Skip to content
Finance Lessons

Factor Models

Final Exam: Factor Models

A graded, no-retry final exam on factor models — CAPM's failures, Fama–French three- and five-factor models, momentum and its crashes, estimating loadings and premia, alpha vs factor exposure, and the factor zoo. Pass mark 70%.

12 min Updated Jun 6, 2026

Six lessons hammered one idea from every angle: the return you can explain by exposure to known, paid-for risk factors is not skill — it is beta wearing a costume. You watched CAPM’s single market beta fail to explain why small and cheap stocks outrun their betas, met the Fama–French three- and five-factor lineup (SMB, HML, RMW, CMA), survived momentum and its violent crashes, learned to estimate loadings and premia without fooling yourself, and finally separated true alpha from hidden factor tilts in the sprawling factor zoo. This is the capstone, and it does not forgive: every answer locks the instant you submit, the distractors are the exact traps the lessons warned you about, and your score stays sealed until the final question.

Warning:

How this exam works

This is a graded exam. Questions arrive one at a time. Once you submit an answer it is final — there is no going back, no second try, and a wrong answer simply fails that question. Your score stays hidden until the very end, where you need 70% to pass. Read every option before you commit.

Question 1 of 24

A portfolio manager points to a basket of high-beta stocks and argues it must deliver high returns because CAPM says return rises with beta. Empirically, what actually happens across the cross-section of stocks?

Select an answer to continue.

Course Recap

Big picture

Factor models — the whole arc

  • Factor Models
    • CAPM fails
      • Security market line is too flat
      • High beta does not mean high return
      • Size and value go unexplained
    • Factors & premia
      • Factors are compensated systematic risks
      • Premium = pay for bearing that risk
      • Economic story, not data-mining
    • Fama–French 3-factor
      • SMB: long small, short large
      • HML: long value, short growth (book-to-market)
      • Long–short spreads, not single bets
    • Momentum & FF5
      • Carhart adds UMD (winners minus losers)
      • Momentum crashes — not free money
      • FF5 adds RMW profitability, CMA investment
    • Estimating loadings
      • Slopes = loadings, intercept = alpha
      • Long, multi-regime windows
      • Fama–MacBeth premia; a low t means no premium
    • Alpha vs exposure
      • Beating the index is not alpha
      • R-squared is variance, not return
      • Information ratio = alpha / tracking error
    • Smart beta & the zoo
      • Hundreds of mostly redundant factors
      • Decay out of sample, ~58% post-publication
      • Demand t above 3, not t above 2
From CAPM's broken single beta to the factor zoo's decay: explain returns by paid-for exposures, then test whether any real alpha is left.

Key Takeaways

Success:

What you now carry out of this course

  • High beta is not high return. CAPM’s single market beta is too flat to explain the cross-section — size and value slip right through it.
  • A factor is a paid-for risk. SMB, HML, RMW, CMA, and UMD are long–short spreads compensating systematic exposures, not single outright bets or proof of skill.
  • Loadings are exposure, the intercept is alpha. Estimate them on long, multi-regime windows; a Fama–MacBeth premium with a low t-stat (near 1) is no premium at all.
  • Beating the index is not alpha, and R-squared is variance, not return. An R-squared near 1 just outs a closet indexer charging active fees for beta.
  • Attribution can make alpha negative. When factor contributions exceed the return, the manager subtracted value — judge skill by the information ratio (alpha over tracking error), not a lucky year.
  • Momentum is compensation for crash risk, never free money. Its rare, brutal reversals are the price you pay for the premium.
  • A backtest is not a live return. Anomalies decay out of sample (around 58% after publication) and costs can shave a 5% backtest to roughly 0.6%.
  • More factors is not more insight. The factor zoo is mostly data-mined redundancy — demand a t-stat above 3 and out-of-sample survival before you believe.

Mark lesson as complete