01
Authorize securely
Add the hosted MCP URL, sign in with Google, and approve short-lived OAuth access. No customer passwords or API keys.
Private quantitative infrastructure for AI
Institutional research. Conversational speed.
Connect Claude, ChatGPT, Codex, Cursor, or another compatible AI to a private quantitative research engine. Test ideas, challenge assumptions, compare variations, and preserve the evidence behind every result.
Built for serious investors, professional traders, quantitative researchers, and investment teams evaluating research workflows.
Illustrative demo · Helvetic Research
Connected
Best Sharpe1.21
Max drawdown-12.8%
Best parameterVIX 24
Works in
Claude Web/Desktop
Claude Code
ChatGPT
Codex
Cursor
OpenCode
AntiGravity
0MCP research tools
0+Risk & trade metrics
0Robustness grade factors
0%Reproducible & audit-ready
A controlled research workflow
Your AI becomes the interface. Helvetic becomes the engine.
The conversation stays familiar while strategy execution, assumptions, metrics, storage, and authorization remain structured and auditable.
AI interfaces
Claude
ChatGPT
Codex
CursorRequests in, evidence back
Helvetic Research MCP
OAuth protected · reproducible · streamable HTTP
Structured market research
Market coverage
Equities & indices
FX & rates
Commodities
Crypto
Requests in, evidence back
OAuth + PKCE secured
Streamable HTTP MCP
AI workflow handoff
02
Describe the research question
Start conversationally, describe strategy rules, compare assumptions, or ask your AI to turn an idea into a reproducible Helvetic research run.
03
Challenge and document
Compare variations, optimize parameters, test robustness, and export reproducible reports with assumptions and disclaimers.
Research engine
More than a backtest.
Explore the research modes your connected AI can orchestrate.
85 tools, one conversation
Every step of the research process, callable by your AI.
From symbol discovery to reproducibility audits — each capability is a structured MCP tool with recorded assumptions and research-only boundaries.
85 tools, one conversation
Every research tool, callable by your AI.
Try asking:
Tool descriptions are illustrative. Your connected AI orchestrates them automatically — you never call them by name.
Evidence, not just answers
See the result. Inspect the assumptions. Preserve the report.
Connected AI clients return research into the conversation while Helvetic preserves structured results and produces branded, reproducible outputs.
VIX-filtered S&P 500 strategyWalk-forward validated
Annual return12.4%
Sharpe ratio1.21
Max drawdown-12.8%
Research report previewPDF / Excel / CSV
Helvetic Research MCP
VIX Risk-Off Strategy
Return12.4%
Sharpe1.21
Drawdown-12.8%
Includes equity-curve and drawdown charts, strategy specification, assumptions, engine version, dataset version, risk metrics, and research-only disclaimer.
Sample interface and illustrative figures shown for demonstration only — not a real result, track record, or investment advice.
Built for scrutiny
Trust comes from what the product explains.
Review the architecture, methodology, data boundaries, privacy controls, and research-only mandate before connecting an AI client.
Security
Provider-backed identity, OAuth-protected MCP, hashed tokens, and revocable AI connections.
Review security →Methodology
Structured execution, reproducibility metadata, and robustness testing.
Review methodology →Data coverage
Clear provenance, provider availability, daily-history limits, and modeled-data boundaries for every research result.
Review data →Privacy
Private workspaces, retention controls, deletion, and connection revocation.
Review privacy →Research boundary
No execution, no brokerage access, and no investment advice.
Read disclaimer →Reproducible by design
Research Methodology
The principles used to turn a conversational strategy idea into auditable quantitative research.
One backtest, one defensible grade
A 0–100 robustness score from eleven weighted checks. Hover a factor to see how it's scored. Anything the engine can't assess scores nothing — so skipping the hard checks can never round the grade up.
A
86/100
excellent robustness
Evidence coverage 100%
Hover any factor to see exactly how it's scored.
Is this Sharpe real, or just luck?
Try enough strategies and one will look brilliant by chance. The Deflated Sharpe Ratio (Bailey & López de Prado) computes the Sharpe you'd expect to reach by luck alone after N attempts — your result has to clear that bar.
0.0luck bar → your Sharpe3.0
Both labs above run the engine's real formulas — the same scorecard weights and Deflated Sharpe math that grade every result. The full methodology covers the deterministic pipeline, the golden-reference validation suite, and the published methods behind each number.
Read the full methodology →Questions, answered
What serious researchers ask first.
Built on published methods — the Probabilistic & Deflated Sharpe Ratios and PBO/CSCV overfitting tests — with full governance documentation behind every number.
Is my research private?
Yes. Every result is account-scoped and isolated, you control retention and can delete anytime, and connected AI clients are revocable.
Is this investment advice?
No. Helvetic is research-only — analytics, backtests, and simulations. It never trades, connects to brokers, or makes recommendations.
What data does it use?
Daily history for equities, ETFs, indices, FX, commodities, and crypto (Yahoo → Tiingo / Coinbase) plus FRED & ECB reference data. Every result records its data provenance.
How do I trust the numbers?
Every run is reproducible from a strategy and dataset hash, graded by an 11-factor robustness scorecard, and exportable as a self-verifying audit bundle a third party can independently re-derive.
Which AI clients work?
Any MCP-compatible client — Claude, ChatGPT, Codex, Cursor, OpenCode, Gemini and more — connected over secure OAuth.
What does it cost?
Free to start, no card required. Paid plans raise your monthly research credits, kept-result retention, seats, and workspace capacity. See pricing →
Get started
Start with the research workflow.
Free access is available now — connect your AI assistant and start researching. Upgrade to Pro or Max when you need a larger research allowance, longer retention, more seats, and more workspace capacity.