Hi everyone, i recently did a regression on what input variables are heavily correlated to wins / losses this season. From the regression (r-square of 82%); these 5 variables emerged as being heavily tied to our wins and losses with p-values < 5% in decreasing order of coefficients:
* Reb % – highest coefficient
* Opponent Turnover %
* Turnover % — so, win the turnover battles
* True shooting % – somewhat obvious: shoot well and we win
* Opponent points in the paint
I, then, did a deep dive to see which of our starters impact the outcome of the games the most:
**Reb %:** Regressions didn’t show any starters that had p-value <5%… probably because Jaylin Williams is leading our team in reb %
**TS %:** three players shooting are x-factors in defining wins / losses… SGA (highest coefficient by far), Josh Giddey & Chet Holmgren (both had similar coefficients)
Hopefully the stats show that Giddey isn’t the focal blame point in our losses. I do think he’s had a down year and his “mistakes” are louder because his shooting is left wide open. His efficiency on his push shots / floaters are down and regressions showed that his performance around the rim actually impacts our wins / losses. A lot of this is due to what Andrew Schlect on Down To Dunk mentioned on his change of roles this season where he’s much less of a ball handler and more of a spot-up guy. There’s probably opportunities on optimizing lineups for the remainder of the season to develop & boost his confidence a bit more.
Check out the full article here:
by DillyDillyKang
8 Comments
Nice work. A lot of people simply won’t accept this because they have had their views tainted bc of the “allegations”. I don’t even know if you can call them that tho
What are you trying to prove lol these numbers don’t mean a whole lot
Nice analysis.
I think the difference between two recent games (Dallas and Magic) is interesting.
In the Magic game, we got out-rebounded 48-30 and were fairly even on turnovers and other metrics. Yet we dominated them because our offense was uber efficient. I waited for the âwe donât rebound well enoughâ cries. Silenceâ- crickets because we won.
The Magic have a much better defense than the Mavs, but somehow the Mavs seemed to muck up our offense to a great degree and our defense was not good.
Some of this is match-ups and momentum (which I realize are probably not measurable). We didnât really seem ready to play against Dallas, but came out very efficiently against the Magic and Kings.
In other words, after the poor showing against the Mavs, I expected us to lose against the Kings for the 3rd time. Basketball doesnât work like that and I should know better.
Do you have any comments on how âintangiblesâ affect the game or is this not a thing from a statistical perspective?
Giddey isn’t the only reason we lose games when we lose, but he’s a major factor and it’s becoming harder and harder to deny that.
We get away with a lot of his shortcomings during the games we win too, cause the other players around him are much more efficient.
There’s also the Dort factor, he’s had a stellar season from three thus far. If his shooting numbers trend down, it’ll be even harder to justify Giddey+Dort minutes. Dort at least has defensive value so the minutes will remain more consistent with him. Giddey not having that will be a major problem come playoff time.
This feels like the tussle between advanced analytics and “my eyes”. I think though the common point is shared – Josh is not in his “best” position. This is on Mark to fix, if it can be fixed.
Everything else? That’s where being a fan shines…as many of us have opinions, and many of us dislike the other folks opinions in favor of our own đ
This actually foes with the eye test for me as I have said that when we lose its not really Josh’s fault that we lose. Another variable you didn’t have which I think is significant is the minutes played and when those minutes are played. Someone playing 25 minutes isn’t going to have the same effect as someone playing 35 minutes spread throughout the game.
The shooting from three isnt the problem because even when we have nothing but shooters out there they still double team and blitz Shai. Dort is shooting 43% from the year from three but they still leave him wide open, because a 70% shot at the rim is more efficient than a 43% percent shot from three.
So what would make Giddey fit more is him attacking thru cuts and finishing in the paint better. This would take advantage of leaving him alone at the perimeter. Teams don’t leave player’s who can finish in the paint alone on the perimeter. This gives them a head of steam and puts the defense out of position.
Only a matter of time before two certain Reddit users jump in here with the worst opinions. đ
(Good analysis tbh)
This is really interesting work! And I highly doubt that remaining 18% has anything to do with Giddey as an individual, so you’ve effectively answered the question that you set out to.
I would be curious to see the difference between live-ball turnovers (which we force a lot of) versus side outs and other non-transition opportunities.
I think our output turning defense to offense is the biggest strength of the team and I’d be willing to bet that there would be a tangible difference between a night where we forced more live-ball turnovers (i.e. if there two nights where the thunder force 12 turnovers, and have 12 of their own, I know that having more live ball turnovers forced than the opponent would positively correlate with OKC win%, but I’m curious what the specific factor would be)
It’s been awhile since I did this kind of work and it would be a pain to determine manually which turnovers are live/dead, but I’d assume live ball would be around 1.5x more valuable.
That aside, really really interesting stuff! Great work