Statistical Analysis. Humor. Knicks.

Tuesday, September 2, 2014

Knicks 2014 SCHOENE Is A Real Bear

ESPN’s prediction method SCHOENE has rung a bell with Knick fans by predicting the ‘Bockers to only win 37 games in 2014. It’s an alarmingly low number, considering the team is coming off taking the Atlantic with 54 wins and a hard fought second round series. For most fans it’s hard to understand why anything could be so pessimistic on New York’s prospects. In 2014, the Knicks are coming back with largely the same cast. The 5 starters of their final game in 2013, ‘Melo, Chandler, Felton, Shumpert, and Prigioni all will be in orange and blue come November, as will 5th man J.R. Smith, backup center Kenyon Martin, and even the oft-injured Amar’e Stoudemire. Heck New York added some decent complementary pieces in World Peace, Udrih, and Bargnani.

So what gives?

I’ll admit I’m bearish on the Knicks 2014 outlook, with the hopes that I’m wrong. But my reasons are primarily based on what has happened outside of the MSG headquarters. I feel that most of the top of the Eastern Conference has improved at a greater rate than New York, which would result in a Gotham slide down to 5th place (“Gotham slide”™ is a registered trademark of KnickerBlogger.Net.)

On the other hand, there is reason to be concerned about the actual changes with the roster. Gone are Novak, Copeland, and Kidd, along with their TS% of 60.1, 58.3, and 53.2 respectively. World Peace, Udrih, and Bargnani will likely take the minutes of those three along with Brewer and White. One idea might be to take those 5,692 minutes from 2013 and give them to the new three, then look at the aggregate numbers. For the noobs, using their career averages seems like a reasonable place to start.

Year eFG% ts% FT/FG OREB/36 TO/36
2013 51.5% 55.0% 19.6 1.6 1.7
2014 50.3% 54.4% 21.9 1.7 2.0

Looking at the projected numbers, the Knicks are likely to take a drop in shooting efficiency. Last year New York was 8th in eFG, but if they shot 50.3%, they would been tied for 12th. If we pretend the gains in rebounding and free throws offset the loss in turnovers, then it reasons that this projected 2014 New York team will see a decline in offense.

If we use the last 2 years stats for our players, the drop-off from 2013 is even larger.

Year eFG% ts% FT/FG OREB/36 TO/36
2013 51.5% 55.0% 19.6 1.6 1.7
2014 49.7% 53.9% 21.5 1.7 2.0

Now this kind of “Maimonidenian mathematics“, as Father Knickerbocker would call it, isn’t very precise. The NBA isn’t as simple as taking individual statistics across teams and roles and copy and paste numbers into excel to generate a new team total. And that doesn’t even account for the improvement/decline of players or the yearly variation in stats. But it does provide us with a starting point to a conversation on what the 2014 might be like. And it does show why an objective assessment, like SCHOENE, would also be bearish on the Knicks 2014 odds.

29 comments on “Knicks 2014 SCHOENE Is A Real Bear

  1. Owen

    Good post. I agree it’s a decent starting point for a conversation. Still think it’s way out of line though, at 37.

    BTW, HAPPY BIRTHDAY!!!!!!!!!

  2. Hubert

    Using Jason Kidd’s career numbers as a baseline is definitely going to produce a bearish result when you take them out, but we know that’s not the kind of production we were getting from him.

    Likewise, if you want to assume Bargnani’s numbers from the last two injury-marred years and project them over a large amount of minutes, we’re going to look pretty terrible that way, too. But we hope that’s an unreasonable expectation.

    If those things are factors in his SCHOENE projection, I can see how he can get to 37 wins.

  3. Mike Kurylo Post author

    Hubert: Using Jason Kidd’s career numbers as a baseline is definitely going to produce a bearish result when you take them out, but we know that’s not the kind of production we were getting from him.

    Wait, what?

    Kidd’s season was what it was. It wasn’t balanced & evenly spaced, but if you want to throw the whole thing out because he malfunctioned at the end, then you’re ignoring how good he was early on. He played in 2013, and overall played well, so you can’t just choose to ignore that because he wasn’t worth a bag of socks by June.

  4. Brian Cronin

    Yeah, I’m fine with being bearish on the Knicks in relation to the other teams in the East. A prediction of fifth is totally reasonable, which is a notable drop from second just last year. But there’s fifth and then there’s 37 wins, ya know?

  5. Frank

    Mike Kurylo: He played in 2013, and overall played well, so you can’t just choose to ignore that because he wasn’t worth a bag of socks by June.

    Interesting way of putting it, because basically after Jan 1 he was a mummified caricature of himself. He “played” nearly 1600 minutes in 2013 with a TS of 43.2. If you look at the minutes that the Knicks played with their 2 most important players on court (Tyson/Melo), the Knicks were 5 points worse on offense per 100 possessions when he was on the court than when he was off. He was a glorified version on Ronnie Brewer, who just happened to flame out a month before Kidd did. Kidd played well for about 1.5 months, and horribly for 4.5. So on balance, losing him is probably a net positive for this team on the floor. I can’t really comment on whatever “leadership” he brought, but overall I think that’s an overrated concept anyway.

    And remember also – the Knicks played their best ball with Pablo and Felton late in the year, and Felton/Kidd early in the year. Who is to say the Knicks wouldn’t actually have been BETTER with Pablo/Felton at the beginning?

  6. Hubert

    You said:

    “One idea might be to take those 5,692 minutes from 2013 and give them to the new three, then look at the aggregate numbers. For the noobs, using their career averages seems like a reasonable place to start.”

    If SCHOENE is replacing Jason Kidd’s career numbers with the career numbers of Benoh Udrih, I can understand why it would predict a precipitous dropoff.

    But, regardless of how well Kidd played, that’s not really happening. Because 1) no matter how well Kidd played overall, it was not up to his career numbers, and 2) We’re actually replacing Kidd’s minutes with an additional 800-1000 minutes of Pablo Prigioni, who was superior to Kidd in every category except rebounding. And that didn’t factor into the equation.

    So SCHOENE, while objective, would appear to be very limited, if it’s using the methodology you described.

  7. DRed

    Hubert:
    You said:

    “One idea might be to take those 5,692 minutes from 2013 and give them to the new three, then look at the aggregate numbers. For the noobs, using their career averages seems like a reasonable place to start.”

    If SCHOENE is replacing Jason Kidd’s career numbers with the career numbers of Benoh Udrih, I can understand why it would predict a precipitous dropoff.

    But, regardless of how well Kidd played, that’s not really happening.Because 1) no matter how well Kidd played overall, it was not up to his career numbers, and 2) We’re actually replacing Kidd’s minutes with an additional 800-1000 minutes of Pablo Prigioni, who was superior to Kidd in every category except rebounding.And that didn’t factor into the equation.

    So SCHOENE, while objective, would appear to be very limited, if it’s using the methodology you described.

    You sure Woody is going to play Pablo?

  8. Hubert

    Mike Kurylo: Wait, what?

    And to simplify it even further, all I really said was Jason Kidd last year was not as good as his average career numbers, so using them as a baseline would not be sound. I didn’t think that would draw surprise from anyone.

  9. Hubert

    DRed: You sure Woody is going to play Pablo?

    Was Mike (or SCHOENE) sure he’s going to play MWP, Udrih, and AB 5,692 minutes?

  10. DRed

    Hubert: And to simplify it even further, all I really said was Jason Kidd last year was not as good as his average career numbers, so using them as a baseline would not be sound.I didn’t think that would draw surprise from anyone.

    From what I understand, SCHOENE averages a players 3 previous seasons, uses that to find similar players, and then projects the next year based on those numbers. It’s more complicated than that, but that’s the gist of it.

  11. Hubert

    After looking them up, some of Jason Kidd’s numbers were better than his career numbers last year. His TS% was 25 bps higher, for instance. Rebounding was only slightly lower. But then the AST% was 17.8 to 38.5, as you might expect.

  12. ruruland

    DRed: From what I understand, SCHOENE averages a players 3 previous seasons, uses that to find similar players, and then projects the next year based on those numbers.It’s more complicated than that, but that’s the gist of it.

    The methodology used to project offense is easy to understand, though of course the very nature of any kind of projection makes it quite fallible.

    Trying to project defense on the other hand….

    If AB is playing at the level he had the previous two years, does anyone honestly believe Woodson will use him when he has so many other options (going small, using multiple wing players, starting Melo at the 4 and bringing Amar’e and Martin off the bench, using Tyler as the back-up center)?

    With three quality point guards, two of whom thrived on the floor together, to go along with what should be three quality shooting guard and two forward who can play both the three and four, there is only upside.

    AB does not have the chops, respect/clout that Kidd did. No way Woodson would ride him out.

    It would be interesting if AB was taken out of the equation entirely and replaced with Shumpert in the starting lineup (increasing the minutes for THJ and MWP).

  13. Mike Kurylo Post author

    Hubert: And to simplify it even further, all I really said was Jason Kidd last year was not as good as his average career numbers, so using them as a baseline would not be sound. I didn’t think that would draw surprise from anyone.

    I see the confusion. I didn’t use Kidd’s career average. I used his actual 2013 numbers.

    Frank: Interesting way of putting it, because basically after Jan 1 he was a mummified caricature of himself. He “played” nearly 1600 minutes in 2013 with a TS of 43.2.

    You’re just making my point stronger. Yes Kidd was garbage after January. But I’m looking at his total numbers for the year. So hence at some point he was producing at a high rate.

    In other words:

    total Jason Kidd numbers = numbers after January + numbers before January

    If the first variable of that equation is good, and the second part awful, then the last variable must have been awesome. Hence at some point the Knicks benefited greatly from Kidd’s presence on the court. Hence those numbers in their totality can’t be ignored.

  14. DRed

    ruruland: The methodology used to project offense is easy to understand, though of course the very nature of any kind of projection makes it quite fallible.

    Trying to project defense on the other hand….

    If AB is playing at the level he had the previous two years, does anyone honestly believe Woodson will use him when he has so many other options (going small, using multiple wing players, starting Melo at the 4 and bringing Amar’e and Martin off the bench, using Tyler as the back-up center)?

    With three quality point guards, two of whom thrived on the floor together, to go along with what should be three quality shooting guard and two forward who can play both the three and four, there is only upside.

    AB does not have the chops, respect/clout that Kidd did. No way Woodson would ride him out.

    It would be interesting if AB was taken out of the equation entirely and replaced with Shumpert in the starting lineup (increasing the minutes for THJ and MWP).

    No, but this team, while deep, is also injury prone. On the plus side, Bargs is one of those players, so there’s a decent chance Woody won’t have to make a decision. On the negative side, I can easily see Stat and Kenyon missing time together and Bargs getting serious minutes as a PF/backup center. Also, while Bargs was okay on offense at some points in his career, his defense/rebounding make him a net minus unless he blows away his career highs on offense. A power forward with a TS% of 55 who can’t rebound or play team defense is still pretty crappy.

  15. Unreason

    Isn’t it important to establish that SHOENE is worth discussing before discussing it. From NickC’s helpful post a couple of threads back I learned that its predicted Ws have been off in the past by an avg of 5. It’s being used to predict Ws to see how ranks playoff prospects are likely to work out.

    Being off by 5 on average suggests it would have been incorrect on about 90% of playoff rankings of the past 2 seasons. 5 Ws. To me, that’s so poor it isn’t worth chewing on why it says what it does about the Knicks.

    Unless I’m wrong about it’s predictive power I’m more interested in why it’s so consistently incorrect than the implications of its predictions.

  16. Donnie Walsh

    DRed:
    Also, I still don’t understand just handing minutes to THJ.

    Everytime I see THJ I think of the THCJ. Since he was here first, can we come up with a better KB knickname for Hardaway?

    Here are some suggestions:

    Son of Sam
    Bitchkid
    Li’l Tim
    Sin of the Father
    Anfernee Hardaway Jr.
    The Honorable Cock Jowles

  17. mokers

    Unreason:
    Unless I’m wrong about it’s predictive power I’m more interested in why it’s so consistently incorrect than the implications of its predictions.

    I think the SCHOENE people are pretty open to this. If you read the bit about it at Basketball Prospectus, they talk about how hard it is to predict performance because of all the interaction effects and they go on to talk in particular about rebounding.

    Donnie Walsh: Everytime I see THJ I think of the THCJ.

    Glad I’m not the only one, we should put out a newsletter.

  18. Z-man

    Melo has never played on a team that won less than 42 games (other than the 36-30 during the lockout year.) Nor has he ever been on a team that missed the playoffs. Why do we even bother debating this garbage? Are the roster changes (ours and other teams) so significant that it can be reasonably predicted that we would lose 17 more games than last year?

    Let’s see:
    we lost: Novak, Kidd, Wallace, Brewer, Camby, Thomas and Copeland.

    we gained: Bargnani, Udrih, World Peace, Hardaway, and have roster spots left to fill with the likes of Tyler, Aldrich, Murry or someone not yet named.

    17 games?

    I would hope that anyone defending this “analysis” is rushing to Vegas to put their life savings on the under.

    If a reasonable margin of error is +/- 4 wins, and the Knicks win at least 48 games, can we pledge never to revisit SHOENE’s methodology again?

  19. stratomatic

    I think this a good way start the conversation. I’m bearish on the season mostly because other teams improved, but I agree that the team got a little weaker. The only thing that would change my mind is an upside surprise from Shumpert or Hardaway.

  20. massive

    This is going to be an interesting season for sure. It’s going to be hilarious when we win 55 games and people in the media are screaming Carmelo for President and people here are screaming Tyson/Shump Shump for President.

    This team will be great again this year. 37 wins is ridiculous and not worth debating honestly. Well, unless you believe in the Bargspocalypse. Then 37 wins is reasonable.

  21. cgreene

    Also the minutes replacement method was completely flawed. Kidd, Copeland, Brewer, Novak, Camby, Wallace is being replaced by Shumpert, Prigs, Bargs, Metta, Martin, Udrih probably in a similar order as what I described. So those should be the plug ins.

  22. MJG1789

    Z-man:

    Let’s see:
    we lost: Novak, Kidd, Wallace, Brewer, Camby, Thomas and Copeland.

    we gained: Bargnani, Udrih, World Peace, Hardaway, and have roster spots left to fill with the likes of Tyler, Aldrich, Murry or someone…

    Sanity. Jason Kidd had a great month, but even when he was good he wasn’t Michael Jordan. The rest of the year he was a pile of suck. Of the rest, only Copeland made significant contributions. Bargnani might end up having a bad year, but unless we are projecting him to strangle Melo, Tyson, Shump, etc., why are we suddenly a disaster? And who are these eastern conference powers beating us?

  23. Unreason

    mokers: I think the SCHOENE people are pretty open to this. If you read the bit about it at Basketball Prospectus, they talk about how hard it is to predict performance because of all the interaction effects and they go on to talk in particular about rebounding.

    I don’t mean any disrespect to the people behind SCHOENE. They’re probably just blindly groping toward something trying to illuminate it using what’s they have access to. More power to them.

    To make progress I think they should propose, measure and empirically verify ideas about what causes team’s relative success over the course of a season. Maybe they already are. I’m sure that’ll be a long slog. Without some of that kind of progress, I don’t see how they can tell whether the problem is due to strongly interacting forces – i.e. an inherently short prediction horizon like weather systems – or if it’s a highly predictable phenomena, but they aren’t able to understand or measure the things that matter.

    Good predictive models are based on knowledge of causal relationships and theory-driven ideas about what is and isn’t important and the functional form of their interrelationships. Starting with data that happens to be available and working backwards from it to try explain the outcome you’re interested is a good recipe for making up a lot of plausible sounding stuff that’s pretty far off base. Even when data mining approaches are successful at predicting outcomes, they say pretty much nothing about the mechanisms leading to it.

    Given how poorly this one seems to predict what it’s supposed to at this stage in its development, I’m a little surprised it gets play in a national forum.

  24. Jack Bauer

    Mike Kurylo: Wait, what?

    Kidd’s season was what it was. It wasn’t balanced & evenly spaced, but if you want to throw the whole thing out because he malfunctioned at the end, then you’re ignoring how good he was early on. He played in 2013, and overall played well, so you can’t just choose to ignore that because he wasn’t worth a bag of socks by June.

    I think you meant he wasn’t worth a bag of socks by mid JANUARY

  25. Hubert

    Mike Kurylo: I see the confusion. I didn’t use Kidd’s career average. I used his actual 2013 numbers.

    Ah, I see. I misread your post. Rather foolishly, I might add.

    I do agree with your overall point. Kidd contributed TREMENDOUSLY to one of our two best stretches of the season (the 18-5 start) and that will have to be replaced.

    I’m confident that it can be based on the fact that he contributed less than zero to the other best stretch of the season (the 16-2 finish).

    The biggest contributor to that stretch that we lost was Copeland.

Comments are closed.