What Works on Wall Street Ch. 1: The Folly of Forecasting

"Models beat human forecasters because they reliably and consistently apply the same criteria time after time." — James O'Shaughnessy

Investment Context

O'Shaughnessy used the Compustat database to backtest decades of market data from 1951 onward, answering one question: which strategies actually make money, and which are noise?

His first conclusion was devastating to the traditional industry: measured against strict quantitative models, human experts almost always lose.

But read the conclusion precisely. Models win not because they are clever but because they are consistent. That distinction runs through the whole book and is the easiest thing to misread.

The Wall Street Translation

1. Human Inconsistency

People cannot process large volumes of data without bias. We tire, we get emotional, we fall in love with stories, we panic in crises. Even an expert with an excellent set of rules will eventually break them, because intuition insists this time is different.

This finding did not originate in finance. The psychologist Paul Meehl showed in 1954 that simple statistical rules consistently outperform expert judgment in clinical diagnosis, and dozens of studies across fields have since replicated it. O'Shaughnessy's contribution was importing a known regularity into equities and verifying it on a large sample.

2. The Discipline of a Model

A quantitative model never tires, never liquidates because of a frightening headline, and applies identical criteria to every stock in the database without exception.

3. A Necessary Qualification

Consistency is an advantage only when the rules themselves are sound. A bad rule applied consistently loses money faster and more completely than one applied haphazardly.

So the book's real claim is not "use a model and you will win" but "if you have reliable rules, a model guarantees you execute them." Whether the rules are reliable is the serious question Chapter 5 takes up.

Actionable Trading Rules

1. Discount analyst price targets: Sell-side ratings are shaped by structural incentives and behavioural bias, and their historical accuracy trails simple models.
2. Do not buy on a feeling: In a system as complex as the market, warm sentiment about a company's new product carries almost no information.
3. Put the rules on paper: Write buy and sell conditions in advance. Their value lies not in sophistication but in deciding for you when your emotions are running.

Relevance to a Retirement Portfolio

For retirees the key lesson has nothing to do with stock selection: your largest risk is not picking the wrong stock but changing your mind at the wrong moment.

A written allocation plan — "60/40 stocks and bonds, rebalanced annually" — works for exactly the reason this chapter describes: it makes a consistent decision for you during a crash, when the urge to flee is strongest. The return from that discipline usually exceeds anything a stock-selection method contributes.