AlphaGen is a quantitative research platform that lets you validate, in minutes, whether an explanatory variable has real predictive power over an asset price.
Welcome to AlphaGen
AlphaGen is a quantitative research platform built for one job: deciding — fast and rigorously — whether an explanatory variable carries real predictive power over an asset price.
Plug in a price series and a candidate variable (weather, satellite vegetation indexes, fundamentals, macro indicators, your own proprietary signal…) and AlphaGen runs the same battery of statistical analyses, signal calibration and risk-managed backtests a quant team would build by hand. The work that used to take a senior analyst two to four weeks comes back as a fully structured response in a few minutes.
What you get out of every calibration
- A continuous intensity signal between 0 and 100 — directly mappable to bullish / neutral / bearish positions.
- An institutional-grade backtest with multiple position-sizing schemes: Target VaR (volatility-targeting), Fully invested (compounding), Fixed notional (constant exposure reference).
- 11 performance statistics on the daily PnL — Sharpe Ratio, Hit Ratio (overall / long / short), Profit Factor, Max Drawdown, Annualized Return & Volatility, Ulcer Index & Ratio, Abs. Exposure Rate.
- A per-variable decomposition so you can see which variables actually drive the model and which ones are noise — with the weights learned during calibration exposed in the response.
- A fully customizable dashboard of charts, heatmaps and KPI tiles (drawdown, high water mark, monthly / yearly heatmaps, NAV vs benchmark…) plus a one-click multi-sheet Excel export.
Who uses it
| Role | What AlphaGen helps with |
|---|---|
| Prop traders & quants | Screen ten times more variables per week with a standardized methodology you can audit and compare. |
| Asset managers | Build multi-factor models where every component is validated with the same statistical pipeline. |
| Industrial hedgers | Convert operational data (NDVI, weather forecasts, stocks levels) into actionable signals to time forward purchases. |
| Risk management | Independently re-run a trader's hypothesis with a full audit trail and reproducible results. |
The mental model
Project → Calibration → Prediction
You create a project to scope a study (e.g. "TTF gas" or "EU wheat"), you calibrate a model on historical price + variables, then you predict on fresh data without retraining. Calibrations and predictions live side-by-side, are versioned, taggable, and exportable.
Where to start
-
New to AlphaGen? → Your first calibration — a 20-minute UI walkthrough on a real TTF gas + weather-forecast dataset. No code, no notebook.
-
Prefer the API? → API quickstart — end-to-end notebook tutorial, calibration and prediction in under five minutes.
Tip — every page in this documentation is cross-linked. If a term shows up that you have not seen before, click it: there is usually a dedicated page that defines it from first principles.
Was this article helpful?