An aside first - I'm assuming we're talking about players playing starter level minutes through this discussion - being successful/useful in a smaller role and how to measure that is a different discussion.
Smoothbutta wrote:I agree you can't take on-off in a vacuum to judge players, of course. But I disagree with the sentiment that you can't compare across teams (when looking at several seasons). Within a team in a single season is often worse, with some examples later below.
I'm not going quite as far as saying one 'can't', but I do feel it's of very limited use/clarity (I'm sure we could all agree that a +15 is more impressive than a 0, but the difference between a +10 and a +7, hard to tell), even at career level lengths. I don't think any of your examples really get into any within a team situations - which ones were you thinking there? I do agree that we tend to see improved/more stable measurements at larger sample sizes.
Smoothbutta wrote:You are saying it is harder to have high On-Off on a good team? I disagree with this, I think it is more about specifics of replacement players. For example, Mutombo's replacement during some of his best On-Off years was Christian Laettner who was horrid on defense and therefore makes Mutombo look even more amazing on defense while Hakeem's was Kevin Willis who was good on defense too, making Hakeem seem like a smaller impact player defensively than he really is. For the 2016 warriors, Steph and Dray have insane On-Off even though it's a good team because their replacements were guys like Livingston/Speights/Barbosa.
Not trying to say that, I wouldn't use 'hard' at all, it's just more or less likely - and I'd actually go with the reverse. A high on/off is the product of having a good on rating and your team having a relatively poor rating when you're off (some balance thereof). Being on a good team makes having a good on rating (as a starter) almost a given.
It is harder to be on a good team in the first place, but given the team is already good or bad it's more likely to find a high on-off on a good team.
Smoothbutta wrote:In the case of the 2011 Bulls, Derrick Rose's replacement was CJ Watson who was not an amazing player as seen also from his On-Off on other teams, so it is concerning Derrick Rose's On-Off couldn't be really high in an MVP season. I am sure the other factors like Coaches rotations impacts this a lot too. I would probably conclude that Derrick Rose's on court impact was not reaching the heights of the peak years of someone like Steph/Lebron/Nash etc. and that no one player on that Bulls team had insane impact based on the On-Off data. But of course Drose is Drose and with context from advanced stats like BPM you can see he did a ton for the team.
A) It's difficult to differentiate between Rose/Deng with impact measures for that season. B) I agree with a baseline that they should be seen below those guys listed and that neither were 'true' MVP level. C) I find the on rating for the duo, +9.6 in 2752 minutes, meaningfully more informative than how either did in their few minutes alone or how the team did in the bench minutes without either playing. On/off gives that off portion a lot of weight, past the point of usefulness (for comparing players across teams) imo.
