A non-random walk down Wall Street

What this book actually teaches

What I'd tell a client

Andrew Lo and Craig MacKinlay's thesis is the academic counterpunch to Burton Malkiel's "random walk" view of markets: stock prices are not a pure random walk, the efficient-markets hypothesis in its strongest forms is empirically wrong, and there are statistically identifiable patterns in returns — though exploiting them is far harder than identifying them. The book collects and extends a decade of the authors' published research, anchored by their famous 1988 variance-ratio test that rejected the random-walk hypothesis for weekly U.S. equity returns.

The arguments build in three layers. First, the empirical case against random walks: using variance-ratio tests across multiple time horizons and asset classes, the authors show that short-horizon returns exhibit positive autocorrelation — momentum — that a true random walk should not produce. The effect is small but statistically robust and survives a range of specifications. Second, they investigate why: market microstructure (bid-ask bounce, nonsynchronous trading), behavioral effects, and slow information diffusion all contribute, and the relative weight varies by horizon and asset.

Third, the implications for investing and for the efficient-markets framework itself. Lo's broader project — developed more fully in his later "Adaptive Markets" work — is that markets are better understood as evolving ecosystems where strategies work until they're competed away. Predictability exists, but it's regime-dependent and decays as participants discover and arbitrage it. The book also covers long-horizon mean reversion, the size and value effects, and methodological problems (data snooping, survivorship bias) that have inflated published return anomalies. The chapter on data-snooping bias is genuinely important and remains relevant to any quantitative researcher today.

Who this is for: quantitative finance students, practitioners doing systematic strategy research, and serious investors who want to understand the academic debate around efficient markets. It is dense, mathematical, and assumes comfort with econometrics.

Weaknesses are real. The book is a collection of academic papers stitched together; it reads as such, with notation shifts and uneven accessibility. Many of the documented anomalies have weakened or disappeared in the decades since publication — exactly as Lo's adaptive-markets framework would predict, but frustrating for readers hoping for usable signals. The statistical machinery requires real econometric background; readers without it will get the conclusions but not be able to evaluate the evidence. And the practical investing guidance is thin compared to the diagnostic work.

Verdict: essential for anyone doing serious quantitative research and a valuable corrective to overconfident efficient-markets claims, but the wrong starting book for general investors. Pair it with Lo's later "Adaptive Markets" for the synthesis.

About Andrew W Lo

Read more from Andrew W Lo and explore the full bibliography on ClearValue Books.