Feature and Signal Engineering

Under development

This lesson is part of the course scaffold and is being actively written. The learning objectives and outline below define its final scope.

Learning objectives

By the end of this lesson you will be able to:

• Transform raw market data into standardized, forecast-ready features and document each transformation's rationale.
• Compute the information coefficient of a signal, track its time series, and test whether it is distinguishable from zero.
• Measure signal decay across horizons and signal turnover, and quantify their joint effect on realizable performance after costs.
• Combine several correlated signals into a composite using a defensible weighting scheme and show the marginal contribution of each.

Outline

1. From raw data to features — cleaning, transformations, standardization
2. From features to signals — forecasts, scores, cross-sectional neutralization
3. Information coefficient — computation, IC time series, significance
4. Signal decay — horizon analysis and decay curves
5. Turnover — measurement and its interaction with trading costs
6. Signal combination — correlation among signals, weighting schemes
7. A signal research workflow — templates, reporting, and a signal's paper trail

Prerequisites

• Momentum and Trend Following