Momentum strategies have been notable successes in most places researchers have looked. One glaring exception is Japan. There, trying to pick stocks based on price momentum has made no money for near 30 years. This failure has led some to call into question momentum’s viability, suggesting that perhaps momentum’s success elsewhere may be the result of data mining. We reject that interpretation.
We argue that because value and momentum strategies are strongly negatively correlated, they need to be studied as a system. In this context, we show that the results in Japan are perfectly consistent with value and momentum working everywhere at similar levels and are entirely within the range of statistical noise.
Going further, viewed as a system with value strategies, either through ex post optimizations or through Fama-French three-factor model intercepts, we show momentum strategies are actually a success in Japan. In sum, we find the Japanese momentum results supportive of — not contrary to — the idea that momentum is a strong ex ante efficacious strategy around the world.