Part 2: Getting the most out of NBA Lineups - Miami Heat

[watto84]

Part 2 of NBA lineup analysis with machine learning. I look at the Miami Heat with insights on how to get Goran Dragic going.

Check it out: http://www.zigzaganalytics.com/home/par ... miami-heat

[Crow]

3 man raw plus minus shows trios of Dragic with Richardson and Whiteside and Richardson and
J Johnson doing pretty well, best of big minute trios. But these four aren't in a quad very often. The one top minute 5 man lineup with all 4 did fabulous at plus 32 per 100 possessions but it was pretty low minutes. I'd try that 5-10 minutes per game. What other insights can you give on that quad? Is there a reason machine learning didn't focus priority light on it already? Why hasn't Spo focused priority attention on it already? Two minutes looking at the data and my eye went right to this above everything else with Dragic.

[watto84]

The last time I ran the heat team specific scout which was on November 19 (a while back now). I looked at which lineups for the heat had hit their targets as per the algorithm. 5 lineups had hit it...3 had olynyk at the 5 and only 1 with whiteside. All of these lineups were a positive +/-. It was at this stage I was a big proponent of getting olynyk more minutes (early on his minutes were a little inconsistent) but the heat were just so much better with him out there. Whiteside was/has been poor in general with a combo of knee issues but also poor body language/sulky as he adjusts to the heat taking more and more 3s.

The best whiteside lineup at the time in terms of hitting the targets was whiteside, Richardson, waiters, t.johnson, dragic

[Crow]

Have Whiteside issues, still can test / use / maybe benefit from Dragic, Richardson, J Johnson lineups. By the surface plus minus seems more worth pursuing than most or all of the other variations getting run. Johnson back available last game, zero use of the trio.

[Crow]

The Dragic trios mentioned were among his best big minute trios last season. The quad was his best quad. The biggest minute 5 man lineup with the quad was again his best big minute lineup by a large measure. And yet they were all used lightly to very lightly. Too lightly. Maybe there is a reason... but I don't see it. And I challenge this underutilization.

[Mike G]

The rest of the question might be: how do you make up for more minutes under this one lineup? You go with 2, 1, or zero starters for longer stretches. Dink lineups become major experiments.

[Crow]

Yes there are minute limits and ultimately grand rotation strategy lineup tradeoffs but there is so many dink lineups / minutes that you don't get to really tough choices for quite awhile. They could bump up the minutes for 1-3 lineups easily without creating death bench lineup results. And it is far easier to maximize a trio than a specific 5 man lineup. Maximizing 3 out of 4 on court, even easier.