Top 30 most improved Playoff Lineups

Posted on 09/05/2010 by


Yesterday’s Post today.

Reader Alex ask in thursday’s post (A Half Baked Predictive Look at the 2010 NBA Playoffs):

“Hey Arturo – Pretty cool. It looks like Boston was pretty unusual in that their top 6 guys (by minutes) in the regular season weren’t even the guys they played in the playoffs, since they went from 6th to 2nd. There can’t be many teams who do that, excluding injuries, right?”

That’s a leading question if I ever heard one. So I dug up the playoff data and took  a look.

I’ll be using the WP48 for the top 6 players for each team based on minutes played. For reference, based on the WP48 for the top 6 for the regular season we would expect:

So you need your top 6 to be above average (.130) to be better than .500. And around 160 and up puts you in championship contender mode.

First off let’s look at the 2010 data (for perspective):

So based on the actual playoff lineups and regular season performance Chicago, Boston and Cleveland were the most improved and OKC, Orlando got worse due to lineup changes and Utah & Milwaukee lost key personnel. Note that the range remained within plus or minus half a  game over a seven game series (except for the Deer losing their starting center). Boston jumped from an also ran to a championship contender.

So was this a historic occurrence? Let’s look at the thirty biggest jumps in expected performance from the regular season to the playoffs.

Actually the answer is no. Historically there’s always someone who makes a significant leap based on their playoff lineup. The Houston 1995 Team (the Drexler trade team) was a sixth seed that was actually much better.  The top 5 were good teams with bad luck (88 Chicago lost to the Bad Boys and the Jordan Rules, 96 Miami got swept by the best team of all time, 94 Utah ran into Hakeem, 87 Milwaukee pushed the Celtics to the limit, 93 Portland should not have lost to an ok Spurs team).

On average teams improve slightly (about .02 over a seven game series) going into the playoffs but if I look at the standard deviation for the delta I get .46 games or about half a game for that series (or about 6 games for a full season). So I can reasonably expect to see variation within the plus/minus two standard deviation limit every playoffs. In English, that means someone is likely to improve close to a game and someone else drop off close to a game based on their playoff lineup.

In my next post, I will talk about my magical unicorn.

Posted in: Uncategorized