Predictive value is the practical, economic significance of a market signal or investment indicator — not merely its statistical correlation with future returns, but its ability to produce investment decisions that generate profits large enough to exceed transaction costs, taxes, and the risks of implementing the strategy. A signal with high statistical predictive power but low economic predictive value — because the edge is too small relative to costs, or too infrequent to be actionable — has little practical value for real-world investment management.

Statistical vs. Economic Predictive Value

In academic finance, statistical significance (a low p-value) is the threshold for claiming a relationship is real rather than random. But statistical significance does not guarantee economic significance. A signal that predicts returns with a t-statistic of 3.5 may generate excess returns of only 0.5% annually — meaningful in a study but trivial after transaction costs, management fees, and taxes in a real portfolio. Economic predictive value requires that the signal generates excess returns large enough to be worth the costs and effort of implementing it.

Decay of Predictive Value

A critical issue for investment practitioners is that documented signals often lose predictive value after their discovery and publication. As researchers publish findings about a profitable anomaly, capital flows in to exploit it, and the resulting competition reduces or eliminates the predictive edge. This decay of predictive value is why rigorous continuous research, out-of-sample validation, and the combination of multiple robust signals is more reliable than relying on any single documented anomaly.

---