Offensive+Defensive Power Rankings

This thread shows a method to figure out the average contribution of each team to their matches.

I feel that the rankings could be inaccurate since it does not take the defence into account.

To include defence, we can calculate both offensive power, and defencive power of each team

so for example:
total score of team abcd = OPR of abcd + OPR of abcd’s allies - DPR of abcd’s opponents.

I don’t have the knowledge to do the rankings in a reasonable time, so I am asking the people who did the OPR rankings to calculate the rankings using this method.

This method will result in a more accurate OPR and also a DPR.

Thank You,
-Borna

As Karthik clarified in that thread, numbers are just numbers. What you do with them, how you interpret them, is completley up to you.

I believe the OPR numbers are a perfectly accurate measurement of a team’s average offensive contribution, after penalties and everything. Maybe there’s a way to calculate DPR. Maybe there’s a way to adjust the numbers relative to the rest of the teams in each of its competition.

Again, the numbers are just numbers. If they don’t help you, or you think they’re inaccurate, I’d reccomend you simply ignore them. I also don’t really see much of a suggestion from you. All I see you saying is that you want someone to figure out how to calculate defensive rankings. If I were you, I would at least suggest something, instead ofj ust coming and telling people to “calculate the rankings using this method.” I don’t see a method here.

Once again, the OPR numbers are perfectly accurate if you take them for what they are. They’re a way to induce some idea about a team’s average contribution from the matches they played and their scores.

I think there was a misconception.
The method that i proposed is to use the equation on the first post and make the N equations and N variables using that. I don’t see a mathemathical problem with that. if there is one can you explain it.

Thank You
p.s. Actually I am pretty happy with the OPR

Maybe I’m confused, but I don’t think I understand you.

Right now, we have n equations for n teams. Hence, we can use those n equations to solve for the n opr numbers for each team. If you propose we somehow use the same n equations, or the same coefficient matrix M and sum vector s to solve for both the offensive and defensive power rankings, I’d say that’s quite impossible. As a general, solving for 2n variables with n equations doesn’t work.

In order to really get DPR

numbers, I believe you either have to change the matrix and vector, to generate 2n numbers in the vector p we’re solving for, or set another set of n equations in n variables to solve for the defensive power numbers. I believe the answer will actually be to set a new system of 2n equations with our 2n variables being the OPR
and DPR
numbers from each time. How you go about setting those equations, well, that’s why they pay the pros the big bucks

I am wondering if this would work.

for each match,
red score = OPR(red1) + OPR(red2) + OPR(red3) - DRP(blue 1) - DPR(blue 2) - DPR(blue 3)
blue score = OPR(blue1) + OPR(blue2) + OPR(blue3) - DRP(red1) - DPR(red2) - DPR(red3)

41 regionals * 70 matches on average * 2 = 5740 equations
each equation only has 6 variables in it so i’m not sure that it would work.