# Kelly-Thorp and secretary problem



- if you have an expectation (and variation) of the edge, how long should you wait in a stable environment?

if you have finite number of attempts for a given project. **Finite number of bets correction to Kelly-Thorp when you extract/improve information?**

- compare with [Michael Mauboussin – The Babe Ruth Effect: Frequency Versus Magnitude](https://www.valuewalk.com/2015/03/michael-mauboussin-the-babe-ruth-effect-frequency-versus-magnitude/)

## [secretary problem](https://en.wikipedia.org/wiki/Secretary_problem)

How to make quick choices...  If your goal is to end up with a thing (house/spouse/job) in the top # %-ile, reject the first x% and then the select the next one that is superior to  all the others...

![secretary](https://knowen-production.s3.amazonaws.com/uploads/attachment/file/3328/bride.jpg 'secretary')


it looks that the product of [Prob(max) by time] is roughly constant, so then the main choice is which percentile (of best secretaries) to choose?


# Parents

* Short-run versus long-run
* Chess, Poker, and Markets⏎

# Attachments

* [](https://knowen-production.s3.amazonaws.com/uploads/attachment/file/3328/bride.jpg)