Very interesting. I'd like to see the comparison between a regression on 'weakest link' 'strongest link' and average PER across starters.
Another factor is that this doesn't weight for time on the court. Someone could go into the lineup data and extract the PER for the lowest player on the court for each lineup and then create a weighted average based on how long each lineup played.
Another factor is that this doesn't weight for time on the court. Someone could go into the lineup data and extract the PER for the lowest player on the court for each lineup and then create a weighted average based on how long each lineup played.