Factor betas and hedge ratios begin with a regression. A wrong uncertainty estimate can turn noise into a false alpha claim. This topic separates errors in the conditional mean from errors in its covariance. It connects the OLS baseline to diagnostic tests, robust standard errors, and GLS.
The programme as a whole bridges undergraduate statistics into research design, systematic trading diagnostics, and the empirical discipline required by financial machine learning. Regression foundations are the first gate: if the covariance statement is wrong, every downstream alpha claim is wrong for the same reason.
By the end of this note you should be able to:
- restate the classical linear regression assumptions and Gauss–Markov, and map each common finance failure mode to a later topic;
- detect and respond to heteroscedasticity, serial correlation, specification failure, and multicollinearity;
- walk a CAPM or APT-style regression through a diagnostic checklist and choose HC, HAC, or a model rebuild;
- derive GLS for known , recognise WLS as the diagonal case, and state feasible GLS honestly.
In this lesson
- Contents
- The big picture
The full lesson is for members.
Read the full lessonNew to SelfMadeQuant.com? See membership options or browse the Econometrics and Financial Time Series syllabus.