I’m back with a special request from the APBR.org (Basketball Nerd Forum) to have a look at what the machine learning scout can tell us about the Milwaukee Bucks.
As I write this post, the Milwaukee Bucks have just lost to the Celtics in game 5 and are now down 3-2 in the 1st round series.
In this post, the machine learning scout dives into the detail of the Bucks most used lineups and starts to answer:
So let’s get straight too it…..
I mentioned the turnover struggles of the “Poor Giannis Line-up” that has Maker starting in place of Henson. I took a deeper look at this with game tape to see if this was just a statistical anomaly (basically bad luck when Maker was on the court) or was there something more to it.
One of the downfalls of analytics is with stat geeks not verifying numbers through the eye test. Although we would love to argue the numbers are never wrong, sometimes they are and tell a very different story to what is happening, especially with small sample sizes. That is why I wanted to really look at what was happening with Maker and not just assuming it was his fault for the turnovers.
This lineup with Maker is committing 2.8 turnovers per 4.5 minutes on the court and in total it has committed 28 turnovers (small sample). In 3 games alone this lineup committed 16 turnovers, so I went and had a look at each one in these 3 games.
What I found was 20% of these turnovers were due to trying to throw lobs at the ring to Maker. In most cases Maker had either set a pick and rolled well to the basket, back cut or got out on the break ahead of his defender.
Unfortunately the bucks are trying to bring Maker and his athleticism into the game with a little too much eagerness and are throwing lobs either in traffic or the pass is just not well executed.
The bucks are trying to use Maker much like the Rockets do with Clint Capela around the basket in these situations, but are just not executing cleanly. It’s not to say it’s a bad look, they just need to show some poise and perhaps work on executing at practice a little better before running it in a game.
Let’s take a look at the examples…..
So what did we learn from the scout and what can the bucks do moving forward?
Firstly their True Shooting percentage is excellent (6th in the league) through the looks they are getting in the paint, and basically non-existent 3pt shooting. They really need to look at getting Tony Snell some more looks from 3 (shoots 40% from 3 but only gets 3.6 attempts a game) and finding another perimeter sniper in the off-season.
The looks they are getting in the paint will only increase if teams see them as a threat from 3. Right now defenses can pack the paint as the bucks are very one dimensional.
Rebounding for the 2 most used lineups was also a big issue (particularly with Bledsoe on the court), the Bucks have far too many terrible rebounding stretches that is really hurting them. As a team they rank 3rd last in the league in Rebounding %.
Bledsoe I wouldn’t say is killing them, but it’s a problem when there is next to know drop-off when Brogdon or Dellavedova are in the game for him (or their line-ups actually perform better) and as I mentioned that one Dellavedova lineup has been very good in limited play.
When Bledsoe isn’t on the court, the Bucks are getting to the free throw line more often as a unit and rebounding more consistently.
Finding more 3’s (which will open up the lane for even more paint points), rebounding, potentially moving on from Bledsoe and showing more poise with how they involve Maker in the offense are the main points we have learned from the machine learning scout this time around.