Use this checklist before you change a paper-trading rule, prompt, risk setting, or benchmark. It keeps the review focused on evidence quality instead of recent simulated profit or loss.
Trading Boy does not execute live trades, hold funds, or provide financial advice. This checklist validates simulated paper-trading evidence only. It does not approve live trading decisions or predict future returns.
Copy this checklist into a journal, agent review prompt, or team note. Fill it out after the paper sample closes and before any rule change is made.
| Checklist item | Pass condition | Risk if skipped | Next action |
|---|---|---|---|
| Sample window | Start date, end date, market, timeframe, and rule version are recorded. | Old and new results get mixed together. | Split the sample or restart the benchmark. |
| Trade inventory | Entries, exits, skips, missed trades, and exclusions are counted. | The sample overweights active trades and hides restraint. | Add missing records before judging the result. |
| Rule fit | Each paper entry maps to a written setup and invalidation. | Outcome bias makes weak decisions look planned. | Tag rule breaks and review them separately. |
| Risk behavior | Paper size, max drawdown, exposure, and stop distance stayed inside plan. | Positive paper PnL hides unsafe process drift. | Reduce simulated size or tighten controls. |
| Outlier review | The largest winner and loser are isolated and explained. | One unusual result controls the conclusion. | Report normalized and raw views. |
| Market context | Trend, volatility, news, liquidity, and broad regime are labeled. | The rule looks stronger than it is in one favorable regime. | Collect more context or segment the benchmark. |
| Journal completeness | Another reviewer can understand thesis, trigger, invalidation, and exit reason. | The review depends on memory instead of evidence. | Improve journal fields before changing rules. |
| One next action | The review chooses one change, continuation, or stop decision. | The next sample cannot be compared fairly. | Document one owner and one follow-up date. |
Sample window: A four-week ETH paper-trading sample using prompt version 2, one setup type, and a fixed simulated risk cap.
Inventory: The sample includes 22 simulated entries, 9 skips, 4 missed-trade notes, and 2 excluded entries where the rule version was wrong. The excluded entries are documented but not used for the benchmark decision.
Finding: The paper return is positive, but the largest winner contributes most of the result. Risk stayed inside limits except for two late entries where stop distance widened after the setup moved.
Decision: Mark the sample useful but incomplete. Keep the same setup, add one late-entry guardrail, and rerun the benchmark. Do not rewrite the whole prompt or increase simulated size.
Mark paper results inconclusive when the sample is too small, the rules changed midstream, skipped trades are missing, outliers dominate the result, or the market context is too narrow. Inconclusive is not a failure. It means the evidence should not drive a major workflow decision yet.
Change one rule when the same weakness appears repeatedly and the journal shows enough evidence to isolate it. A late-entry problem should produce a timing guardrail. A sizing problem should produce a risk-control change. A vague thesis should produce a better prompt or checklist field.
Use the checklist in order. First confirm that the sample is clean enough to review. Then check whether the process followed the rules. Only after those two steps should the simulated return matter. This order prevents a strong paper outcome from hiding rule drift, missing skips, or oversized simulated risk.
If the sample fails early, stop the review and fix the evidence layer. Add missing journal fields, separate mixed rule versions, or rerun the benchmark with a stable prompt. A clean smaller sample is more useful than a larger sample that cannot be audited.
Use this checklist after the paper trading journal template has enough entries and before the benchmark review worksheet becomes a decision record. If you need a broader explanation, read how to evaluate paper trading results. If the next question is whether the evidence is strong enough for a separate live-capital discussion, use the paper trading readiness review.
For AI workflows, add AI trading agent without live execution and AI paper trading agent evaluation. The same checklist applies, but the review should also include prompt version, persona version, model notes, and agent output quality.
It is a structured checklist for reviewing simulated results before changing a paper-trading rule, prompt, risk limit, or benchmark.
Check sample size, rule stability, journal completeness, skipped trades, drawdown, outliers, market context, and whether the next action changes only one variable.
No. It validates simulated paper-trading evidence only. Trading Boy does not execute live trades, hold funds, or provide financial advice.