Analysis Two Man Wowy Combination Tracker

[scott]

Yesterday in the Vassell thread I posted some data about how he went from making virtually every lineup combination worse last year to making virtually every lineup combination better. This inspired me to build a matrix for every two-man combination among our top 11 players (sorry DJG and CB fans... this is actually quite a bit of work to put together so I cut it off at these 11 guys).

I plan on updating this throughout the season, maybe in 15 game increments. There are notes on how to red these charts on the left but I'll explain a little more below.

One thing to note is that these numbers *exclude* low leverage situations, which are possessions where the outcome of the possession has very little impact on the outcome of the game (you can read more about this here) in an effort make the data more meaningful. So if you try to match these numbers up with on/off numbers that include low leverage possessions, you'll get a different result. As an example of the impact this has, when Wemby is on the court we have a 9.6 NetRtg in All Leverage situations, but only a 7.4 NetRtg when we exclude Low Leverage situations.

This data is from databallr.com, which I highly recommend. They have a lot of great tools available, and the WOWY Lineups explorer is a lot of fun.

How to read these matrixes:

• The Matrix on the top tells you how the player on the Top Improves the NetRtg of the team with the player on the left on the court.
  • This is calculated by: [Net Rating When Both Players are On] - [Net Rating When Player on the Left is On but the Player on the Top is Off]
  • A practical interpretation would be: take a look at Barnes on the left intersected with Castle on Top. This means that Lineups with Barnes and Castle on the court are +9.9 versus lineups with only Barnes on the court.
  • The cells where a player intersects with himself is the player's On/Off (Net Rating when On minus Net Rating when Off)
• The Matrix on the bottom is the NetRating when the two players on the court together.
  • The cell where the player intersects with himself is the team's NetRtg when the player is On.

What stands out to me this early on in the season:

• He's only played 6 games (5.5 really) but Dylan's impact is HUGE. Can't wait to see him back
• As previously noted, Devin is making lineups better!
• Champ and Barnes have similar consistently positive impacts and to me those three guys highlight what we all already know: that spacing is extremely important and makes our team better
• Despite everyone being please with Keldon's play this year, the team isn't as good in his lineups. With that said, I don't blame that on Keldon... I think that's a function of Keldon primarily playing with bench lineups
• Fox's impact is definitely a bit shakier to start, but he's also played half his games this season without Wemby, Castle and Harper
• Ignore that huge negative number for Korndog and Olynyk... that's only 1 minute of action. Seems like Mitch is wisely avoiding playing them together
• Sochan's numbers match the eye test.

What stands out to you?

I definitely encourage everyone to play around in databallr, as you can dive in well beyond just NetRtg. Here's a sample image of what you can see (and you don't have to limit combos to just 2 players either)

 

[LeBowen]

Great work, with some obvious skewed data due to small sample sizes.
As in Wemby/Barnes having a negative net rating because Barnes' hot streak coincided with Wemby's injury.

The amount of these data analysis topics you made over the past year or so would make you a top tier analyst at ESPN.
As I said when this place became a thing, once the general public starts noticing it, media outlets will steal a lot of our stuff on regular basis.

 

[ambchang]

The great news is the number of positive matchups we have this year compared to last year. Other than keldon and sochan most players are mostly +ve

Sochan though, this year he took the Vassell spot of being negative in every single situation. Just shocked at how badly he regressed

 

[CGD]

Damn, if im understanding this right, those Harper numbers are nuts.

What do we think this tells us about our ideal 5 (every season). Is it Harper, Devin, Barnes, Wemby and maybe Champ?

 

[scott]

Damn, if im understanding this right, those Harper numbers are nuts.

What do we think this tells us about our ideal 5 (every season). Is it Harper, Devin, Barnes, Wemby and maybe Champ?

I wouldn't use two-man combination data t try to read too much into what an ideal 5 is, but it does give us some indication on what pairs fit and which shouldn't share the court. For example, last season there was very clear evidence that the combination of Vassell and Keldon together was very bad (which isn't necessarily the case this year).

One thing I do like to look for is combos where both sides are +ve. So for example, Harper and Wemby... the team is good with either of them on the court by themselves, but Wemby lineups get better with Harper AND Harper lineups get better with Wemby. That isn't always the case, but to me that signals a very effective pairing.

I have another thread I update every 10 games with 5 man lineup combination statistics, so that will get updated soon. Of course, we still don't have a ton of good data because we've had so many injuries.

 

[brazil]

great stuff scott.. dat Dylan Harper dude seems to know how to play ball

 

[scott]

Who loves an overreaction based on an extremely small sample size?

Fox + Dylan lines have posted a +166.7 NetRtg so far this season (200.0 ORTG, 33.3 DRTG).

(Note: 2 minutes played)

 

[scott]

Because I like to continue to show off the cool things you can do with databallr.com (which is free btw). Check out these stat-line shifts between Fox and Devin, which has proven a combo that significantly improves both players NetRtg (here expressed by +/- per 75 as opposed to per 100). Both players see a significant bump in efficiency when on the court with one another, which to me is an indication of much having another high level scoring option on the court opens things up.

Can't wait to see when Wemby and Castle are back... we should be able to have at least 2 and probably 3 of Wemby/Fox/Castle/Harper/Devin on the floor at all times.

 

[Exstatic]

Not as big a believer in +/- as some, but the synergy of TS% and 3G% when they share the floor is eye-popping.

 

[The Truth #6]

Who loves an overreaction based on an extremely small sample size?

Fox + Dylan lines have posted a +166.7 NetRtg so far this season (200.0 ORTG, 33.3 DRTG).

(Note: 2 minutes played)

Fire up the River Boats, the title is in the bag!

 

[ginobilized]

What a gift, thanks scott!
Harper really has some eye-popping numbers. I'm surprised by Vassell/Fox numbers because they don't seem to have chemistry together but, clearly, each benefits from the other being on the floor.

 

[ambchang]

Who loves an overreaction based on an extremely small sample size?

Fox + Dylan lines have posted a +166.7 NetRtg so far this season (200.0 ORTG, 33.3 DRTG).

(Note: 2 minutes played)

So we should roll them out for 48 mins a game 82 games of the year and all 16 (won’t need more than that) in the playoffs? Play at a hyper pace and we can win 300 to 50 every game dude.

 

[scott]

Will the league let us play 2 on 5 is the question

 

[heyheymymy]

Olynyk being negative in almost every combo probably has to do with the injury and return when wemby etc are out injured but going into the season I thought he would have much more instantaneous chemistry with the rest of the guys.