News for hackers. Stuff that matters.
Re:Quant Job Interview Questions (2009) [pdf] by philipodonnell on Saturday November 12, 2016 @06:03PM (#12940700)
This sounds like a variant of the two envelope problem: https://en.m.wikipedia.org/wiki/Two_envelopes_problem

So is this one of those questions that's not possible to answer you're just trying to see how they handle the stress of the unsolvable?

Re:Quant Job Interview Questions (2009) [pdf] by simias on Saturday November 12, 2016 @06:36PM (#12940835)
But here you know the content of the "envelope" before you decide whether to switch or not.

If the adversary was only allowed one pick I guess I'd only re-draw if I had less than 0.5 since statistically you'd be more likely to have a greater number than a weaker one. I'm not sure what the maths are but a simple implementation gave me an average value of about 0.625 using this strategy, so definitely an improvement over a single pick.

But I have the feeling that since the adversary can chose the same strategy maybe you should be a little more aggressive and redraw if you are at 0.5? But then the adversary would do the same so I guess the solution would eventually converge to a value a little above 0.6?

That'd be my guess at least.

Re:Quant Job Interview Questions (2009) [pdf] by murbard2 on Saturday November 12, 2016 @06:10PM (#12940730)
It isn't. There's a clear, well defined answer.
Re:Quant Job Interview Questions (2009) [pdf] by twic on Saturday November 12, 2016 @06:47PM (#12940888)
I'm just a programmer, but this seems really easy to me. What am I missing?
Re:Quant Job Interview Questions (2009) [pdf] by SonOfLilit on Saturday November 12, 2016 @08:29PM (#12941294)
See my post. As a mathematician, maybe you're missing some interesting theory of why if the opponent has an average of 0.625 and you got 0.55 you should redraw. However, as a programmer, the first thing you're missing is running a quick Monte Carlo simulation and seeing whether your idea actually wins :-)