Introducing Harvey Liu Zhu factor zoo for after-work quants

Takeaway. Hundreds of published factors create a multiple-testing problem: a t-stat that looked “significant” under one test is often not enough once the whole factor zoo is considered.

…and the Cross-Section of Expected Returns (Campbell Harvey, Yan Liu, Heqing Zhu, 2016) is the scepticism paper every after-work factor coder should read before celebrating a backtest star.

Fit for learners: ...and the Cross-Section of Expected Returns—multiple testing and the factor zoo. Required scepticism before coding the 314th factor.

This Hub note is a plain-language intro. Not investment advice. Keep it next to BAB, Carry, and every deep-factor Hub note.

The problem

Researchers and bloggers keep inventing factors. Journals and arXiv fill with t ≈ 2.0 stories. If many hypotheses were tried (including unseen ones), false discoveries explode. Naïve p-values mislead.

The idea / core intuition

• Document how crowded the cross-section literature became (the “zoo”).
• Argue for higher hurdle rates on t-statistics under multiple testing—not the single-test 1.96 folklore alone.
• Encourage better research design: out-of-sample, economics, replication—not only in-sample stars.
• For learners: treat every new factor notebook as one more draw from a large search—debug your excitement.

This paper does not say “all factors are fake”; it says “your bar is probably too low.”

Why after-work Python learners care

Python makes it easy to mine 50 sorts before dinner. Harvey–Liu–Zhu is the cultural seatbelt: code less greedily, validate more honestly, and never confuse a pretty seaborn chart with discovery.

Tiny evening exercise: simulate 100 independent noise “factors,” each a random long/short on toy returns; count how many clear |t|>2 by luck; write what that implies for your next real factor idea—learning only.

One-evening reading checklist

1. Define “factor zoo” in plain language.
2. Explain multiple testing without jargon overload.
3. Note why a higher t hurdle appears in the discussion.
4. List two non-t-stat checks you will add to every factor notebook.

Download PDF

Download PDF

Landing / SSRN: SSRN abstract

Important disclaimer

Not investment advice. This Hub note is educational only. Harvey–Liu–Zhu is a research methodology warning—not a ranking of which ETFs to buy or avoid.

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—learn to script carefully first; sceptical factor work needs clean code more than hot takes.