Your first backtest
Run one baseline backtest with AI Researcher, then use the completed run to explore your results. Complete AI Researcher setup first.
Choose a strategy and test period
Choose one strategy definition from ~/fundpro/strategy_library and a date range supported by your cluster's data. A baseline records how the existing strategy behaves before you change its rules or optimize it.
If you need help choosing the file or checking available data, ask:
Use AI Researcher to list my strategy definitions and check the available
backtest data for the underlying in [strategy file]. Help me choose a
test period for one baseline run.
Replace bracketed placeholders in these prompts with your own file, dates, or backtest ID.
Run one baseline
Send this prompt in the same task or conversation:
Use AI Researcher to run one baseline backtest of [strategy file] from
my strategy library over [start date] to [end date]. Keep the strategy
rules unchanged. Summarize the settings, return, maximum drawdown,
Sharpe ratio, and trade count. Include the completed backtest link
and the location of the saved results.
AI Researcher uses the MesoSim plugin to submit the run and retrieve its results. Resolve any missing strategy settings it identifies. If the run fails, ask it to explain the reported failure before changing the strategy.
Inspect the result
Open the backtest link in your MesoSim portal and compare the summary with the run's settings and analytics. Use the following checklist to read the response; these are the expected pieces of information, not sample performance figures.
| Part of the response | What to check |
|---|---|
| Strategy and settings | Correct file, underlying, date range, sizing, and execution assumptions |
| Run status and link | The backtest completed and the link opens the intended run |
| Performance summary | Return, drawdown, Sharpe ratio, and trade count, with any unavailable metrics identified |
| Saved artifacts | Paths to the strategy, run evidence, and report so you can return to the work |
| Interpretation | What happened in this run and what remains uncertain |
For example, ask:
Explain the largest drawdown in this baseline. Identify the relevant
period and trades, and distinguish observations from possible explanations.
Make one comparison
Once you understand the baseline, change one rule and keep the comparison conditions consistent:
Use AI Researcher to compare this baseline with a version using
[specific rule change]. Use the same test period, sizing, and execution
assumptions. Preserve the baseline and show both runs side by side.
For broader research, continue to Explore your results. That guide covers Merlin and Q-API through AI Researcher, as well as dashboards and natural-language data exploration in Quantify.