APBRmetrics

This is just a brief start on the topic. I am not sure how fast or far I'll go with this. Might go far and with help.

But for today:

The top 6 lineups for the Warriors in playoffs overall included only 2 of their 5 most used regular season lineups. They completely eliminated the only negative performing one of these and removed the only one with Zaza, the biggest minutes one. Another lineup was virtually eliminated. The 4 lineups added to the 6 most used playoff lineups were each hardly used at all in the regular season. If you planned how to play the Warriors in the playoffs based on what lineups they played in the regular season, sorry. They changed pretty dramatically. The top 10 lineups got about 60% of the total playoff minutes and provided almost 80% of the net points.


The top 5 lineups for the Rockets in playoffs overall included only 2 of their 5 most used regular season lineups. They barely played 2 of the most used regular season lineups and I think completely eliminated another. They too were quite different, at least in terms of top lineups. The top 10 lineups got about 64% of the total playoff minutes and provided 114 net points. All the smaller lineups combined lost by 83.

Pretty similar levels of lneup concentration and clearly the most used lineups drove the success of each. Dinking around provided a modest bonus gain for the Warriors but was bad news for the Rockets.

Next increment: Playoff lineup data for Warriors against Rockets and vice versa. Then comparison of the overall regular season and playoff lineup data to the conference finals data. And maybe the regular season matchup's lineup data vs. playoffs.

I sorta gave these teams a bit of credit for "surprising" opponents in the playoffs above. But they also surprised themselves with the lack of much testing knowledge from the regular season to inform playoff lineup decision-making. Rockets tested 1 lineups used in playoffs over 200 minutes. The Warriors 1 over 100 minutes. So for 99 plus % of the lineups used in the playoffs the coaches were playing lineups tested less than 100 minutes and most far, far less. I guess these elite coaches & organizations don't need hardly any actual 5 man data to make lineup decisions and be light years ahead. Smh.

Of the 5 most used Warriors lineups used against the Rockets in the regular season (all used modestly), only one was used in playoffs for any notable time and was neutral.

Of the 5 most used Rockets lineups used against the Warriors in the regular season, only one was used in playoffs for any notable time and was mildly negative.

They basically lined up and matched up completely fresh with zero meaningful direct play lineup information for themselves or opponent.

The Warriors went with 6 lineups in playoffs against Rockets for about 2/3rds of total time and 5 of the 6 were positive. 4 strongly. Rockets also used 6 lineups for 2/3rds of time but 5 0f 6 were negative. 4 strongly negative. So in freelance wing it vs. wing it, Kerr and the Warrior talent prevailed.

Half of the Rockets'most used lineups won on efg%, 3pt fg% and ftas but they must on gotten smashed in other facets of the game.

I was going look at lineup management game to game and lineup by lineup. I still might but I might shift to the eastern finals first to see if there was any higher degree of prior information informing the lineup management there.

It may be necessary to go to lineup clusters to get patterns to analyze.

Eastern conference finals- Celtics.

Most used lineup then, little tested in regular season but very good. Given the most minutes and it sucked to -12 per 100 possessions. 2nd most used, fairly well rated in regular season and fantastic then. In playoffs, mildly positive. 3rd most used, very lightly tested. Did fine but only half as good as regular season. 4th most used, lightly tested but great in regular season. In playoffs did way better. 5th most used, tested less than 10 minutes, if at all. Does very poorly in playoffs.

So only 1 of the most used 5 lineups in this series was tested over 100 minutes and barely. 3 way underperformed in playoffs compared to regular season. One stunk with no testing. One did well despite limited testing.

That isn't what you'd want to see in terms of testing or playoff results. That is poor testing, though injuries gives some level of excuse. But as for selection / implementation, it was mediocre at best. Not enough.

I'm skipping the Cavs, at least for now.