Rapid Refinement of Alpha Generating Factors

Developing and testing a series of increasingly sophisticated forecasting and risk models have traditionally required extensive programming and custom linking of disparate statistical applications and data sources. An alternative to this effort for quantitative portfolio managers and researchers is to subscribe to third party modeling services which provide access to a specific model or signals.

ModelStation® Model Construction

Ranging from simple linear factor models, to cross sectional or time series regressions, to non-parametric routines, the ModelStation Model Construction module supports the full range of model estimation methods. Either in lieu of or as a complement to using proprietary or third party models, this module offers several key features that make the process quicker and more flexible:

Forecasting Key Benefits:

• Rapid discovery of alpha generating factors
• Accelerate the estimation of effective factor models
• Forecast asset returns using multifactor models
• Keep forecasts and factor coefficients up-to-date in changing market conditions

Forecasting Feature Summary:

• Produce single factor or multi-factor forecasting models using linear or non-linear routines
• Extraction of historical predictions with no look-ahead bias
• Time series and cross sectional regressions with or without weightings
• Proprietary pre-regression genetic algorithm to surface optimal factors
• Support for any factor type including technical, fundamental, or macroeconomic
• Robust regression
• Automatic model re-estimation and updating when new factor data becomes available
• Detailed reporting and charting for discovered models
• Export reports, charts, and model data to external systems

Risk Modeling Key Benefits:

• Build proprietary risk models for specialized universes or investment styles
• Understand and quantify your exposure to risk factors
• Keep risk forecasts and factor / stock exposures up-to-date in changing market conditions

Risk Modeling Feature Summary:

• Time series and cross sectional regressions with or without weightings
• Multiple optimization objectives including Information Coefficient and Directional Accuracy
• Rich data transformation library allows for the creation of complex factors
• Automatic model re-estimation and updating when new factor data becomes available
• Detailed reporting and charting for discovered models