Introducing Taming the Factor Zoo for after-work quants

Takeaway. Publishing a pretty t-stat is not enough—Taming the Factor Zoo asks whether a new factor still matters after controlling for the existing zoo, using a double-selection LASSO style framework.

Taming the Factor Zoo: A Test of New Factors (Guanhao Feng, Stefano Giglio, Dacheng Xiu, 2020) is the “does my signal add anything?” paper after-work factor miners need.

Fit for learners: Double-selection LASSO framework to ask whether your new signal adds anything beyond the zoo.

This Hub note is a plain-language intro. Not investment advice. Keep it next to Harvey–Liu–Zhu and Thousands of Alpha Tests.

The problem

Hundreds of factors already exist. A new characteristic that “works” alone may be redundant once the zoo is in the regression. Naive single-factor tests overstate novelty.

The idea / core intuition

• Treat evaluation of a candidate factor as a high-dimensional inference problem among many controls.
• Use double selection (LASSO-style) ideas to select controls and test the candidate’s incremental role rigorously.
• Goal: fewer false “new factors,” clearer incremental contribution claims.
• Complements multiple-testing hurdles: even a high t-stat may not be new information.

Why after-work Python learners care

It is the antidote to “I found a 12th z-score that backtests well.” Before celebrating, ask: residual after the zoo? Perfect mindset once you can run regressions in Python.

Tiny evening exercise: invent a candidate signal that is 0.9× an existing toy factor plus noise; regress the candidate on the old factor; note how “alpha” shrinks—learning only.

One-evening reading checklist

1. State the incremental-testing question in one sentence.
2. Explain why single-factor tests mislead in a zoo world.
3. Sketch what “double selection” is trying to do (plain words).
4. List one notebook habit this paper should change for you.

Download PDF

Download PDF

Landing: NBER w25481

Important disclaimer

Not investment advice. This Hub note is educational only. Feng–Giglio–Xiu is methodology research—not a ranking of tradeable factors.

What this is not

• Not CEF / subsidy framing
• Not Quant or Macro course CTA
• Not a “factors are dead” clickbait thread

Next step

Single course CTA: SoloQuant Academy’s Introduction to Python Programming—linear algebra anxiety drops once loops and arrays feel natural; then LASSO notebooks become approachable.