StratCraft
AI-native quant research

One researcher.
The capabilities of a quant team.

Turn market hypotheses into data, factors, algorithms, sweeps, backtests, model comparisons, and reproducible evidence, with AI guiding the workflow and you remaining in control.

Start researchingSee the research workflow

Drop your email. One ping when it's live, nothing else.

Workflow
AI-guided
Decisions
Human-controlled
Research record
Evidence-first

A connected research system, not a strategy generator.

StratCraft organizes capabilities that are usually handed between quant researchers, data engineers, ML engineers, and trading-system programmers. It lowers the programming and coordination barrier without removing statistical standards, market risk, or accountable judgment.

Form a hypothesis

Turn a market intuition into a computable, falsifiable question with explicit constraints and acceptance criteria.

Build candidates

Use AI to create and inspect strategy, factor, and model candidates. Generation widens the search; it does not prove a result.

Run evidence-producing experiments

Review and start data, sweep, evaluation, and C++ backtest operations. Keep inputs, parameters, Artifacts, evidence, and lineage inspectable.

Compare and combine

Start with transparent rule-based and linear baselines. Compare more advanced models only when their evidence justifies the added complexity.

Remember and iterate

Preserve successes, failures, rejection reasons, and context so the next research decision starts from evidence instead of a blank prompt.

Human-in-the-loop

The AI-guided research loop

AI helps at every stage. You review the evidence, approve operations, and decide what enters the next iteration.

  1. 01Hypothesis
  2. 02Candidate
  3. 03Data
  4. 04Sweep
  5. 05Evaluate
  6. 06Combine
  7. 07Backtest
  8. 08Save Evidence
  9. 09Explain
  10. 10User Decision
  11. 11Next Iteration
AI-guided research loopHuman in control
Evidence before promotion

Generate less blindly. Reject with evidence.

Generation expands the search space. Governance decides what deserves to survive. Availability varies by operation; no check implies profitability or production approval.

Point-in-time dataChronological OOSCost and turnoverDuplicate exposureSimple baselinesReproducible rejection reasons
Clear delivery boundaries

Open Research Foundation

Loginless Community AI Studio, direct BYOK or supported local models, base research operations, local C++ execution, Artifacts, evidence, and lineage.

AI Assistance

Local BYOK and local-model routes are independent of login. Optional hosted inference is disclosed separately and never changes entitlement.

Commercial Quant Lab

Separately admitted advanced discovery, evaluation, composition, model research, automation, and governance capabilities.

StratCraft is in early product development and preparing its open-source release for Windows, macOS, and Linux. Autonomous research and live deployment are future governed directions, not current product claims.

© 2026 StratCraft