FinAcademy › Risk Track I › Regression

Module R4 · QTA 7–9, FRM 5–6

Regression, done properly

OLS with one regressor, then several, then everything that goes wrong: heteroskedasticity, multicollinearity, outliers. CAPM beta, APT and the performance ratios ride on top because they are regressions.

4 lessons · 11 interactive figures · 28 exercises

Module progress
—0%
  1. 1OLS with one regressor
    A line through a cloud, chosen to make the squared vertical misses as small as possible. Beta is that line's slope.
    7 exercises
  2. 2Inference on coefficients, dummies, and beta
    Is the slope different from one? Is the intercept different from zero? That is the CAPM tested as a regression.
    7 exercises
  3. 3Multiple regression, the F-test and factor models
    Add regressors and R² can only rise. Adjusted R² and the F-test are how you tell whether the new ones earn their place.
    7 exercises
  4. 4Regression diagnostics: heteroskedasticity, outliers, specification
    The estimates can be right while the standard errors are wrong. Diagnostics are how you find out which.
    7 exercises
Module test

20 questions drawn at random from the 28 exercises of this module, 48 minutes on the clock, marked instantly with the worked solution and a link back to the lesson for every miss.