[Bob Steele]

I don't think that win-loss record is a very good indicator of competitiveness.
It certainly says something about a robot.
I do think that being picked for eliminations is a much better indicator.

My reasoning here is that matches are completely random and a team can (and often does) win matches not based on their own ability.

Teams are picked (and note that I say picked and not seeded) because other teams think they are competitive.

The issue here is that what about the seeded teams? Most would be considered competitive.

I also think that a robot needs to be reliable...

We could consider these three factors to be Robot Seed + Robot "attractiveness as a partner" + Robot Reliability

I would propose that a value for competitiveness (for a robot) be a synthesis of something like this:

[(some constant) X Number times seeded/ number of regionals] + [(some constant) X number times picked for eliminations / number of regionals] + [(some constant) X number of matches that the robot is functional on the field / number of matches scheduled for robot]

I would have to do some numerical evaluation with teams that we consider to be competitive to come up with values for the constants. I would think that my initial reaction is that the weight (or constant) for picking should be higher than the other two.

Just my ideas.
A competitive robot does not mean a great team... and we are talking robots here.