Connect QuanterLab as a custom MCP connector in Claude.ai (or Claude Desktop / Code) and run scanners, backtests, walk-forwards, and regime analyses with natural language. Your QuanterLab account is the boundary — Claude only sees what you'd see in the platform UI.
https://quanterlab.com/mcp as the server URL.Claude knows about every QuanterLab module you'd interact with in the platform UI — scanners, backtests, walk-forwards, regime analyses, and your saved work. Ask in plain English; Claude picks the right tool.
Reference list of the 31 MCP tools Claude can call on your behalf, grouped by the QuanterLab module they wrap. You don't need to memorize tool names — Claude picks them from natural language — but power users may want to see what's available.
Indicator-rule strategy backtesting plus walk-forward optimization on a single ticker.
list_indicators Browse the technical indicator catalogue (RSI, MACD, Bollinger, etc.) with their parameter shapes.start_backtest / …_progress / …_results Run a backtest with entry/exit/confirmation indicator rules. Returns Sharpe, drawdown, win rate, trade list.start_walkforward / …_progress / …_results Rolling re-optimization across folds; reports per-fold OOS metrics + decay ratio so you know if the edge is stable.Fit-and-trade backtests using OU / Kalman / cointegration models, plus universe scanners.
list_stochastic_methods Catalog of available methods (OU z-score, Kalman MR/trend, cointegration pairs, …).start_stochastic_backtest / …_progress / …_results Single-asset (or pair) backtest with a chosen stochastic method and signal logic.start_mr_scan / …_progress / …_results Universe scanner for mean-reverting stocks (Hurst → OU half-life → ADF/KPSS pipeline).start_pairs_scan / …_progress / …_results Find cointegrated pairs across an index (correlation → Engle-Granger → half-life → β stability).HMM regime detection plus per-regime threshold optimization for adaptive strategies.
start_regime_optimization / …_progress / …_results Detect market regimes via HMM, grid-search per-regime thresholds, and report static-vs-adaptive performance side by side.Rank stocks across an index, then build risk-aware portfolios from the survivors.
run_screener Multi-factor (Value / Quality / Momentum / Growth) composite scoring across an index. Tunable weights and metric mix per cluster.optimize_portfolio_hrp Hierarchical Risk Parity weights — robust to estimation noise, no expected-return guess required.optimize_portfolio_mvo Markowitz max-Sharpe / min-variance with Ledoit-Wolf shrinkage and weight constraints.optimize_portfolio_ivol Inverse-volatility weighting — concentration risk gets diluted, low-vol assets get larger weights.Forward-looking risk metrics on a portfolio you've already built.
analyze_var Daily and horizon-scaled VaR / CVaR at 95 / 99 / 99.5%; per-asset risk decomposition; rolling-window history.simulate_monte_carlo Forward simulation across 1m / 3m / 6m / 1y horizons; percentile bands, probability-of-target outcomes, ending-value distribution.Read-only access to objects you've saved in QuanterLab — backtests, scanner runs, configs, projects.
list_my_backtests / get_backtest Your saved backtest results, summary list and full record.list_my_filter_results / get_filter_result Your saved scanner / filter outputs.list_my_trading_configs Configurations you've exported from the strategy builders.list_my_projects / get_project Project boards with attached items + report metadata.Add the server to your claude_desktop_config.json:
{
"mcpServers": {
"quanterlab": {
"url": "https://quanterlab.com/mcp"
}
}
}Restart Claude Desktop. The OAuth flow opens in your browser on first connection.
claude mcp add --transport http quanterlab https://quanterlab.com/mcpThen run /mcp inside Claude Code to authenticate.
Sign in to Claude.ai first and add the connector there. The Chrome extension reads from the same connector list.