becomes visible.
| Field | Entry output | Skip or exit output | Why it matters |
|---|---|---|---|
| Decision type | Paper entry, paper add, paper exit, or no action. | Skip, missed setup, excluded, or close review. | Separates action from review status. |
| Rule version | Prompt, persona, setup, and risk version. | Same version labels. | Prevents hidden prompt drift. |
| Setup name | The written setup that qualified. | The setup that failed or was incomplete. | Connects output to the rules page. |
| Thesis and invalidation | Why the paper idea exists and what proves it wrong. | Why the thesis is absent, invalid, or closed. | Allows post-trade review without guessing. |
| Paper risk | Planned simulated size, stop logic, and drawdown context. | Risk reason for skip or exit. | Stops confidence from overriding the risk rule. |
| Review question | One question for the future reviewer. | One question explaining the skip or exit. | Turns output into journal evidence. |
Template text
Decision type: [paper entry | paper exit | skip | missed setup | excluded]
Rule version: [prompt version, persona version, setup version, risk rule version]
Setup name: [the exact setup or blocked setup from the written rule]
Thesis: [why the simulated decision is being considered, written before outcome]
Invalidation: [condition, price, time, or context that proves the paper idea wrong]
Paper risk: [simulated size, maximum planned loss, drawdown context, exposure conflict]
Skip or exit reason: [required when decision type is not paper entry]
Journal tags: [rule fit, behavior tag, data quality tag, market regime]
Review question: [one specific question for the later checklist]
Weak output example
A weak output says the setup looks strong or confidence is high without naming the setup rule, invalidation, risk limit, or review question. That output may sound helpful, but it is hard to audit.
How to apply the format
Start with the AI trading agent rules workflow, then put this output format into the AI trading agent prompt template. The prompt should tell the agent that incomplete fields require a skip or excluded status rather than a confident paper entry.
Use the same format during the simulated AI trading agent review process. If the format is missing fields, do not patch the record after the fact unless the missing information is clearly available. Mark the sample caveated instead.
When the output format changes, record the change in AI trading agent prompt versioning. Format changes can affect results because they change what the agent is forced to consider before acting.
Handling missing or uncertain fields
The template should make uncertainty explicit. If the agent cannot identify the setup, invalidation, market context, or paper risk, the output should become a skip or excluded record. Do not let the agent fill missing fields with confident language after the fact. That creates a clean-looking journal, but it weakens the review because the original decision was not actually complete.
Use short allowed values where possible: entry, exit, skip, missed setup, excluded, clean rule fit, broken rule fit, data caveat, and risk caveat. Allowed values make weekly review faster and reduce the chance that similar problems are split across many labels. Free-form notes are still useful, but they should explain the label rather than replace the structured fields.
AI trading agent output format FAQ
What should an AI trading agent output format include?
It should include decision type, setup name, thesis, invalidation, paper risk, skip reason, market context, journal tags, confidence caveat, and review question.
Why require an output format for a paper agent?
A required output format prevents vague agent summaries and makes entries, exits, skips, and mistakes easier to review consistently.
Can this output format be used for live trade orders?
No. This template is for simulated paper-trading review. It is not a live order schema, signal feed, or financial advice.