Nerds on Wall Street: Math, Machines, and Wired Markets

5.9. Does This Really Work?

It takes forever to distinguish luck from skill with 100 percent certainty, so definitive claims in this regard must be viewed with caution. But these approaches are not weird black-box ideas. They are based on a few very fundamental ideas:

• Market inefficiencies are out there. Use the scientific method to systematically evaluate potential market inefficiencies on an ongoing basis. Look at stocks individually and in groups.

• All the stocks, all the time. Search for these inefficiencies in all the stocks, all the time. The progress in computer technology is one of the wonders of history. IBM's Deep Blue used an analogous relentless approach to chess.

• Keep what you find. Don't lose potential value added in poor implementation. Avoid overly costly trades entirely. Trade effectively to make balanced, diversified bets to add value at the chosen level of risk.

A computerized, quantitative investment process is emotion-free, is without distractions, and is willing to take contrarian positions when they are warranted. All you need is skill, luck, or both in forecasting returns. There is no limit to the amount of information available to you for this, and almost no limit to how many times you can pore over it looking for exploitable inefficiencies. The goal is maximizing predictability. A useful view of maximizing predictability, popularized by Andy Lo at MIT, is illustrated in Figure 5.8.

You really have three things to consider when you are setting out to be a ...