post-competition analysis of strength of allies vs strength of opposition

Sometimes it feels like our end-of-qualification position is all out of kilter with how the drivers and robot performed; we’ve had it go both ways (good standing with a robot having a bad weekend, 38th with a bot that seemed pretty solid).

One of the subjective things seems to be “our partners carried us” or “we just drew weak partners/great opposition in qualifications”. I was going to do a little analysis of our past season qualification matches, and compare the average scores and ranking of alliance partners to the average scores and ranking of opposition in our qualifications to see if that subjective feel has basis in fact (recognizing that my robot being a loser will drag down my alliance partner’s rankings and scores).

**Are there existing tools for doing this analysis, or will I need to roll my own? **It won’t be hard, but I feel really lazy… <done for season, whew>

I know the feeling for sure - this year we seemed to end up about where our robot deserved to be, but 2702 has ended at the bottom of the standings some years when our robot was far better than that.

One mildly accurate evaluation of a given team’s strength is OPR. You could look at the average OPR of your alliance partners, average OPR of your opponents, and see how those averages compare with the OPR distribution at the actual regional. Actually, I could probably throw that kind of average calculation into OPRNet in about an hour.

Example: at Waterloo this weekend, we have 33 opponents and 22 alliance members. If our alliance partners had an average OPR of 5.5 and our opponents had an OPR of 11, we could say we had a tough schedule.

There’s a bunch of ways that people analyze schedule difficulty in pro sports. Here’s a good starting point: http://en.wikipedia.org/wiki/Strength_of_schedule

At Buckeye, our robot had more functionality and was performing better than at FLR, and yet at FLR we were 8th seed overall and Buckeye found us near the very bottom of the heap (and yet we were picked 7th for the elimination rounds).

One of the interesting things about the coopertition bridge was how seeding turned out vs. individual robot performance – good scouting was absolutely critical!

OPRNet v24 (available here) now can do schedule analysis. From eyeballing my own team’s schedule difficulty, as well as 1940 and 1551’s, it doesn’t look like it is that useful of a tool (there isn’t much correlation with what the tool spits out vs how a team did … maybe I should include the team in their alliance’s strength)

Does the program include the elimination matches? Being on 188’s alliance for eight elimination games would certainly skew the results toward “high alliance OPR” but not have anything whatsoever to do with seeding.

OPRNet always ignores eliminations matches for precisely the reasons you ask.

Well, maybe we were just awful, then. :ahh:

I’ve always used DPR (I think you call it SAA) as a measure of schedule strength. Never actually done a statistical analysis, but I’ve found usually when someone says ‘Dang our schedule stank!’ they’ve had high DPR and when they coast to the top 8 it’s a small number.