This paper raises some interesting questions about the Efficient Markets Theory (not that I am a fan of it).
This paper investigates whether security analysis ability, estimated from past trading experience, affects individual investors' future stock purchases and portfolio performance. Fixed-effect panel regressions provide strong evidence that the ability to forecast the direction of future short-term excess returns significantly influences investors' future equity purchases. Furthermore, the past trading experience that helps estimate such ability also significantly improves future portfolio performance. However, there is no evidence that other types of stock selection ability impact future stock purchases. Overall, our findings indicate that individual investors do learn from their trading experience.
If investors learn from their experience, and thus improve over time, it means there is something to be learned, which means EMT can't be totally true.