How to Use Alpha Research Copilot

Core features

• - Choose ticker, market, horizon style, risk profile, and analysis mode.
• - Review data warnings, metric explainers (i), and interpretation panels where present.
• - Hand off to Auto Research with preset parameters, or continue manually in Research Lab.

Key outputs

• - Research-screen verdict (candidate / watchlist / reject) with conviction and risk notes.
• - A compact evidence snapshot (IC, Sharpe, drawdown, turnover, etc.) for go/no-go decisions.

Core features

• - Enter ticker, market, style, risk, and benchmark; the system materializes experiment settings.
• - Optional overrides for transaction cost and turnover cap to stress implementation realism.
• - Creates an experiment and routes you to the structured experiment / report view.

Key outputs

• - Full experiment output: validation, portfolio diagnostics, and research memo.
• - A reproducible baseline you can reopen in Research Lab for deeper tuning.

Core features

• - Configure universe, target, features, model, validation, portfolio, and diagnostics.
• - Advanced options: multi-model, walk-forward windows, purged K-fold, stress/regime/SHAP where enabled.
• - Accepts query presets from Quick Insight, Auto Research, or the Factors page (factor IDs and context).

Key outputs

• - Fold-level validation metrics, model comparison, portfolio backtest, and final verdict.
• - After a successful submit, the app usually navigates straight to that run's experiment detail page (no need to look up an ID first).

Core features

• - Library: search and filter by category and market scope; cards show metadata, actions (detail, analyze, Research Lab), and an info (i) popover (description, formula, attribution when available).
• - Library: create custom factors with name, type, description, and `definition_json` (use `source_type` for provenance such as `self_built` or `signal_book`).
• - Exploration: choose universe, horizon, and factors (with optional search/category/market filters); run IC-style diagnostics.
• - Exploration: metric cards and charts include tooltips (i) and a results interpretation panel; link to Research Lab with selected factors.

Key outputs

• - A maintained factor catalog shared with Research Lab and downstream workflows.
• - Optional pre-check evidence (IC, decay, quantiles, correlation, coverage) before a full experiment.

Core features

• - Pick multiple completed experiments, choose equal or optimized weights, then run the analysis.
• - Each experiment row has an i control that opens that experiment's detail page in a new navigation context.
• - Review net Sharpe, drawdown, turnover, cost impact, concentration, diversification, and curve panels.
• - Interpretation panel and metric tooltips mirror other diagnostic pages.

Key outputs

• - Portfolio-level metrics and time-series evidence (cumulative return, drawdown, gross vs net).

Core features

• - Filter by stage, conviction, alpha family, and related fields where shown.
• - Sort and compare cards; open linked experiments or follow-up flows when available.

Key outputs

• - A unified view of signal progression from research toward deployment review.

Core features

• - Automatic handoff: Auto Research and Research Lab redirect to the finished run's detail view when submission succeeds.
• - Return visits: use links on Signal Book cards, Portfolio Analysis experiment rows (they show #id), browser history, or bookmarks—no need to type a URL by hand.
• - If you land on a detail URL, the number after /experiments/ in the address bar is only an identifier for that saved run.
• - Inside the detail page, use tabs to move between validation, factors, portfolio, and narrative conclusions when the build exposes them.

Key outputs

• - A persistent experiment record with metrics, charts, and memo/report views where configured.