On February 15 2010 05:54 paisano wrote:
No, it would be the intermediate-value theorem for continuous functions; one question though, are you sure winning % isn't discreet; assuming Savior can only play a finite number of games, won't the fraction w/p (w - won; p- played) be rational, and thus the function is undefined for the irrationals? If I'm wrong, I'd appreciate some help on this, as it seems you can't really apply the theorem here.
Win % is discrete, which is why the property doesn't hold for every fraction. 80% happens to be one of them. You cannot apply the intermediate value theorem.
I believe this was a Putnam problem from years back.
+ Show Spoiler +
Let w be the number of wins and T be the total number of games played right before Savior's win rate was strictly below 80%. That means
w/T < 4/5
and
(w+1)/(T+1) >= 4/5
First equation gives 5w < 4T, or 5w <= 4T - 1, since w and T are integers. Second gives
5w + 5 >= 4T + 4, or
5w + 1 >= 4T, or
5w >= 4T - 1.
Combined with the first equation, we see that we MUST have 5w = 4T - 1. In other words,
w = 4T/5 - 1/5 and
(w + 1)/(T + 1) = (4T/5 + 4/5)/(T + 1) = 4/5 = 80%, as desired. Basically, the first game that gives Savior a win rate of at least 80% is the game that puts him at EXACTLY 80%.
More generally, the trick works for any fractions p/q such that q - p = 1. In this case, q = 5 and p = 4.