Statistical Analysis. Humor. Knicks.

Friday, September 19, 2014

Math Soup for the Soul: Possibilities for Eastern Conference Playoff Seeding

These days, the comment section of Knickerblogger is filled with exchanges like this:

“Good god. The Knicks are a combined 1-4 against teams below .500 since the Carmelo Anthony trade! Our defense looks like swiss cheese carved into the shape of a matador!”

“Oh please. They are 6-2 against teams OVER .500. Which group do you think we’ll be playing against in the playoffs?”

For better or for worse, KnickerBloggeristas are married to Carmelo Anthony, and that commitment has understandably made a lot of folks anxious. I figure, what better way to get control of our emotions than to turn to the numbers? The cold hard numbers.

Let’s start with some pie charts showing how tough the various Eastern Conference playoff teams’ schedules are. I sorted teams by win percentage and placed them in categories (bottom third, middle third, top third) and then compared those categories against teams’ remaining schedules.

mmmm basketball pie charts...

So, as you can see, the Knicks’ schedule looks a whole lot easier on the surface than Atlanta’s or Philadelphia’s. Let’s look a little closer:

Winnables is now a word!

What we see here are some reasonable best case and worst case scenarios for each team. I categorize “winnables” as games vs. the bottom 2/3s of the league, and “gimmes” as games vs. the bottom 1/3 of the league. A lot of this was already clear: Orlando is pretty much locked in at the 4 seed. Miami has a distinct chance of moving up to the 2 seed.

There also are, however, some interesting things to note:

1) Philadelphia’s chances of moving up revolve around a worst case finish for New York and a best case finish for itself.
2) Chicago, despite being only a half game ahead of Boston, is much safer from Miami.
3) And… The fifth seed is still well within reach for New York.

Let’s look closer by ditching those crudely cut categories and use something fancy. Pythagorean Wins is a formula based on a team’s scoring differential, and is known to be more accurate in predicting wins than even wins when looking between seasons. It predicts based on the Knicks’ performance pre-tonight that they should be 34-32 (their exact record). Looking through about a half dozen teams, I’d estimate that it is about 1-2 games off on average, except in Texas. Texas teams for some reason don’t rhyme well with its equation, as the Spurs are supposed to be seven games worse than they are.

Anyway, I applied BR’s formula in reverse. I took the average score of games already played between teams that will play in this final month and applied them to the formula. Wanting all factors included, I also adjusted for the proportion of home games vs. road games (.12 game bonus) and the number of back to backs (.15 game bonus) played and played against. Finally, I gave teams that play end-of-season against a team that is likely to rest its starters (games vs. any of the top three seeds) a .25 game bonus.

Words, lines, numbers, and colors.

As you can see, things look less rosy for New York in this model. However the scores applied to this formula were from the old Knicks, in the pre-Melozoic era. The new Knicks may be better or worse than their predecessors.

One factor that these stats don’t weigh is individual match-ups, for instance the Celtics have beaten Atlanta by an average of 17.5 points in their two games so far this season. They have won their games against their other two possible playoff opponents by an average of 0.7 points (Philadelphia) and 3 points (Knicks). Additionally an older team, Boston may be happy to rest starters in their back to back games. In order for them to get a more favorable match-up against the Hawks, the Knicks will have to move up. And with two games left against New York, Boston will have a say on the Knicks position. Interestingly, Miami has two games left against Atlanta. If they go all out for the win, they risk helping Boston get a more favorable first round team.

There’s one last thing I want to show you all before we bag and tag this one – something for the pragmatists, cynics, and optimists alike. Here’s what Pythagorean Wins has to say (adjusted for strength of schedule) about the number of wins this Knicks team would get in a full season based on their performance in the games listed.

48 wins makes me mellow.

So now go ahead you optimists. Post that comment: “Jordan Bulls, we’re coming for your 72 wins! After we get acquainted with Larry O’Brien this year of course.”

And you cynics: “I hope the Rockets enjoy our lottery pick next year. I’m through with this joke of a franchise. I’ll be watching the Knuggets.”

And finally, ye’ steady pragmatists: “48 wins. That’s just what I posted three weeks ago. Big surprise. Excuse me, but I have a date with a regression analysis that has curves in some very interesting places, if you know what I mean.”

Ah but wait! What’s this? This huge bowl of math soup is doing something to my stomach! It’s giving me a gut… feeling. A gut feeling. Yes, that’s it. And as we all know, gut feelings are the most reliable form of statistical analysis. My gut is saying:

#1 Chicago vs. #8 New Jersey (Haha Utah, you should have traded us Deron. Lottery pick = gone)
#2 Miami vs. #7 Philadelphia
#3 Boston vs. #6 Atlanta
#4 Orlando vs. #5 New York

31 comments on “Math Soup for the Soul: Possibilities for Eastern Conference Playoff Seeding

  1. jon abbey

    the Nets are playing very well but probably have to go 13-3 or 14-2 to make the playoffs, that’s asking a lot. Holllinger’s model still gives them absolutely no chance (0.0).

  2. cgreene

    Interesting piece. The optimist would say that if as currently constructed the team is a 48 win team over the season then with some positive tinkering over the summer for more of the right role player that we can easily get into the low to mid 50′s. Maybe that’s contention. Maybe it isn’t. But I say “C’MON!!!” to anyone who wouldn’t have taken that one year ago today. Then you have the Chauncey $12M to play with in 2012. I see a crack of blue in the sky, Hoolahoop and THCJ!!!

  3. Doug

    cgreene: But I say “C’MON!!!” to anyone who wouldn’t have taken that one year ago today. Then you have the Chauncey $12M to play with in 2012.

    Like the guy with 12 million dollars in cap space is going to spend it on some role player. COME ON!!

    #GOB

  4. Jafa

    I think both NYK and PHI will pass ATL, IND will get the 8th seed and MIA will get the 2nd seed setting up:

    CHI vs. IND (this should be a sweep for the Bulls)
    MIA vs. ATL (this should be a sweep for the Heat)
    BOS vs. PHI (this should be a nightmare for the Celtics)
    ORL vs. NYK (this should be painful to watch for Knicks fans)

  5. latke

    Yeah, we need to get put Shawne Williams and Jared Jeffries on like twelve protein shakes a day. If they can get up to say 280 pounds, maybe they can keep Howard from getting position directly under the basket every time.

  6. Caleb

    I would much rather be playing Orlando than any of the other big 3 teams, although Boston is old and creaky enough that the Knicks might catch an injury-related break.

  7. ess-dog

    God I love pie charts. That was a pretty positive featurette. I think we all would’ve signed up for 43 wins at the beginning of the season, much less 47. Just think, if we’d beat Cleveland twice and Indy once we could be looking at a 50 win season right now.
    The question is, will we continue to lose to crappy teams? If TD can continue to play well and Jeffries just plays semi-decent, we have a shot at 47 wins and the 5th seed, since the Hawks just look terrible right now. They have to be kicking themselves for that Joe Johnson signing.

  8. cynickfan

    The math geek that I am – I find all of this very fascinating. But the real question I ponder is this: Of the top 4 – Chi / Mia / Bos / Orl – who would we want to face in the first round to maximize the chances of an upset?

    We have beaten chi twice, but with Noah in there, I don’t like our chances.
    We have taken Miami down and we can probably take them down again
    We have been buried by Orlando and that is not likely to change
    We have lost to boston, but closely. But they are creaky and without Perk.

    To maximize our upset chances, I think we would want Bos / Mia / Chi / orlando – in that order. Assuming these four teams finish 2 / 3 / 1 / 4 we want to finish 7 / 6 / 8 / 5. I don’t see Orlando getting to the 3rd spot, so I would absolutely NOT want to pass Atlanta. The moment we hit the playoff magic number we should shut it down – we are more likely to upset at the bottom of the bottom half than at its top.

  9. tenebrous

    latke: Yeah, we need to get put Shawne Williams and Jared Jeffries on like twelve protein shakes a day. If they can get up to say 280 pounds, maybe they can keep Howard from getting position directly under the basket every time.  

    can definitely be done.

  10. CRJoe

    Am I the only who would much rather take MIA over Boston??? It´s a joke to compare a team that has had their core together for 4 seasons, has been to the finals twice, and has a 9 player deep roster with a team that has 2 great players, 1 good player and a bunch of scrubs… And if we go by the numbers we all know how MIA is doing against decent opponents AND against the Knicks…

  11. DS

    “(Haha Utah, you should have traded us Deron. Lottery pick = gone)”

    I bet the Jazz’s #1 goal is to get Jimmer Fredette.

  12. Jafa

    CRJoe: Am I the only who would much rather take MIA over Boston??? It´s a joke to compare a team that has had their core together for 4 seasons, has been to the finals twice, and has a 9 player deep roster with a team that has 2 great players, 1 good player and a bunch of scrubs… And if we go by the numbers we all know how MIA is doing against decent opponents AND against the Knicks…  

    CRJoe,

    I don’t’ want either MIA or BOS, but if I had to choose, I choose Boston. Don’t forget, Miami’s 2 great players have both been to the finals too, and Wade has as many rings as the Celtics players.

    The other thing is that Boston has trouble with our style of play, evidenced by how many points we put up on them when we played them (They won 105-101 on Oct. 29 and 118-116 on Dec 15). If we can get them to go up and down the court, we could make it an interesting series and even have a chance at an upset. Miami, on the other hand, can play our style probably better than us (if they had the knock down shooters we have on the corner) and even when we won, it was with great defense, not offense (in our 4 games with them, we scored 91, 98, 93 and 91 points). Remember, we average 106.8 PPG.

    So, are you willing to bet on our offense or our defense?

  13. fushknicken

    Great post, super interesting stuff. 48 wins pretty much, one thing that I wish someone could look into is whether the differential between above .500 and below .500 teams is meaningful. It seems to be somewhat of an arbitrary cut-off point to me, and that the point differential in the smaller samples induced by the threshold has too much variance. Does, say the lakers’ pythagorean record have such a large disparity in point differential? Are there any teams with a large disparity? What is the distribution of this difference? I would hypothesize no, and no and wide-Gaussian, possibly bi-modal for the two conferences. Proof anyone?

  14. Caleb

    It’s useful shorthand but above- and below- .500 is an arbitrary cutoff; the quality of your opponents is on a sliding scale.

    e.g. if you play a 65-win team and a pair of 40-win teams, you have a much harder schedule than if you play a pair of 42-win teams and a 20-win team.

    For a more precise look, (if you’re an ESPN Insider) look at Hollinger’s rankings and look at SOS (strength of schedule). Since – at the end of the year everyone’s SOS is right around .500 (at least within a couple of points), you can see at a glance whose remaining schedule is hardest.

    In order of hardest remaining schedule, to easiest. (The number is SOS to date… he list looks backwards, but that’s because an easier schedule to date means a tougher schedule from here on).

    Atlanta .482
    Boston .485
    Philadelphia .485
    Chicago .486
    Orlando .495
    New York .495
    Miami .497

    Miami has the highest percentage of road games left, Philly the most home games, but it’s pretty evenly split all around.

    All in all I don’t think strength of schedule will be a big factor.

  15. Jafa

    fushknicken: Great post, super interesting stuff. 48 wins pretty much, one thing that I wish someone could look into is whether the differential between above .500 and below .500 teams is meaningful. It seems to be somewhat of an arbitrary cut-off point to me, and that the point differential in the smaller samples induced by the threshold has too much variance. Does, say the lakers’ pythagorean record have such a large disparity in point differential? Are there any teams with a large disparity? What is the distribution of this difference? I would hypothesize no, and no and wide-Gaussian, possibly bi-modal for the two conferences. Proof anyone?  

    Check the pudding!

  16. John Kenney

    Guys- I don’t ever wish us to lose an individual game. but the silver lining is that as long as we still have a better record than the rockets, it doesn’t hurt us to possibly move up a draft spot. We aren’t winning the title this year anyways, the celtics/bulls/heat are all about the same for me, and i really want us to get faried or one of the morris twins.

  17. Jafa

    John Kenney,

    I respectfully disagree. I simply cannot support any form of intentional tanking after the unintentional tanking of the last 10 years. We have known losing too long, and unless we have a chance at getting a LeBron James type of player, then we just have to draft like we did last year and like the Spurs do every year.

  18. Robert Silverman

    As long as the Knicks make the playoffs and the Rockets don’t, the Knicks will have the lower draft pick

  19. d-mar

    I know I’m in the minority, but I’d much rather play Orlando than Boston, Miami or Chicago. I know Dwight eats us for lunch and we can’t stop him, but if our biggest defensive challenge is a pure low post scorer who can’t shoot free throws, I’ll take it. The rest of the team just doesn’t scare me that much (I know, Nelson torched us last game, but he was Toney-esque from the outside)

    This is NOT to say that we wouldn’t be pretty heavy underdogs, I just think Orlando is more flawed than the others and could be riper for an upset.

  20. Caleb

    @23 I completely agree – although the Perkins trade hurt Boston, IMO, and they are always an injury away from trouble. If Shaq isn’t back their frontline is awfully thin.

    Miami and Chicago will be a lot tougher, IMO – I wouldn’t put much stock in the regular season results.

  21. JK47

    Miami’s half court offense is pretty shoddy, and in the playoffs there’s usually lots of half court play.

  22. max fisher-cohen Post author

    fushknicken, I tried a couple of other samples to test how the formula works when you do a division like this: the Heat and the Lakers.

    The Lakers project as a 44 win team vs. .500 or worse teams and a 65 win team vs. .500 or better teams.

    Lakers:
    overall: 60 wins
    vs. .500 or worse: 44 wins
    vs. better than .500: 65 wins

    Heat:
    overall: 59 wins
    vs. .500 or worse: 47 wins
    vs. better than .500: 62 wins

    Kings:
    overall: 26 wins
    vs. .500 or worse: 23 wins
    vs. better than .500: 26 wins

    It’s possible that many teams play down to their competition, or that I shouldn’t be adjusting linearly for SOS, or that this four team sample is just skewed. In any case, you can see that although the projections all seem to favor the “vs. better than .500″ schedule, the difference between the Knicks’ projection and these others is still huge.

    A more conservative estimate (though still not totally thorough) would involve me halving the effect of SOS

    Lakers:
    54 wins vs. .500 or below
    59 vs. better than .500

    Kings:
    29 wins vs. .500 or below
    24 vs. better than .500

    Heat:
    58 wins vs. .500 or below
    56 wins better than .500

    Knicks:

    vs. 500 or worse: 26 wins
    vs. better than .500: 63 wins

  23. nicos

    Unless the Knicks go on a rampage the next 3 weeks they’re going to be an underdog no matter who they face. I think Orlando would probably give us the best chance of winning- when Jameer Nelson struggles shooting they’re very beatable. If he’s on the Knicks can probably forget it but if he struggles enough that the Knicks guards can go under the high screen at least some of the time and keep him out of the paint they have a decent chance no matter what Howard does. Their forwards are also their weak spot defensively so I think that gives the Knicks some hope as they don’t really have good match-ups for either Carmelo or Amar’e (putting Howard on Amar’e risks both foul trouble and leaving the middle wide open neither of which Orlando’s going to want to do. Barring Orlando I’d say Chicago- less playoff experience Boston/Miami and an offense that can sputter at times- I also wonder if they have that extra playoff gear or if they’re playing pretty much at maximum level right now. Unfortunately if we stay at 6/7 it’s probably going to be Boston or Miami in which case hope for injuries- or that TD and Billups can shoot 75% from three all series long!

  24. John Kenney

    @21 I wasn’t saying the players should tank or have that mindset. I’m saying that as a fan, when I think about the losses we’ve been having, while frustrating, they might help the team some in the long run, given we stay at least at the 7th seed.

  25. jon abbey

    John Kenney: @21 I wasn’t saying the players should tank or have that mindset. I’m saying that as a fan, when I think about the losses we’ve been having, while frustrating, they might help the team some in the long run, given we stay at least at the 7th seed.  

    ha, this occurred to me in a different way the other day, which was that losing two to Indiana helped to ensure NJ misses the playoffs and cuts down the chances of Williams staying there a bit.

  26. dsulz

    I don’t really get how you come out with 48 wins. If we’d started with Melo from the beginning of the season, perhaps.

    Overall though, I think all this analysis, great as it is, is kind of beside the point. Past behavior is the best prediction of present behavior. We were two games over .500 pre-Carmelo. With Carmelo we have a .500 winning percentage. I see no reason to believe why we will do particularly better down the stretch. We have 14 games left. Using my first grade arithmetic skills we can see that if past winning % is a predictor of future winning % we will win 7 or 8 of our final remaining 14. Which will leave us with 42-43 wins.

    You could also make the argument that we’ll finish worse because we’re playing a lot of shitty teams down the stretch and we’ve made a habit of losing to shitty teams since we got Melo.

    Nice post Max, I don’t mean to rag. But 48 wins? C’mon.

  27. max fisher-cohen Post author

    dsulz: Nice post Max, I don’t mean to rag. But 48 wins? C’mon.  

    The 48 wins IS from the beginning of the season, as are the 70 and 20 wins. The formula I used predicted 43 wins for the knicks this season (see the chart in the middle with the caption “words, lines, numbers and colors” for predictions for all seven teams this season). Would be a little lower after the loss last night.

    If I take all the games since the carmelo trade and before last night’s debacle, I get about 43 wins again (not surprising since we haven’t been much better/worse since the trade).

    I’m not sure what you’re saying though about teams continuing to win at the exact rate that they have in the past. The Knicks have been around .500. By that logic, they should follow a pattern of win/loss/win/loss/win/loss throughout the entire season. That hasn’t been the case. They’ve had two six game losing streaks along with a stretch of 13 wins in 14 games. Much of that variation had to do with the quality of their opponents. Before last night, the Knicks had a moderately favorable schedule (especially considering how they’ve fared in the past against their upcoming opponents).

    That said, that advantage is only marginal. That’s why the formula only predicts 43 wins. The real reason they have a shot at the 5th seed was because Atlanta has a particularly difficult schedule against opponents that they have had real trouble with (as evidenced by their disaster against Miami last night).

Comments are closed.