Quote:
Originally Posted by Nathan Streeter
But does that make OPR a poor indicator of a team's performance - compared to actual points scored by a team - or a better indicator? From the standpoint of "how much value does team ____ add" in the context of a 3-team alliance, I think it actually makes OPR a better metric. If a team is only capable of a skill set that adds little value (when partnered with two event-average teams), then I don't think they should have a high OPR.
|
The issue lies in event-event comparisons. Because each defense can only be crossed a maximum of 2 times by an alliance, a team's OPR at any given event is influenced by the average ability of teams at the event. Let's consider 2 teams that are capable of identical performances:
Team A is at an event where the average pair of random partners can fully damage 1 defense. This means that in order to get the breach, Team A must fully damage 3 defenses.
Team B is at an event where the average pair of random partners can fully damage 3 defenses. This means that in order to get the breach, Team B must fully damage 1 defense.
Let's say they can both score low goals well. Now perhaps Team A has limited time to score low goals, but Team B may have plenty of time to pop in a few low goals. Team A will (on average) have an
inflated OPR based on the level of play at their competition by 20-
n*2 points over Team B's, where
n is the number of additional low goals Team B has time to score in a match.
For example, this may explain in part why Team 1983 had an OPR of 45.73 at Auburn Mountainview while Team 67 had an OPR of 37.73 at Waterford. Based on my (subjective) viewing of the matches, both teams had highly capable nearly identically performing robots. Sure if I had to give a slight edge to what I saw this weekend, it would go to 1983 but I would say the difference in terms of robot capability was on the order of 2 points or less (one low goal).
Again, subjective/anecdotal information, but the lack of linearity in defense scoring REALLY makes OPR less valuable as a yardstick. It still trends with performance, but an OPR of 40 at one event is not necessarily the same as an OPR of 40 at another event. And this is more true than in past years because of the mentioned scoring mechanics for defenses.