I got data from the ISR District 1, and compared it with final rankings, giving me a correlation for each category.
closer to -1: the more of that category, the more associated it is with a higher rank
closer to 0: little to no affect on final rank
closer to 1: the more of the category, the more associated it is with a lower rank
Category
Correlation
ELO
-0.78
OPR
-0.70
Stage One Activated OPR
-0.59
Teleop Balls Outer OPR
-0.55
Auto Balls Outer OPR
-0.46
Endgame Climb and Level OPR
-0.44
Endgame Climb OPR
-0.43
Teleop Balls Inner OPR
-0.42
Auto Balls Inner OPR
-0.38
Auto Line %
-0.38
Auto Balls Low OPR
0.00
Teleop Balls Low OPR
0.23
Fouls
0.25
if you do not know ELO or OPR, I very much encourage looking them up and reading up on them
Interested me to see how high outer port is ranked, meaning teams that can just make in outer turn out better week 1 than teams who can hit inner (probably due to lack of aim, will change over time as teams get more accurate). Also interested to see Climb so low, I always viewed it as a much larger impact, but this does mean that endgame is not as heavily weighted as years past, meaning you can cycle enough to make up for a teammate with no climb. Also interesting to see Teleop Low Balls actually hurting overall rank. I attribute this to these teams not being able to do much else, ie climb or even have an auton sometimes, that lowers overall score in games and leads to losses, hurting rank.
I looked into it and I only have data for the alliance fouls, not per team, sorry. I am interested in knowing I will see what I can do. Maybe take the fouls per match, split it amongst each team on the alliance, average them throughout the event, then do correlation from there?
How is dealing with Foul points per alliance any different than Power Cell points per alliance? Or is your original analysis based on scouting data that’s per-team?
You can calculate Penalty OPR with:
Ax=B
where A is the matrix representing who played who, X is the matrix represnting penalty OPRs, and B is the matrix representing the sum of penalty points the opposing alliance has scored against each team.
Key here is that it is the opposing alliance penalty points, not your own alliance.
It’s been a long time since I did any linear algebra, but I did get a vague understanding of OPR from the Blue Alliance blog a couple years back. I’d think you’d just run it using opponents’ Foul points rather than your own total score or Outer-Power-Port points or whatever.
I feel that in the coming weeks, OPR will highly depend on inner balls during teleop, it seems that in ISR #1, many teams are depending on auto and endgame to increase their OPR. Great post Ben!
Correct, I think that as time goes on, teams will become more and more accurate, and thus the better teams will have a more accurate shot. This leads to Inner goal having a higher affect on final rank, and I think that this increase in points and cycle time will drive down other aspects of the game (ie crossing auto line and certain parts of endgame). However it will be interesting to see the fluctuation between endgame and inner as more teams get better at it. Thanks Ethan Pong!