2 ms·
It's also solving a different problem: it maximises the when probability that you get the absolute best match, rather than maximising the expected value of you
by dmurray 7y ago
It's also solving a different problem: it maximises the when probability that you get the absolute best match, rather than maximising the expected value of your choice.
- scast 7y agoYes, exactly. I was looking to solve the second problem, not the former. Following the secretary rule is good strategy for a lot of things, but I think we actually have more information here.
- ISL 7y agoFurthermore, I suspect it requires that each measurement is independent and identically-distributed. If your measurements are correlated (as prices generally are -- if they were lower than average yesterday, they are more likely to be lower than average today), the secretary-problem approach will not work.