The goal of the (awfully complicated) user standings is to rank users according to their performances in all Oracle contests while balancing factors such as crowding (it's a lot harder to win when there are twice as many people playing), overall user strength (it's a lot harder to rank high when everyone pulls out 45ish averages) and hangovers (the penalty for missing a match does not have to be too drastic). So without further ado, let's get into the explanations. (Math freaks can see detailed formulas at the bottom)
The formula used is
(Rank Score) + (Average Score) + Bonus Points
and will be explained for each section:
Rank Score
This section is quite simple. The winner's score gets higher as more people participate, and the score difference between participants gets bigger as you climb in the standings.
A win should net you roughly 3500 points, 10th place roughly 2000 points, 30th place roughly 1000 points and 60th place roughly 500 points. This section is adjusted for contest length (halfseason contests are worth half the points)
Average Score
The higher your average, the better your score. Climbs fast over 42, drops slowly under 42.
A perfect average would be worth 5000, 45 points is worth 2500, 42 points is worth 1000, 21 points is worth 500. Multiplied by a factor for participation (Predicting half the matches will penalize you by ~30%) and adjusted for contest length.
Bonus Points
Bonus points awarded are added to the Rank and Average scores
25 points for a perfect prediction
25 points for 1st place prediction for a given match
20 points for 2nd place prediction for a given match
15 points for 3rd place prediction for a given match
10 points for 4th place prediction for a given match
5 points for 5th place prediction for a given match
10 points for each day in 1st place on the overall leaderboard
8 points for each day in 2nd place on the overall leaderboard
6 points for each day in 3rd place on the overall leaderboard
4 points for each day in 4th place on the overall leaderboard
2 points for each day in 5th place on the overall leaderboard
Maths, formulas and more frightening stuff even...
Detailed formulas for the things above, if you really want to know...
Rank Score
[500*(N/63)*{(Ln(C)/Ln(2))(Ln(R)/Ln(2))}] where:
N is the number of matches in the Contest;
C is the number of contestants who have entered predictions in this particular season;
R is the user's rank for that particular season.
Average Score
[500*(N/63)*((P/N)^.5)*IF{A>42;(A40);2*(A/42)}] where:
N is the number of matches in the Contest;
P is the number of matches predicted by the user;
A is the user's average score per match (without bonus points).
When a user has an average of over 42.00 points, his basic score is 500*(A40), therefore 1500 for an average of 43, 3000 for an average of 46, etc...
When a user has an average of under 42.00 points, his basic score is 1000*(A/42), therefore 1000 for an average of 42, 500 for an average of 21, etc...
That basic score is then multiplied by the participation factor(sqrt(P/N)) which is used to penalize users who have missed a match. Using the square root balances the results by assuming that a user averaging 45pts/match for 62 matches wouldn't have scored a 0 on its 63 match, yet without awarding him the full 45 points he might have obtained.
