Statistical Analysis. Humor. Knicks.

Monday, October 20, 2014

Playoff Odds, Based on Pre-Season Predictions

ESPN’s forecast for the N.B.A.’s Eastern Conference came out last week. The prognosticators collectively saw the Knicks as the 5th seed, behind their area rivals the Brooklyn Nets. While such pre-season crystal-balling is officially meaningless, it does give a starting point to perhaps do some real analysis.

Luckily for a numbersmith like yours truly, ESPN published the consensus wins and losses for each team. Taking that into account I decided to run a simulated season based on those win totals. Nothing too complicated, just randomizing each of the 82 games based on the individual team’s winning percentage. Doing that I was able to come up with a total simulated number of wins for each team, and even calculated the playoff seed for each team.

Taking that data, I ran 10,000 seasons, and came up with the following odds for the 2014 season:

Team Total seed1 seed2 seed3 seed4 seed5 seed6 seed7 seed8
Miami Heat 100% 67% 21% 8% 3% 0% 0% 0% 0%
Chicago Bulls 100% 15% 31% 27% 19% 7% 1% 0% 0%
Indiana Pacers 100% 10% 25% 29% 24% 10% 1% 0% 0%
Brooklyn Nets 100% 7% 20% 26% 30% 14% 2% 1% 0%
New York Knicks 99% 1% 4% 9% 20% 47% 13% 5% 2%
Atlanta Hawks 76% 0% 0% 0% 2% 7% 26% 23% 19%
Cleveland Cavaliers 69% 0% 0% 0% 1% 5% 20% 22% 20%
Washington Wizards 68% 0% 0% 0% 1% 5% 20% 21% 21%
Detroit Pistons 59% 0% 0% 0% 1% 3% 14% 20% 21%
Toronto Raptors 17% 0% 0% 0% 0% 0% 2% 5% 9%
Milwaukee Bucks 8% 0% 0% 0% 0% 0% 1% 2% 5%
Boston Celtics 4% 0% 0% 0% 0% 0% 0% 1% 2%
Charlotte Bobcats 1% 0% 0% 0% 0% 0% 0% 0% 0%
Orlando Magic 0% 0% 0% 0% 0% 0% 0% 0% 0%
Philadelphia 76ers 0% 0% 0% 0% 0% 0% 0% 0% 0%

On an initial glance, things don’t look very good for New York. The Nets have 7 times the probability of winning the first seed, and has a 1 in 4 shot at for making the top two seeds. Meanwhile, the Knicks have only a 1 in 20 chance of being at least the East’s second best team. On the sunnier side, the Knicks have a 34% chance at earning home court playoffs for the first round, and their is almost no possibility of missing the playoffs altogether.

To go a little more granular, and get an idea on what these seasons look like, I ran 10 random seasons for the Knicks:

Seed Wins
4 57
5 48
5 46
5 50
5 45
3 56
5 43
2 52
6 46
3 50

For Knick fans, it should be difficult to be upset at any of these outcomes. Even the worst seasons, where the team only manages a 5th or 6th seed is a relatively mild downside. There is little difference between the 5th and 6th seed when the playoffs come around since the 3rd or 4th seed will be about the same strength. Those seasons where New York manages a 3rd or 4th seed is about the same outcome, except for the advantage of home court. Finally, that 2nd seed would be an cause for celebration.

While exercises like these are mostly for entertainment sake, they give a glimpse into a possible future for the boys in blue. Perhaps they are more telling of what the perception is of the Knicks than any real prediction.

33 comments on “Playoff Odds, Based on Pre-Season Predictions

  1. mokers

    I don’t think you can necessarily be upset by these outcomes, but I think there is a lot of reason to be more optimistic about the results than you mentioned here. Here is the reason:

    http://espn.go.com/nba/story/_/page/ESPN-Forecast-panel-130415/espn-forecast-panel

    There is nothing scientific about the panel at all. No offense, but it’s not like the survey is an accumulation of people who have run simulations. You are running your simulation on survey data, so while this is a nice exercise, we are also culling from the same type of group that gave us this:

    http://espn.go.com/nba/preview2012/story/_/id/8517191/2012-13-nba-predictions-atlantic-division-champion

    It’s an interesting way to get a conversation started, but somebody wanting to do some math (not me at the moment) could come up with some much better simulation data. You could take some of the advanced stats, say PER, WP, WS, on/off, TS% and USG%, plug in what you think will be playing time numbers for each team and run the monte carlo that way. You may still get close to the same numbers, but I think it would be a much more useful dataset. You could then have discussions about playing time assumptions, make adjustments for expected aging, etc.

    I think all this tells us is if conventional wisdom will be more accurate than last year, when “Paul Pierce and a rebooted Boston roster look to finish atop the Atlantic for a sixth straight season”.

  2. Hubert

    BigBlueAL:
    I really cant wait for the NBA to start.Most excited I am for an NBA season since the 1992-93 season.Not coincidentally 1992 was the last time I had the misfortune of watching such a boring Yankee team.

    If there was ever a time for NY to become a Knicks town again this is it.Yankees are as bad as they have been in 21 years, the Mets suck as usual.Ditto the Jets.The Giants arent looking too good.This is the time for the Knicks to own NY again like they did from 1992-1994.

    BBA I think you and I are about the same age and have the same recollections (I’m 36). I love telling people about what it was like in the early 90’s. They never believe me. “No, really. No one went to baseball games. No one talked about them. It was all Knicks. When it wasn’t Knicks, it was actually hockey. Yes, hockey. It’s a sport played on ice with skates.”

  3. Hubert

    As for this piece, I mean, I appreciate the hard work, Mike. It’s an entertaining exercise. But I’m sure even you know it’s built on a shaky foundation! The consensus wins & losses of ESPN prognosticators isn’t really something from which meaningful probabilities can be drawn.

    If the prognosticators are right, we’re fucked. But there is good reason to believe those predictions have some terrible biases built into them.

    Everyone making predictions on Indiana is assuming another leap forward in the development of George and Hibbert. That is being treated as a guarantee in the consensus predictions, when we know it isn’t. The predictions also seem to be biased towards the results of two tight playoff series instead of the 82 game season that preceded it.

    And don’t even get me started with Brooklyn. There are so many extreme variables present that no team in the NBA should have a higher standard deviation in predicted outcomes. Yet the consensus seems to be rather uniform: all the variables will work out in their favor, they will be better than last year, and they will win the Atlantic.

    I know you know this, I know this was just a creative exercise (and a good one), I just needed to vent about the herd mentality of pre-season predictions.

    Where did my calm go?

  4. EB

    This is pretty cool Mike! But I instinctively dislike anything that disagrees with JR Smith’s predictions!!! Can we run this using his predictions? I wonder if he can name all the teams in the NBA, or would we get his Chinese team winning the 6th seed?

  5. Hubert

    One thing that might be an interesting exercise:

    What biases do you think are impacting the consensus predictions for each team the most, and how does that skew the resulys?

    Miami – consensus seems to assume they won’t try as hard, skewing them down. Is there any real reason to assume they will drop off by 6 games this year?

    Chicago – I think people are forgetting what a dominant regular season team they had been before last year and skewing their picks down. Some people should be picking these guys to finish ahead of Miami. They did it for the first two years of the big 3 era, after all.

    Indiana – As I said above, I think the results of their two playoff series are being given far more weight than the 82 game season, and the unpredictability of young player development is being ignored, skewing their results up.

    Brooklyn – I think people are just being ignorant and going along with the crowd. No team has a greater chance of ending up completely out of the playoffs than this team. But the real possibility that this is a disaster isn’t being taken into account, skewing their consensus higher.

    Knicks – the opposite of Brooklyn & Indiana, our 82 game season is being given less weight than our playoff series, and every variable is being assumed to turn out negative, skewing the consensus picks for us down.

  6. johnno

    mokers: we are also culling from the same type of group that gave us this:
    http://espn.go.com/nba/preview2012/story/_/id/8517191/2012-13-nba-predictions-atlantic-division-champion

    This is pretty hilarious — last year, out of 35 “experts,” 29 predicted that the Celtics would win the Atlantic and 3 each picked the 76ers and Nets. You would have thought that at least one would have picked the Knicks, if for no other reason than to stimulate laughter and ridicule. By the way, on ESPN.com right now, they are doing their countdown of all of the NBA players and they have a picture of Marcus Camby in a Knicks uniform holding a basketball. It looks like he’s playing a game. I didn’t realize that such a picture actually existed. Imagine that — there was a point last year at which he was healthy enough to get in a game. Who knew?

  7. mokers

    johnno: This is pretty hilarious — last year, out of 35 “experts,” 29 predicted that the Celtics would win the Atlantic and 3 each picked the 76ers and Nets.You would have thought that at least one would have picked the Knicks, if for no other reason than to stimulate laughter and ridicule.By the way, on ESPN.com right now, they are doing their countdown of all of the NBA players and they have a picture of Marcus Camby in a Knicks uniform holding a basketball.It looks like he’s playing a game.I didn’t realize that such a picture actually existed.Imagine that — there was a point last year at which he was healthy enough to get in a game.Who knew?

    Exaclty. You would think one person would have picked the knicks just because it would go against the grain. You would think that anybody who picked the 6ers would have been banned from making predictions in the Atlantic for one year.

  8. Mike Kurylo Post author

    DRed:
    You’d think once in 10,000 seasons Lebron would blow out his knee and the Heat would fall apart.

    Obviously that’s not what the simulation is looking at. Quite honestly the point of the article is to show how statistical distribution works. A simulation blind to things like “what is the chance LeBron gets hurt for the year” still has the Heat not winning the division a third of the time.

    In sports it often seems like the players will things to happen. But at times I feel that isn’t the exact representation of how things work. Players can run a play perfectly and have it fail, or botch a play and succeed. There’s an element of chance involved that needs to be acknowledged that makes things less certain than you would imagine.

  9. BigBlueAL

    Hubert: BBA I think you and I are about the same age and have the same recollections (I’m 36).I love telling people about what it was like in the early 90?s.They never believe me.“No, really.No one went to baseball games.No one talked about them.It was all Knicks.When it wasn’t Knicks, it was actually hockey.Yes, hockey.It’s a sport played on ice with skates.”

    Im 33 (born in 1980). People are shocked when I tell them as a kid the team I fully expected to win a championship first in my lifetime was the Knicks not the Yankees lol. The Giants won the Super Bowl in 1986 (well the 1986 season with the Super Bowl played in Jan 1987) and my dad bought the season video which I watched almost on a daily basis but at that time I was just really a Yankee fan. Didnt start rooting for the Knicks until the 1989-90 season which was when I first had cable and watched them on MSG.

    From 1992-1995 it was really all about the Knicks for me (though the Yankees started becoming good again in 1993). The current Knicks dont have anywhere near as good a chance of winning a title as the the Knicks from 1993-1995 did but this is the first time since then they have better odds of winning a championship than the Yankees do lol.

  10. Hubert

    ’93 was a mini-renaissance. The Knicks won 60, Simms & LT had one last hurrah and won 11 games, and Jimmy Key & Paul O’Neil started the Yankees revival.

    Then the Knicks blew it in game 5 (and forget Charles Smith, we missed 15 FT’s in that game 5 and shot 57% from the line in a one point loss; it was horrifying), the Giants lost 42-3 to SF, and the Yankees ended up setting a record for spending the most days tied for 1st without ever being in 1st.

    But yeah, back then you would have bet on the Knicks winning a title before any of those three teams.

  11. Hubert

    MJG1789:
    http://www.thenbageek.com/articles/who-will-benefit-from-sportvu

    An interesting article about SportsVu suggests that rebounding numbers may not be as cut and dried as WOW would have it, so immediate ad hoc reasoning is employed.

    That’s almost comical.

    “You see, it’s one thing to have a ton of data. It is entirely another thing to know what to do with it.”

    Now watch me give you data and intentionally misconstrue to fit my pre-existing belief!

    “Reggie Travels 4.3 feet to Lopez 6.4 feet? That’s probably because Reggie’s much faster and more accurate at geting to the most likely spot where the ball is going to come off the rim.”

    “More of Lopez’s rebounds are contested? That probably means that Lopez isn’t blocking out as well”

    “It’s amazing that two people can look at the data and come to exact opposite conclusions.”

    Yes it is.

  12. DRed

    Yeah, the thing about that Sports Vu data is it doesn’t tell us a whole lot (yet). One, it’s from an 18 game sample. Two, we can’t see enough of the data to draw conclusions from it. You can’t look at that and say “Ha, suck it nerds, we’ve proved Lopez is better than Reggie because he gets 1% more of the rebounds he’s close to over an 18 game sample”. I think it’s fair to ask why a good rebounder isn’t getting closer to more rebounds, and that sports vu data doesn’t seem to answer that question. Is it possible that good rebounders are better at generating uncontested rebounds? Again, that data doesn’t show anything that would answer that question. It’s interesting stuff, though.

  13. nicos

    DRed:
    Yeah, the thing about that Sports Vu data is it doesn’t tell us a whole lot (yet).One, it’s from an 18 game sample.Two, we can’t see enough of the data to draw conclusions from it.You can’t look at that and say “Ha, suck it nerds, we’ve proved Lopez is better than Reggie because he gets 1% more of the rebounds he’s close to over an 18 game sample”.I think it’s fair to ask why a good rebounder isn’t getting closer to more rebounds, and that sports vu data doesn’t seem to answer that question. Is it possible that good rebounders are better at generating uncontested rebounds?Again, that data doesn’t show anything that would answer that question.It’s interesting stuff, though.

    Yeah, I’d love to see Sports Vu try to break down why Evans is two feet closer to his rebounds- I’m sure it’s a mix of Evans having superior anticipation on the one hand and the fact that he doesn’t bother to contest shots (just 13 blocks last year) and just heads to boards early a la Lee or Love. It’s going to take awhile to figure out how to use sports vu data in a useful way but I don’t think there’s any question that it’ll revolutionize how we view basic box score stats.

  14. Donnie Walsh

    nicos: I’m sure it’s a mix of Evans having superior anticipation on the one hand and the fact that he doesn’t bother to contest shots (just 13 blocks last year) and just heads to boards early a la Lee or Love.

    I just don’t understand anybody’s infatuation with rebounding. Like defense, it’s a team concept. The better teams tend to force more misses, and thus get more rebounds. But even that doesn’t tell a whole lot because the best rebounding teams almost never win the championship. Last year Miami was in the bottom quarter of the league in rebounding and won 66 games. I don’t know if there is a study that shows a great correlation between rebounding and winning, but I kind of hope this SportsVu thing does expose it because volume rebounding seems as useless as the volume scoring that people here like to denigrate so much.

  15. DRed

    Donnie Walsh: I just don’t understand anybody’s infatuation with rebounding. Like defense, it’s a team concept. The better teams tend to force more misses, and thus get more rebounds. But even that doesn’t tell a whole lot because the best rebounding teams almost never win the championship. Last year Miami was in the bottom quarter of the league in rebounding and won 66 games. I don’t know if there is a study that shows a great correlation between rebounding and winning, but I kind of hope this SportsVu thing does expose it because volume rebounding seems as useless as the volume scoring that people here like to denigrate so much.

    That’s like saying hitting home runs is unimportant because the team with the most home runs rarely wins the world series.

  16. mokers

    DRed: That’s like saying hitting home runs is unimportant because the team with the most home runs rarely wins the world series.

    I think that is a bit more subtle than that. First, I don’t think you can equate a true outcome with a rebound. A better analogy would perhaps be the double play. Good defenses will complete more of them, but it having lots of double plays can also mean things happened you didn’t want, like somebody getting on base. It is totally possible that teams that do poorly generating double plays aren’t bad at winning baseball games. It could also be that teams that do poorly with their rebounding can still be good at winning basketball games.

    But I agree, SportVu isn’t going to help unless we have ready access to data. And we aren’t going to be able to really understand that much until we have many seasons of data.

  17. DRed

    mokers: I think that is a bit more subtle than that. First, I don’t think you can equate a true outcome with a rebound. A better analogy would perhaps be the double play. Good defenses will complete more of them, but it having lots of double plays can also mean things happened you didn’t want, like somebody getting on base. It is totally possible that teams that do poorly generating double plays aren’t bad at winning baseball games. It could also be that teams that do poorly with their rebounding can still be good at winning basketball games.

    But I agree, SportVu isn’t going to help unless we have ready access to data. And we aren’t going to be able to really understand that much until we have many seasons of data.

    There are many paths to the mountain top. Just because getting rebounds is good doesn’t mean you have to be good at rebounding. But if you’re bad at rebounding, you need to be better at other things.

  18. Donnie Walsh

    DRed: That’s like saying hitting home runs is unimportant because the team with the most home runs rarely wins the world series.

    Maybe, but is it false? (I’m not a baseball fan).

    Rebounds to me are more like serves in tennis. The defensive team is supposed to win them, so it’s not really worth tracking unless there is a “break” (ie an offensive rebound). Sure, a rebound here and a rebound there can be the difference between winning and losing, but in general, if a guy averages 10 rebs a game, there’s no way to know from a box score if any of them make a lick of difference to the outcome of a game.

    To me, rebounding seems like an easily quantifiable stat (unlike defense), and because it is easy to record, people put a lot of emphasis onto it. But really, it’s just the end of a good defensive possession, and shouldn’t be treated as anything more.

  19. Frank

    Yeah, that NBAGeek article was as reactionary an article as I’ve read — barely thought out, and totally a pot calling the kettle black.

    Re: uncontested rebounds — that just means there was no one within “x” number of feet of the ball other than Evans — ie. everyone else ran away. Why should he get any credit for those? And lest we all forget, someone has already done this analysis on the Reggie Evans / Brook Lopez effect:

    http://hoopchalk.com/2013/04/16/why-cant-brook-lopez-rebound-he-boxes-out-too-much/

    Now I have no doubt that Reggie’s a good rebounder, especially on the offensive end, where pretty much every rebound he gets is probably contested. But evidence is certainly mounting that he’s figured out how to game the system by attacking every single rebounding opportunity (and being good at it to boot) just as much as a volume scorer has figured out how to get $$ by shooting a lot.

    DRed: There are many paths to the mountain top. Just because getting rebounds is good doesn’t mean you have to be good at rebounding. But if you’re bad at rebounding, you need to be better at other things.

    lol that’s why there are four factors and not just “rebounding”.

  20. thenamestsam

    Brian Cronin:
    On an entirely other tangent, did y’all see that Chris Anderson got catfished? High-larious.

    One of the most confusing stories I’ve ever read. So there was a woman pretending to be him and sending sexually explicit messages to a 2nd woman on the internet suggesting they meet up, etc. And at the same time the 1st woman was also pretending to be the 2nd woman and sending Birdman messages. Then he actually did meet up with that 2nd woman and have sex with her? And somehow they never realized that they hadn’t actually been talking to each other on the internet but had both been individually communicating with a 3rd party? How is that possible? Honestly one of the most convoluted schemes you could ever dream up.

  21. thenamestsam

    As for the SportsVu data the bottom line is that anyone trying to draw any conclusions from the information that’s out there now is doing so because they have an agenda. The factoid about Lopez and Evans is undeniably interesting but it just isn’t close to enough information to draw any conclusions. It’s going to take seasons worth of data and some serious analysis to come to any sorts of answers about this stuff.

    The funny thing is that the NBA Geek article calls Schuhmann’s article “A textbook example of looking at the data and coming to the exact wrong conclusion about what it means.” Only Schumann never presents any conclusion at all other than “No single stat or number exists that’s going to tell you all you need to know about a player. Everything must be taken in context and the more information you have, the better argument you can make. Well, SportVU is a lot of information.” It’s the NBA Geek that dramatically overreaches and comes to an absurd conclusion (absurd based on the lack of support for it – it may be right but it will be years before there will be any support for it) all in “defense” of a position that hasn’t even been attacked. They embarrass themselves with articles like that.

  22. Frank

    thenamestsam: It’s the NBA Geek that dramatically overreaches and comes to an absurd conclusion (absurd based on the lack of support for it – it may be right but it will be years before there will be any support for it) all in “defense” of a position that hasn’t even been attacked. They embarrass themselves with articles like that.

    Yeah, that article was definitely embarrassingly poorly thought out.

  23. flossy

    Frank: lol that’s why there are four factors and not just “rebounding”.

    To quote a certain, much maligned Italian, “It’s called basketball, not rebound ball.”

  24. flossy

    thenamestsam: t’s the NBA Geek that dramatically overreaches and comes to an absurd conclusion (absurd based on the lack of support for it – it may be right but it will be years before there will be any support for it) all in “defense” of a position that hasn’t even been attacked. They embarrass themselves with articles like that.

    Eh, the NBA geek like the TownHall.com or World Net Daily of basketball analysis. It’s just an echo chamber for ideologues who are so far gone that they’ve formed an impenetrable bubble of basketball wingnuttery. Articles like that aren’t meant to convert, they’re preaching to the converted. Nor are their writers open to questioning their own orthodoxy.

  25. The Honorable Cock Jowles

    flossy: Nor are their writers open to questioning their own orthodoxy.

    Except when they revise their methodology, which they have.

  26. EB

    The Honorable Cock Jowles: Except when they revise their methodology, which they have.

    I would separate out NBA Geek and their articles from the methodology of WP. Outside the merit of WP is people’s adherence to WP. And regardless of what WP says about the value of rebounds it could obviously be modified to account for “junk rebounds” if we had a tally of those.

  27. DRed

    EB: I would separate out NBA Geek and their articles from the methodology of WP. Outside the merit of WP is people’s adherence to WP. And regardless of what WP says about the value of rebounds it could obviously be modified to account for “junk rebounds” if we had a tally of those.

    First we have to figure out what a junk rebound is.

  28. Jack Bauer

    DRed:
    You’d think once in 10,000 seasons Lebron would blow out his knee and the Heat would fall apart.

    Not wishing it upon him, but if it did happen to occur I wouldn’t be all that upset.

Comments are closed.