AI Research Tools
Connect backtesting to an AI workflow: specify rules, calculate results and retain the review record.
From natural language to an inspectable record.
The researcher defines the question, AI structures the request, the system calculates, and the researcher checks assumptions, execution and risk. We are packaging this workflow into MCP tools and independent skills.
- Existing core: bar backtesting, cost modelling, parameter research and validation records.
- MCP direction: connect tools in authorised data environments with run and provenance records.
- Independent skills: rule specifications, research plans and reviews that do not require local price data.
Inspect the demonstration, then discuss integration.
Controlled demonstrations are available. Public paid MCP and skill packages are in development. This website does not run research or execute trades inside your account.
Download the free backtest review skillThe sample is a review checklist; it does not connect to systems or accounts. Paid capabilities, versions and prices will be published when deliverable.
One invocation. Inspect every step.
This is an actual invocation of our BarBacktester. Twelve synthetic bars and two specified signals check next-bar execution and costs. It is not market performance, a recommendation or an out-of-sample test.
01 Specify the request
Specify the signals, two-bar holding period and 5 bps per-side costs. No market or account data is used.
02 Calculate with the engine
The engine returns entry, exit, costs and net returns. The language model does not invent result values.
03 Review the calculation
Check signal/entry separation and cost deduction. No live account is used. Two trades cannot establish a strategy’s validity.
Run recorded 2026-10-19T17:02:44.787398+00:00 · Engine revision and file hashes are in the JSON.
Start with one concrete workflow.
Tell us your institution type, current workflow and the outcome you need. Scope, timing and pricing follow discovery.
Discuss your requirements