4 ms·
What really bothers me about these articles is that the authors claim to beat but they only study 37 companies. What about the 38th? Or the 380th. I guarantee
by mattrp 7y ago
What really bothers me about these articles is that the authors claim to beat but they only study 37 companies. What about the 38th? Or the 380th. I guarantee you I can cherry pick 37 companies where a linear regression will accurately predict earnings quarter over quarter... what does that prove?
- mzz80 7y agoA study never proves anything. A p-value only tells you how likely the results are by random chance. Presumably, the companies were chosen beforehand. It is unlikely to have happened by chance that it beats for 37 companies. That’s different then selecting the companies after the results.