Explore your results
After your first backtest, use AI Researcher to investigate the strategy and plan further experiments. Open Quantify when you want to explore stored results or build shared dashboards.
| Objective | Where to start |
|---|---|
| Explain a run or compare strategy variants | AI Researcher with MesoSim |
| Discover predictors or optimize strategies | AI Researcher with Merlin |
| Apply quantitative models, statistical tests, or portfolio analytics | AI Researcher with Q-API and Merlin as needed |
| Query stored results and create charts or dashboards | Quantify, using Metabase's built-in Metabot agent or its query tools |
Investigate with AI Researcher
Keep working in your existing conversation, or identify the backtest and saved campaign when starting a new session. The prompts below are example requests; the appropriate analysis depends on your strategy and available data.
Compare strategy variants
Use AI Researcher to compare backtests [baseline ID] and [variant ID].
Check that their periods and assumptions are comparable. Explain the
differences in return, drawdown, and trade count, and identify what
further evidence would help assess the change.
Use Merlin for strategy research
Access Merlin through AI Researcher for predictor discovery and strategy or portfolio optimization:
Use AI Researcher with Merlin to investigate whether entry filtering
improves [strategy]. Define a walk-forward evaluation and compare its
out-of-sample results with the unchanged baseline. State the proposed
experiment scope and assumptions before starting the campaign.
Use Q-API for quantitative analysis
Access Q-API through AI Researcher for statistical tests, models, performance analysis, and portfolio calculations:
Use AI Researcher with Q-API to compare the performance of backtests
[IDs] and investigate portfolio allocations. Check date alignment,
explain the allocation constraints, and compare the result with
equal weights.
The product documentation describes the available methods. Source code, notebooks, and direct API references remain available for custom integrations under Advanced workflows.
Explore stored results in Quantify
Open Quantify (Metabase) to work with saved questions, charts, and dashboards. Its built-in AI agent, Metabot, can help you ask questions about data and develop SQL in the same interface.
To use Metabot, bring your own AI provider API key. AI usage is charged to your provider account. This is separate from your MesoSim key and from the AI Researcher connection. Ordinary Metabase queries and dashboards remain usable without an AI provider key.
Follow Explore results with Metabot for key requirements and a first query. For the underlying data, see BackDB for run-level metrics and RunDirs for detailed run-file analysis.
Return to your research
Bring the relevant backtest IDs and observations from your analysis into AI Researcher:
In backtests [IDs], I observed [pattern] while exploring the results.
Use AI Researcher to check that observation and propose a focused
follow-up experiment. Reuse the existing research campaign and baseline.