Last Friday, Dan Hackett wrote an excellent forum post trying to quantify the impact of certain Raptors on their teammates using a method known as WOWY analysis. We asked to run the piece as a guest feature and he kindly obliged, with some tweaks from his original post.

I’ve been pondering how to capture defensive impact (and team offense impact) better through stats, and stumbled on a neat stat used in NHL circles called With Or Without You’s (WOWYs). It is fairly simple – it just separates a player’s stats into how the team and individual performs when he is playing with a particular teammate (or lineup) versus not with that teammate (or lineup).

Now, as far as I can find, there’s no readily available version online for basketball, but I figure I can cobble something together from basketball-reference’s On/Off splits and lineup finder. So I thought I’d try to take a look at the Raps and see what impact our players have on their teammates. I won’t be able to see individual stats based on lineups (which is what WOWY really is, and I hope to find a way to look at that too), but the way the team performs when the players are on the court together is really what matters, so I’ll look at that. This will be similar to the approach taken with adjusted plus-minus and RAPM, except I’ll be targeting the impact on lineups with specific players.

What I’m interested in is how our pieces fit together – not so much the Gray’s and Acy’s of the world, but how our primary and secondary pieces either help or hinder the performance of our core pieces.

First, a list of the players and their impact on and off the court last year in general. I’ve also included the most used guys who are off the team now, as well as the new guys.

Player – ORTG On – DRTG On – ORTG Off – DRTG Off – Net Impact

DD: 105.3 / 107.3 / 109.4 / 109.5 / -1.9
Amir: 109.7 / 104.9 / 101.5 / 111.9 / +15.2
Lowry: 107.9 / 107.3 / 104.5 / 108.4 / +4.5
JV: 104.0 / 107.7 / 107.6 / 107.9 / -3.4
Ross: 106.4 / 109.2 / 106.2 / 107.2 / -1.9
Gay: 106.7 / 104.0 / 106.1 / 109.1 / +6.0
Fields: 108.4 / 108.4 / 105.5 / 107.6 / +2.1
Gray: 98.8 / 107.3 / 107.4 / 107.9 / -7.9
Acy: 113.8 / 99.8 / 105.6 / 108.6 / +16.9

Hansbrough: 100.6 / 100.4 / 107.0 / 99.6 / -7.1
Augustin: 100.3 / 102.0 / 106.8 / 99.0 / -9.6
Daye (DET/MEM): 107.4 / 105.5 / 104.4 / 104.5 / +1.9
Novak: 112.2 / 106.5 / 110.6 / 107.0 / +2.1

AA: 106.7 / 110.3 / 106.0 / 106.4 / -3.1
ED: 104.5 / 109.2 / 106.9 / 107.4 / -4.2
Bargs: 104.6 / 110.8 / 106.8 / 106.8 / -6.2
JL3: 105.4 / 111.5 / 106.5 / 106.9 / -5.7

So, yeah. I’m pretty glad we got rid of AA, Bargs and JL3. And as much as Eddy was a nice young piece to have, he wasn’t exactly setting the world on fire in terms of impact. And those new additions don’t exactly inspire confidence (Daye was awesome in DET and terrible in MEM, balancing out to that fairly even score you see above). Now, these are simple plus-minus stats, so mean little by themselves and are prone to problematic conclusions, but that’s why I’m going to look at some lineup data next.

WOWYs:

So, what impact does each of these players have on their teammates? Let’s use an example to go through in detail. I’m going to use Fields as an example, as he is touted by many (myself included) as an impact guy in spite of his individual numbers.

Player – ORTG with Fields – DRTG with Fields – Net RTG with Fields – MP with Fields
DD: 105.3 / 107.3 / -3.9 / 702
Amir: 109.7 / 104.9 / +4.9 / 516
Lowry: 107.9 / 107.3 / +4.1 / 475
JV: 104.0 / 107.7 / -3.3 / 308
Gay: 106.7 / 104.0 / +3.8 / 162

From here, we can extrapolate back from their season total numbers to get a “Without Fields” number, and find the difference between playing with Fields and without him. Keep in mind that a positive value in DRTG change is a bad thing.

Player – ORTG impact from Fields – DRTG impact from Fields – Net RTG impact from Fields
DD: +0.5 / +3.0 / -2.5
Amir: +1.3 / +1.2 / +0.1
Lowry: +5.1 / +0.5 / +4.6
JV: +3.0 / +2.5 / +0.5
Gay: +0.8 / -0.5 / +1.3

So you can see that generally speaking, Fields playing in the same lineup as DD is a bad idea – which supports the backup SG role for Fields, letting him play some with Gay while DD sits, or play SF with Ross at SG (as a note, Ross had a +9.7 WOWY with Fields). In fact, of all the players from last year that were retained, only DD had a negative Fields WOWY. The interesting thing of note here is that I expected Fields to hurt the team offensively, and help enough defensively to make up for it. However, the opposite seems to be true in most cases. He’s a small negative defensively, and a more significant positive offensively. Note that he has good impacts particularly on Gay and Lowry – two aggressive scorers that draw a lot of attention, and can find a guy on a back door cut. Also note that Gay improves offensively and defensively when playing with Fields.

So, here’s how the rest shape up. I’m looking at how our core fits together, plus how some “core hopefuls” such as Fields complement them. So I’ll be doing DD, Gay, Lowry, JV, Amir, Fields and Ross (not Acy or Gray due to small sample size and I don’t care, respectively). Plus I’ll throw Bargnani on there just to show how much of a detriment he was while he was on the floor (if the first set of stats above wasn’t clear enough).

Player – ORTG impact from DD – DRTG impact from DD – Net RTG impact from DD
Amir: -2.9 / -6.6 / +3.7
Lowry: -11.0 / -6.5 / -4.5
JV: -3.3 / -4.0 / +0.7
Gay: +0.5 / -1.5 / +2.1

Does not do well with Lowry. Otherwise is at least passable. Nothing big in either direction here.

Player – ORTG impact from Gay – DRTG impact from Gay – Net RTG impact from Gay
DD: +2.2 / -5.2 / +7.4
Amir: -1.2 / -5.6 / +4.3
Lowry: -2.3 / -8.3 / +6.0
JV: +8.7 / -9.4 / +18.0

Gay definitely had a big positive impact here.

Player – ORTG impact from Lowry – DRTG impact from Lowry – Net RTG impact from Lowry
DD: +2.1 / -2.3 / +4.4
Amir: +5.8 / -1.9 / +7.7
JV: +0.3 / -5.0 / +5.3
Gay: -0.4 / -5.9 / +5.5

Yep, Lowry fits well with this core. Solid PG. But a good point was brought up – was he better in the 2nd half when he settled down and had the Jose-shaped monkey off his back? Let’s see.

Player – ORTG impact from KL 1st Half – DRTG impact from KL 1st Half – Net RTG impact from KL 1st Half
DD: +1.1 / +4.2 / -3.1
Amir: +11.3 / +6.0 / +5.3
JV: -1.9 / +2.1 / -4.0

Player – ORTG impact from KL 2nd Half – DRTG impact from KL 2nd Half – Net RTG impact from KL 2nd Half
DD: +2.8 / -8.3 / +11.1
Amir: +5.0 / -6.6 / +11.6
JV: -1.3 / -10.2 / +8.9

Yep. Yep yep yep. Good observation. Now for Jonas.

Player – ORTG impact from JV – DRTG impact from JV – Net RTG impact from JV
DD: -3.1 / -0.3 / -2.7
Amir: -1.1 / -6.4 / +5.3
Lowry: -7.4 / -2.5 / -4.8
Gay: +5.2 / -4.0 / +9.2

Big swings here, and I expect a lot has to do with his progress from the first half of the season to the second. I think I’ll filter by half season.

*** NOTE: These have now been updated to be more accurate – previous version was an estimate based on the entire season, these are now based on partial season data available from nbaWOWY.com. So slightly different, but still a huge jump in performance from JV.

Player – ORTG impact from JV 1st Half – DRTG impact from JV 1st Half – Net RTG impact from JV 1st Half
DD: -5.9 / +2.9 / -8.8
Amir: -19.0 / -9.1 / -9.9
Lowry: -17.1 / +3.5 / -20.6

Player – ORTG impact from JV 2nd Half – DRTG impact from JV 2nd Half – Net RTG impact from JV 2nd Half
DD: -0.7 / -2.8 / +2.0
Amir: +1.8 / -5.6 / +7.4
Lowry: -1.3 / -5.8 / +4.5

Yep. What a difference. It helped that he played a lot more with Amir in the 2nd half, though that would definitely not account for these big WOWY changes. Why would Casey only play JV and Amir together a total of 24 minutes in the entire first half of the season? Insane. But still, who sees improvement like that mid season? Crazy pants. If he makes half the same progress in his game between the end of last season and the end of this coming one, we’ve got ourselves a player, guys.

Player – ORTG impact from Amir – DRTG impact from Amir – Net RTG impact from Amir
DD: +8.9 / -9.7 / +18.6
Lowry: +11.6 / -8.5 / +20.1
JV: +7.0 / -11.5 / +18.5
Gay: +7.9 / -9.6 / +17.5

Amir is hilariously awesome. He makes everyone around him SO MUCH BETTER. I might have to revisit my stance that we don’t need to move Amir to tank – if we want those losses to pile up, he has to be the first to go. And if we want to be good, we’ve got to have him on the court as much as possible. Just because Amir is so awesome, I’ll list his impact on everyone else as well.

Ross: +15.6
Gray: +7.0
Acy: +3.4
Fields: +9.8

Seriously. The only player on the roster that had a negative plus-minus while on the court with Amir was Gray (I include all the scraps and garbage like Pietrus I’ve left out above), and he’s just a lost cause.

Player – ORTG impact from Ross – DRTG impact from Ross – Net RTG impact from Ross
DD: -5.2 / +2.4 / -7.6
Amir: +1.3 / +2.4 / -1.1
Lowry: +3.9 / +3.9 / 0.0
JV: -11.6 / +5.5 / -17.1
Gay: -1.5 / -3.0 / +1.4

Ross and Valanciunas, as you’d expect from two rooks, just proved too raw to be on the court together. Ross does not match up well with DD either. His slight positive impact on Gay suggests either a good fit or improvement towards the end of the year.

And the piece de resistance… And the single biggest reason to think the team improves this year over last (maybe besides Jonas’ insane improvement).

Player – ORTG impact from Bargs – DRTG impact from Bargs – Net RTG impact from Bargs
DD: -0.7 / +3.5 / -4.2
Amir: +0.8 / +4.6 / -3.8
Lowry: -5.5 / +4.1 / -9.5
JV: -1.5 / +5.0 / -6.5
Gay: +0.4 / +7.8 / -7.4

Fun.

Anyway, I hope you guys get something from this. Feel free to critique and question – I’ve got a whole spreadsheet set up and can make changes where you think it necessary or informative.

This has been a guest post from Dan Hackett, and we’re hoping he’s got some more in store.

  • c_bcm

    very intersting. I like taking these things at face value. Well done sir…Fear Amir!

  • alucart999

    The WOWY analysis seems to also depend on who your replacement is. So for example, if Amir was frequently being replaced by Bargnani (which he wasn’t last year, since Bargnani was injured for much of the season), then that would go some way towards explaining his relatively positive impact on everyone else.

    • DanH

      True. However, you’d expect to see mirrored impacts in that case – as you can see from Bargnani’s WOWY, he can only potentially be responsible for a small portion of Amir’s WOWY numbers.

  • Rebuilding

    I’m confused on something. Why does the advanced summary tab for ORTG differ from the ON/OFF Ortg?
    Amir Advanced tab http://www.basketball-reference.com/players/j/johnsam01.html#all_advanced
    Amir ON/OFF http://www.basketball-reference.com/players/j/johnsam01/on-off/2013/
    Is ON/OFF pace adjusted?

    Second where do you get ON/OFF DRTG?

    • DanH

      Basketball-reference defines a player’s ORTG and DRTG as the values they actually score and allow themselves – ie if Amir scores a basket, it counts toward his points and possessions. If DeMar scores a basket, it does not impact Amir’s advanced ORTG.

      But if DeMar scores a basket while Amir is on the floor, it does impact his On-court ORTG. So, in short, the advanced page is for individual stats, while the ON/OFF page is for team stats.

      • Rebuilding

        Got that. So where do you get On and Off Dratings for player?

        • DanH

          Same page as the ORTG’s – just look at opposition ORTG.

          • Rebuilding

            Why is opponents Ortg = Amir’s team Drtg?

            • DanH

              Team DRTG = Opp Team ORTG by definition.

  • elkabong

    awesome post and you are to be commended on the amount of time and effort it must have taken DanH …… basketball like no other sport relies on the components of the 5 man unit for success,even the best player on the planet needed some help to finally win a title eh!

    looking at last season there seemed to be one worthy unit and your numbers seem to totally back that up. this line in your write up helps to sum up my feelings on that matter “Why would Casey only play JV and Amir together a total of 24 minutes in the entire first half of the season? Insane.” from 3 guard small ball toooooo many times even before the big trade and then continuing that by sticking Rudy at the 4 and benching Val for an incredible amount of 4th quarters none of it made sense to me and bordered on “insane” a good portion of the time from my lazy boy.

    if you look at Demar’s 5 man unit numbers as the guy who was the starter for the entire 82 game season they go 343 – 126 – 85 – 79 – 69 – 67 etc etc etc so how can you accomplish anything as a team when you are in constant flux? how do you really analyze anything of any of our players from this type of data base? how did Casey ever talk himself into keeping his job?

    • Reed

      Casey has to go if the team wants to move forward. The Raps say they use advanced stats yet it doesn’t seem to translate to Casey.

  • Garbo

    Great post!

  • Kohanz

    Check out nbawowy.com and save yourself a lot of time!

    • DanH

      Yeah I’ve seen it. I actually used it for a good chunk of this.

      • Kohanz

        Right, saw the update in the middle of the article. I was just reading the intro and thinking “This guy needs nbawowy”! Vorped.com is another great site to check out that parses play by play to deliver interesting statistics (shot zones in this case).

  • DDayLewis

    Awesome post. This must have taken you ages to compile. Really great work.

    Two things:

    1) If I’m understanding you, these numbers are essentially plus-minus numbers (only adjusted for pace with off/def rating). Any thoughts on xRAPM? I know you mostly focused on lineups, but xRAPM seems to be the best individual +/- stat out there. I bring it up because it suggests that Rudy Gay wasn’t exactly a slam dunk of an acquisition (http://stats-for-the-nba.appspot.com/teams/TOR.html)

    2) Small sample size concerns? http://wagesofwins.com/2011/03/05/deconstructing-the-adjusted-plus-minus-model/

    • DanH

      1) Yeah, xRAPM has a lot of other adjustments in there as well. Plus, xRAPM averages back a few years (I believe, RAPM used to be this way at least) so it is more steady, but less relevant to immediate changes. I like xRAPM a lot, but was looking for a way to look specifically at how the pieces fit together, rather than just the net impact they had as individuals.

      2) Oh, yeah, lots of small sample size concerns. But in sports, sample size problems are almost unavoidable. In the NHL WOWY stuff I’ve seen, a lot of stats guys say 3000 minutes is the point at which they can trust a stat to be projectable. You just rarely get sample sizes like that without huge variance in sports. Especially true when looking at lineups (which by definition play only a subset of the available minutes). That’s part of why I didn’t include Acy in this. Acy has much smaller sample sizes than anyone else, and he looks Amir-esque in them.

      • DDayLewis

        Thanks for getting back to me on this. I’m a fan of advanced stats in basketball, but I’m pretty hesitant to accept +/- stuff because of how much context it measures (which is a good, and a bad).

        It would be pretty interesting to investigate the “why” in these cases, now that we have the “what”.

        Anyway, really great work and thanks for sharing.

  • Someone

    Fun post and analyses for the dog days. One tiny suggestion – a list of abbreviations at the beginning or end would help those us less familiar with these metrics to better enjoy your hard work. Cheers

    • DanH

      ORTG – offensive rating. Points scored per 100 possessions. In this article, the ORTG used is the team’s points scored per 100 possessions.

      DRTG – defensive rating. Same idea. Points allowed per 100 possessions.

      Net RTG – ORTG subtract DRTG. How much your team wins by per 100 possessions.

      Those are the main ones for this article. Good idea for the future to include a glossary.

  • rob

    Amir should be payed 18Mil/year…

  • Fair89

    U2 – With or Without You

    http://www.youtube.com/watch?v=XmSdTa9kaiQ

  • changv10

    Nicely done Dan – a post like that should deserve you a position as moderator for this site!

    It’d be really cool if you could run those numbers in detail for the new raps Hansbrough, Augustin, Daye and Novak – and see if they made their teammates better (In Indiana, Detroit and New York). We may not know their rotations in detail, but maybe sorting by minutes played with that teammate, can give you the top 4 teammates that they influenced. It may be unfair for Daye, since he got stuck into a new environment mid-way … so like you said the Detroit analysis only may be better for him.

    And if they negatively affected other teammates, it would be interesting which rotations would work best here, by assimilating players. I know each person is different, but you may be able to find some commonalities.

    Once again, great post. Keep em’ coming.

    • DanH

      This is a good idea. I left them off because I was looking particularly at fit with the current core, but if they had teammates they fit well with who play similar styles to our core, then it would be interesting to see if that fit continues here.

  • FAQ

    WOW!!! Look at all those numbers… they must mean something, to somebody. Once again this edition of the Raptors is composed of ‘veterans’… hangers-on… newbies… wannabees and scrubs. When will they “gel” as a team, that’s the question all those numbers don’t answer? It seems like the same “gel” treadmill year after year after year after year after…….. sigh