PORIZ rating - technical details, how it works
The ratings are calculated based on a statistical probability table, which
assumes a certain probability of victory based on the difference between the individual
ratings of opponents in a match.
Initially, a rikishi is given an introductory rating of 500 upon his first match
in a rated tournament. He is also assigned an initial Rating Deviation (RD) of 130. After
a match the ratings and RD’s change in the following way: After a bout between wrestlers
W1 and W2: To change the rating and RD for W1 we do:
1. Calculate f = 1 / sqrt(1 + p RD^2) where p = (3 (ln 10) ^ 2) / (400^2 pi^2)
(so p = 1.00724 x 10^-5) and RD is the RD value for the opponent.
2. E = 1 / (1 + 10^(-(r1 - r2) f / 400)) where r1 is the rating of W1, r2 is the
rating of W2 and f we found in step 1.
3. K = (q f) / (1/(RD^2) + q^2 f^2 E (1-E)) where q = (ln 10) / 400 (so q =
5.75646 x 10^-3). If K is less than 50, set it to 50.
4. The new rating of W1 is old rating + K * (w - E) where w is 1 if W1 won and 0
if W1 lost.
5. The new RD for W1 is 1/sqrt(1/(RD^2) + q^2 f^2 E (1-E)).
Go back to step 1 and calculate the new rating and RD for W2.
This produces outcome tailor fit for sumó competition and gives us incredibly
good results.
Wins by fusensho (default) and losses by fusenhai (default) are not taken into
consideration.
Results in the teams contests are included in the ratings for the Open category,
since the team matches are in fact open category matches.
