One problem with using past results to determine who is a good stock picker, if you start with a large sample, is that just by random chance some members of that large sample will be consistently good.
The longer the period your considering, the smaller the number that would make it through on random chance, and the more likely that factor doesn't make the data meaningless. A longer period would mean its more likely that their abilities might change, or no longer fit a new environment, but that typically would not be as big of factor in evaluating people from a large sample as the first, and if they do mentally decline, get distracted, or their skills no longer fit changed circumstances it shouldn't take long to see the results. Sure by random chance they could still get it right, but now you evaluating not the cherry picked best from a large sample, but specific individuals you have already screened out from the large sample, so your far less likely to find consistent outsized results going forward. |
