Karthik & 1114: Effective FIRST Strategies 2012

[Bongle]

Originally Posted by Chris is me

This is an interesting question that I hadn't really thought much about. I can only guess it has something to do with the highly "inseparable" nature of the bridge, as well as the Coopertition Bridge. I think a Hybrid-only OPR and a Teleop ball scoring OPR would probably be more useful than general OPR this year - an option not present in the past but now available via Twitter.

The hybrid and teleop-only stuff is available simply because it's in the team standings page. As long as FIRST breaks out individual stats in the standings page, you can do the same regression math and get bridge/hybrid/teleop ratings.

My own thoughts on why it's not as effective this year as past:
-The 1 CP = 5 points hack isn't perfect
-Many top-end teams (including 1114) sacrifice a ton of potential offensive points by sitting on the CP bridge at the end of a qual match waiting for a cooperating robot. This means their OPR will underestimate their 'true' offensive value
-Teams like 1114 that hop on the co-op bridge early and get rejected (or fail to balance) will have sacrificed lots of offensive scoring and gained nothing for it. Result: elite team with a depressed OPR.
-Seeding high in this game depends on CPs (depends on bridge ability and luck) and match wins (depends on offensive power, bridge ability, and luck). A 1-dimensional measure of a team isn't going to predict a high seed.
-Success in eliminations depends on offensive power, robot shape, centres of gravity, and bridge ability. Again, it won't be captured in a 1-dimensional measure like OPR
-Fouls will elevate a team's opponents' OPR rather than depressing their OPR like penalties used to. Result: a team that causes lots of fouls (and thus contributes lots to their opponents' OPRs) won't get detected
-For bridge power, a team that spends 2 minutes to balance will be rated about the same as a team that spends 10 seconds to balance, assuming they balance about the same # of times

All that said, the correlations between OPR and qualification match win rate are pretty good this year. In 2007 and 2009 it was useless because of the exponential scoring and human scoring respectively. However, it's usefulness as a predictor of who would be best for a team to draft for elims is much reduced. Elimination success this year depends on less-measurable qualities like robot shape and CG and how well the robots work together.