Discussing a classic example of the misuse of statistics as it relates to successful stock investing, the author explains that the best strategy would be an “out-of-sample” approach that ignores the past and observes whether the strategy works in the future.
The author describes a well-known scam in which investors are duped into thinking that a fund manager can consistently select winning stocks despite the success being attributed solely to randomness. Because financial research is susceptible to statistical distortion, more rigorous analysis is needed to reduce the number of “false positives” (i.e., Type I errors) in the data.
How Is This Article Useful to Practitioners?
The majority of empirical research in finance is likely false, according to the author. Thus, the financial products promising outperformance that companies are selling will fail to achieve their stated goals. Investors need to be more skeptical about the trading strategies that fund managers are trying to sell. Academics need to reduce the number of “false positives” in the data by conducting more rigorous analysis.
The traditional two-standard-deviation test for statistical significance is not rigorous enough. It would be more appropriate to use “out-of-sample” testing where the data are split in half and the strategy is tested in both halves of the data. If the strategy works in both halves of the data, the strategy’s success is not likely due to luck. But because researchers know what happened in the past, they can design strategies to exploit past data. Splitting the data into two halves also reduces the number of observations, which makes discovering statistically significant relationships more challenging.
The author states that the only true out-of-sample approach is to ignore prior data and observe whether a particular strategy is successful in the future. Most investors and fund managers, however, do not have the required patience because they demand winning strategies now, not in three to five years.
Investors need to remind themselves that stock investing is inherently probabilistic; thus, luck will often play a role in any success. The author is correct in concluding that academics need to conduct more rigorous analysis in the future to reduce the number of “false positives” and that investors need to be more skeptical about the successful trading strategies that fund managers are trying to sell.