What is this “statistical significance” and most importantly, what sort of performance does one need to turn in to achieve a statistically significant result?
David Dreman wrote about this in Contrarian Investment Strategies: The Next Generation in a section entitled “The Vanishing Support for EMH”, so just how does a person go about proving that they can outperform the market, to the satisfaction of all parties? The biggest problem with statistical significance is that it is a weak tool when there is very little data available. Statistics was designed for use with large sets of numbers, you want thousands of data points in your survey and simply put most money managers haven’t been in the industry long enough to have thousands of quarterly performance figures out just yet! When you have smaller data sets, you need to be looking for larger differences to be flagged as statistically significant. When you have one million data points you won’t need very much of an outperformance to show up on a researcher’s screen at the 95% confidence level (generally regarded as the minimum acceptable level of statistical significance), so when someone has c
Related Questions
- What is this "statistical significance" and most importantly, what sort of performance does one need to turn in to achieve a statistically significant result?
- Is the relationship between the Great Lakes ice cover and climate patterns statistically significant?
- How do I tune a NT workstation to achieve a server-like performance?
