Predictive ability in investment management refers to the capacity of a signal, model, indicator, or investment process to reliably forecast future market prices, returns, or risks better than a naive baseline — typically better than simply assuming past returns will persist (historical mean) or that current prices already reflect all available information (random walk). Demonstrating genuine predictive ability is the gold standard for evaluating investment approaches, distinguishing those with a true edge from those that have merely been lucky or curve-fitted to historical data.

What Constitutes Genuine Predictive Ability

Genuine predictive ability requires several conditions. The signal must generate statistically significant predictions in out-of-sample data — data not used to develop or calibrate the signal. The relationship between the signal and future returns must be consistent across different time periods, markets, and asset classes, not just in a single cherry-picked environment. The magnitude of predictive power must be economically significant — large enough to produce returns that exceed transaction costs and appropriate risk adjustments. And there must be a plausible theoretical reason why the predictive relationship exists and should persist as it becomes known.

The Difficulty of Demonstrating Predictive Ability

Demonstrating genuine predictive ability in financial markets is extremely difficult. Markets are competitive: every participant with the same information and analytical framework reaches similar conclusions, and any truly predictive signal is quickly arbitraged by those who discover it. Financial time series are noisy, with low signal-to-noise ratios that make distinguishing genuine signal from random patterns statistically challenging. And the bar of out-of-sample testing is genuinely demanding — many signals that appear predictive in-sample vanish when evaluated on new data.

Signals with Documented Predictive Ability

Several signals have demonstrated genuine, persistent predictive ability across rigorous testing. Price momentum has predicted relative returns across equities, bonds, commodities, and currencies over hundreds of years and dozens of markets. Value measures (price-to-book, price-to-earnings) predict long-run equity returns at the market level. Trend signals predict the direction of major market moves over 3-12 month horizons. These are not guaranteed to work in every period — but their long-run, cross-market consistency meets the standard of genuine predictive ability.

---