Only 9 of the estimated best 50 offensive RAPM pairs from 2008-11 (I assume that is thru spring 2011?) are still possible. Only 5 of the estimated best 50 defensive pairs for that time period are still possible. Only 7 of the best 25 offensive RAPM pairs from just the 2010-11 season are still possible (if I am understanding the labeling). Turnover hurts and can't be totally controlled but maintaining estimated highly productive RAPM pairs does not seem a high team priority or at least a successful one.
Only 1 of the top 50 5 man lineups from 2008-11 is still possible. A minor one for OKC. Were none of the others possible to save and considered important to save for further use in the future?
Only 149 lineups played 1000+ possessions over 3 seasons? About 5 per team across 3 seasons? Man, that is not conducive for understanding many lineups. Understanding the core unit of the game. Only 19 lineups got 3,000+ possessions of play and evaluation. 4 of the top 12 most used / proven very effective lineups played in one or more of the 3 Finals for the Lakers, Celtics and Magic. No other team got to the Finals in that time period without one of these 12 most used lineups. Not a single lineup used 3,000+ possessions in this time period is still possible to put on the floor. I might have missed one but I only see 2 used over 1000 possessions in this time period still possible. Both for the Thunder. They don't last long. Some of them should probably last longer, or least be given a higher priority to try to make them last longer.
Here is a mini-test that might be worth doing. For the Celtics vs Lakers for 2 recent Finals, how well does individual RAPM explain those series in general and how well does individual and 5 man lineup RAPM explain specific match-ups of the biggest minute lineups? That may be the best case we have to test important lineup vs lineup data.