So, last week we saw which Toronto Raptors lineups have performed well thus far this year, which ones have been all right, and which ones we should never see again. But is there a way to construct a rotation such that (a) we stick mostly to the most effective lineups and (b) the rotation has some flexibility in it for foul trouble or matchups?
Let's find out. First, let's take a look at a typical Dwane Casey rotation (gathered by poring through a few recent box scores). Here's what I managed to piece together, primarily using this game against the Pacers.
There you see each lineup used in the game, with rough substitution times shown, and the year-long net rating for each group. Any group that has been used less than 30 minutes total is given a value of 0. It is interesting to note that at this late stage in the season, Casey used three different lineups that haven't seen 30 minutes of time together yet.
Also note that a minute-weighted average of the net ratings of the lineups gives an expected performance of -1.3. That's losing by 1 point to an average team. You might expect a win against a terrible team, but on the whole not seemingly a great strategy.
Now, let's see what we can do if we stick to lineups that have worked. I'm going to try to stick almost exactly to the patterns above - times at which subs are made, how many bench or starting players are on the court at any time, with a few exceptions that I will bold. Exactly which players start is different, and which players get subbed on and off, but the overall pattern is very close.
Primarily bench units are used at the same time breaks as in Casey's rotation, which means we shouldn't have to worry about relying on data against bench players if the group plays against starters. You'll note that Vasquez in particular is the last sub off the bench - he is one example, as we highlighted last time, of a guy who is much more effective against bench players.
Let's take a look at the minutes load and breakdown for each player in my rotation, to make sure it is practical.
Kyle Lowry: First 9 minutes of 1Q, last 7 minutes of 2Q, first 9 minutes of 3Q, last 8 minutes of 4Q. 33 total minutes.
DeMar DeRozan: First 8 minutes of 1Q, last 7 minutes of 2Q, first 8 minutes of 3Q, last 8 minutes of 4Q. 31 total minutes.
James Johnson: First 11 minutes of 1Q, 2 minutes in middle of 2Q, first 7 minutes of 3Q, 2 minutes in middle of 4Q. 22 total minutes.
Amir Johnson: First 8 minutes of 1Q, last 7 minutes of 2Q, first 9 minutes of 3Q, last 8 minutes of 4Q. 32 minutes total.
Jonas Valanciunas: First 8 minutes of 1Q, last 5 minutes of 2Q, first 8 minutes of 3Q, last 7 minutes of 4Q. 28 total minutes.
Greivis Vasquez: Last 3 minutes of 1Q, first 5 minutes of 2Q, last 3 minutes of 3Q, first 4 minutes of 4Q. 15 minutes total.
Lou Williams: Last 4 minutes of 1Q, first 5 minutes and last 3 minutes of 2Q, last 5 minutes of 3Q, first 4 minutes and last 5 minutes of 4Q. 26 total minutes.
Terrence Ross: Last 1 minute of 1Q, first 7 minutes of 2Q, last 3 minutes of 3Q, first 5 minutes of 4Q. 16 total minutes.
Patrick Patterson: Last 4 minutes of 1Q, first 7 minutes of 2Q, last 4 minutes of 3Q, first 5 minutes of 4Q. 20 total minutes.
Tyler Hansbrough: Last 4 minutes of 1Q, first 5 minutes of 2Q, last 3 minutes of 3Q, first 4 minutes of 4Q. 16 total minutes.
Maybe a little heavy for Amir, but the long and short of it is that he, even in his reduced state, is a key cog to this team, especially defensively. Otherwise those minutes look good, right?
Best part about the low minutes all around is that any foul trouble can be easily absorbed - just shift some of those lineups a little earlier or later to eat some minutes. There are also some other variants on lineups that could fill in due to foul trouble - such as Lowry, DD, Ross, PP and TH, a +30 lineup on the year - that simply didn't fit into the rotation following Casey's patterns.
Actually, do you want to hear the real best part?
Remember how Casey's example rotation had a projected performance of -1.3? This rotation projects a performance of +14.5 points per 100 possessions. Yeah. Not that they will blow out every team with this rotation, but if they stick to a formula like this, using only lineups that have been successful, the Raptors' chances of winning, or even blowing out teams, seem to be much better.
What do you guys think?
(All stats from NBA.com.)