King Kong's score wrong ?

Bruce · Lower Club Player

King Kong's score wrong ?

steven · Tennis legend

Well spotted - you're right, but I was already aware of this. I've taken a shortcut here, since the gorillas were only ever meant to be a bit of fun and to act as a guide to the worst it should be possible to do.

R1 is easy - 50% of the gorillas will get each match right.

In R2, if a player who had a bye in R1 wins, 50% of the gorillas will get it right, but if they get beaten, only 25% of the gorillas will get it right because only 50% of the gorillas who predicted the seed to lose will have predicted the right player to get through to face him in the first place. This is exactly what you just explained, of course. So you're right, they should have had 1.5 points at that stage, not 1.25.

However, this is where the simplifying 'fudge' comes in - the results assume that 37.5% of the gorillas get each R2 match right (and that R3 starts with 37.5% of the gorillas having the right player in branches that had byes in them and 25% of them having the right player in branches that didn't have byes in them), but that approximation (which has 2nd, 3rd and 4th order effects, etc, in later rounds too) only holds when averaged over the draw as a whole if half of those who had byes in R1 lose in R2 and half of them win.

Usually, a lot more players with R1 byes win in R2 than lose, simply because they are the top 8 seeds, so the true gorilla scores are usually a bit higher than the approximate scores shown.

This could be corrected for with quite a bit of work, it's just not a high enough priority at the moment. Conversely, it seems silly to take the gorillas out altogether when they do provide a rough guide to the effect of completely random picks just because the King Kong figure is not completely accurate yet.

Hope that makes sense! smile

Bruce · Lower Club Player

No !

The two matches have one seeded player winning and one unseeded player winning. Hence your stated short-cut method should get the same answer as doing it correctly. i.e. 0.375 * 2 matches * 2 points per match = 1.5

How do you get 1.25 ?

GC77 · County player

Bruce,

Steven has calculated an average over the whole of round 2, therefore matches involving seeds have an average of 0.375, and matches not involving seeds have an average of 0.25, hence an overall average of 0.3125.

So for 2 matches in round 1 the gorillas would get a score of 0.3125*2*2=1.25

steven · Tennis legend

As anyone who frequents the Rusedski board knows, Bruce is the person who does his best to keep me on the straight and narrow as far as accuracy is concerned. A thankless task, I must admit, which he nevertheless does extremely well. smile

Getting to the point, Guy's completely correct (he does have the advantage over Bruce of having the spreadsheet to look at this week, obviously) - in other words, at the moment King Kong will end up with the same score from every 56-player event whatever the results are.

I will get around to making that calculation more accurate one day. At the moment, it's still better to have a slightly flawed approximation to a totally random set of picks to compare with than none at all, I think.

-- Edited by steven at 17:27, 2007-05-17