Checking 2012-13 NBA Win Predictions & Projections

23 Apr

Just before every NBA season most basketball writers, bloggers and authors try to predict what will happen in the next year which is usually fun stuff to read and discuss. But I’ve always felt there’s a half of the picture missing – accountability for those predictions.
Recently started to track single-winner picks which is great but I’d like to continue this blog’s “tradition” with comparing wins projected before the season to the actual ones after the season.
So which writer, blogger or author had the most accurate projections before the 2012-13 NBA season?

To answer this question I gathered predictions in early November from most popular sites and other googled sources [btw, it’s not a closed list so if you want to add your’s predictions let me know in the comments] and compared them to actual results by Root Mean Square Deviation which is essentially average difference between each prediction and result in terms of absolute value.
If you want to play with numbers or check any single entry just download NBA Predictions Accuracy 2012-13.pdf but if you are interested only in the results they are in the table below…

But please keep in mind two important notes:
1) don’t think or assume it’s all about the skills of people involved because randomness plays a part here… and it’s hard for the better example than Minnesota Timberwolves in 2012-13. Literally everybody thought they would be better this season but they were crushed by injuries and in the end the best prediction for this team turned out to be from… last years’ adjusted wins! So if someone predicted 45 wins for them his result was punished by bad luck more than someone who predicted 35 wins for them.
Well, unless someone can predict injuries…

2) in most cases entries next to each other are basically interchangeable in terms of accuracy because difference between them can be translated into “2-3 more/less wins in the right direction would flip them”.

Onto the table…

Author Date Source (data from) RMSE
John Hollinger 10/30/2012 5,94
Bovada updated 10/30/2012 5,97
Zach Harper 10/29/2012 6,27
ESPN forecast 08/20/2012 6,40
O/U Line 10/30/2012 6,47
Joe Schaller 10/30/2012 6,53
Cantor Gamings 10/17/2012 6,67
predictionmachine 10/15/2012 6,74
Joel Brigham 10/29/2012 6,78
Las Vegas Hilton 10/10/2012 6,81
George J Monroy 10/25/2012 6,88
Bovada early 10/12/2012 6,93
Matt Moore 10/29/2012 7,02
Sportsbook early 10/12/2012 7,17
Robert Eckstut 10/29/2012 7,19
Bobbo 10/23/2012 7,19
David Williams 10/30/2012 7,30
theNBAmodeller 10/30/2012 7,37
Bradford Doolittle 10/30/2012 7,41
DanielM ASPM 10/29/2012 7,42
version “dumb” 10/30/2012 7,42
Darryl Howerton 10/29/2012 7,43
Ed Weiland 10/31/2012 7,54
WoW1 10/30/2012 7,61
AccuScore ???? 7,80
WoW2 10/31/2012 7,87
TeamRankings Projections (BETA) 10/31/2012 8,12
Kevin Pelton 10/30/2012 8,16
David Hess 10/29/2012 8,18
SCHOENE 10/30/2012 8,19
Royce Young 10/29/2012 8,51
Derek Ayala 10/09/2012 9,19
ESPN The Magazine and Basketball Prospectus 10/18/2012 9,52
Wins in previous season 9,73

A couple of notes about the results in the 5th season measured:

– The most undervalued teams were Warriors [by an average of 11.1 wins], Rockets [10.7 – thanks Harden!], Grizzlies [9.1], Nets [6.8], Clippers [6.4], Knicks [6.1] and Heat [6.0]. I guess people underestimated how bad Eastern Conference will be… because teams like Nets, Knicks and Heat on any undervalued list are surprising.

– The most overvalued teams were aforementioned Timberwolves [by an average of 12.5 wins], Sixers [11.3 thanks Bynum!], Lakers [9.2 – again injuries], Celtics [7.3 – FA Class], Suns [7.1], Magic [6.5] and Cavs [5.9].

– I have to congratulate John Hollinger on many levels. Not only he won for the second time [after 2010-11] and beat Vegas AGAIN but his fame reached Memphis Grizzlies so they signed him to a deal. I’m sure his main responsibility there will be predicting wins for another season ;-) Unfortunately, we won’t be able to see it.

– last year’s “scouted gem” and “promising prospect” Joe Schaller from proved it was not a contract year and became a “respected veteran” in the group. OK, too many clichés in one sentence but congratulations for him are in order.

– wisdom of the crowd in form of ESPN forecast had 5th strong year in a row… and not only they noticed it but also created a NBA ESPN Forecast. I didn’t see that one coming but that’s an interesting twist for me.

– yes, over/under numbers from Vegas were still good.

– last year’s predictions had a surprisingly short list of competitors and I was a little worried about it but fortunately it was an outlier caused by the lockout. Hopefully more of them will continue to do it in future seasons.

– after last year’s strong showing from members of APBR forum this season was clearly a disappointment.
I guess the moral of this story is “you have to participate to win”…

– I didn’t include predictions from Bill Simmons because on his podcast with Joe House for some teams they both only said “over” or “under” without any specific number.

Have I missed anything?

Anyway, I’ll see your predictions next year!


Posted by on April 23, 2013 in Expanding Horizons


Tags: , , , , , , ,

28 responses to “Checking 2012-13 NBA Win Predictions & Projections

  1. Andrew

    April 24, 2013 at 05:55

    What would this come out to? This was mine, from October…

    Miami 64
    Boston 56
    New York 53
    Brooklyn 50
    Philadelphia 47
    Indiana 46
    Atlanta 44
    Chicago 41
    Toronto 40
    Milwaukee 36
    Detroit 33
    Washington 28
    Cleveland 25
    Charlotte 23
    Orlando 19
    L.A. Lakers 66
    Oklahoma City 63
    Denver 53
    Memphis 49
    L.A. Clippers 47
    San Antonio 46
    Utah 44
    Minnesota 42
    Golden State 41
    Dallas 38
    Sacramento 37
    Phoenix 31
    New Orleans 27
    Houston 23
    Portland 19

    • wiLQ

      April 24, 2013 at 19:22

      “What would this come out to?”
      RMSE for those predictions is 8.55…

  2. thenbamodeller

    April 24, 2013 at 13:52

    For those interested the theoretical limit on the RMSE is roughly 4.5 in the NBA, for comparison.
    Success in predictions in single seasons don’t carry much meaning, a gradual view must be built.
    I suggest that predicting total point margin for a season would be a much better test of predictive ability as it contains less random error than wins due to one result being granular and near-continuous whist the other is binary and discontinuous.

    The Lakers weren’t misvalued because of injury, they’re roughly a 50 win team at full strength. Age and poor roster construction ensured they’d never be great.
    V-zero and thenbamodeller are the same, so you may as well choose one or the other – I don’t need two entries. :P The Twolves were horribly injury prone, completely screwing possibly good predictions, and Asik was undervalued by the market for the Rockets.

    Anyway, thanks for all the info – I’ll be back next year @ thenbamodeller blog, to which I’ll also be posting some playoffs stuff soon (ate, obviously) and other devvelopments I have made over the last six months.

    • wiLQ

      April 24, 2013 at 19:27

      “Success in predictions in single seasons don’t carry much meaning, a gradual view must be built”
      That’s why I try to collect more and more data. 5 seasons is a nice start and there seems to be some consistency among authors. I just wish more of them did it each year.

      “I suggest that predicting total point margin for a season would be a much better test of predictive ability as it contains less random error than wins due to one result being granular and near-continuous whist the other is binary and discontinuous.”
      Very true but that would require a change from predictors themselves – they don’t predict margins but wins so even though technically we could translate it I don’t want to put numbers into their mouths.

      “V-zero and thenbamodeller are the same, so you may as well choose one or the other – I don’t need two entries. :P”
      ;-) Thanks for the info. I fixed it and removed entry from “v-zero”.

  3. Carlos Petry

    April 26, 2013 at 01:08

    These were my predictions before the season:

    Thunder 58
    Spurs 57
    Lakers 55
    Nuggets 52
    Clippers 51
    Grizzlies 48
    Jazz 43
    Wolves 42
    Mavs 41
    Warriors 37
    Rockets 36
    Suns 34
    Blazers 31
    Kings 30
    Hornets 29

    Heat 62
    Celtics 53
    Knicks 47
    Pacers 46
    Nets 45
    Hawks 44
    Bulls 43
    Sixers 39
    Raptors 36
    Bucks 34
    Pistons 34
    Wizards 29
    Cavs 28
    Magic 25
    Bobcats 21

    What would the RMSE be for these?

    • wiLQ

      April 26, 2013 at 18:15

      5.67! I wish I knew about it before or during the season.
      Please let me know when you publish predictions for next year!

      • Carlos Petry

        April 26, 2013 at 18:47

        Wow, I beat Hollinger! That’s awesome! So now it’s just a matter of time before I get a front office job in the NBA, right?

        You’ve probably noticed that my blog ( is in Portuguese (and has been inactive for a while). But if I do make predictions for next season, I’ll make sure to let you know. Even if I don’t, I might send you them just for fun.

        • wiLQ

          April 26, 2013 at 18:56

          Yes, I noticed it – the language was probably the reason I didn’t find – and it felt a little weird it was your last post but congratulations anyway. Show everyone next year it wasn’t a fluke ;-) Thanks in advance for sending them.

          And we don’t know why exactly Hollinger was hired, maybe it was because of his writing skills ;-)

          • Carlos Petry

            April 26, 2013 at 19:24

            Yeah, that was the last post I wrote there. Unfortunately, as you can imagine, blogging about basketball does not make you any money in Brazil. I posted daily previews (and actually had some decent traffic) in October, but then I got a new job that takes way too much of my time for me to able to keep the blog updated. I also stopped tweeting (@Siga_NBA) shortly thereafter. I do hope I can find the time to start writing again in the future.

            Anyway, I’m pretty sure it WAS a fluke! My predictions weren’t analytics-based at all, so I really don’t think I’ve found a winning formula or anything like that. More than likely, I just got lucky. Still pretty cool, though!

            • wiLQ

              April 28, 2013 at 15:08

              “Anyway, I’m pretty sure it WAS a fluke! My predictions weren’t analytics-based at all”
              Oh, that’s a shame… but it doesn’t change the fact it was pretty good guess ;-)

  4. Justin

    April 27, 2013 at 01:40

    I calculated my RMSE as 7.73. You can double-check if you want:

    I’m surprised it’s not more terrible considering I didn’t put too much work into them and I was destroyed in the west (Lakers, of course, but Curry was healthy all year, Denver was better than I thought, and Phoenix and New Orleans threw me off — Nash leaving probably killed morale and Ryan Anderson slumped a little, while Anthony Davis despite nice stats didn’t have a good impact.) My RMSE in the west was 8.90 and in the east 6.35.

    Also, for Wins Produced, were you using their decimal predictions or did you round up? I think a prediction of 57.14 wins is pretty stupid and goes against what we’re doing. I calculated the rounded forms for WOW1 at 7.71 and WOW2 at 7.91.

    • wiLQ

      April 28, 2013 at 15:16

      I double-checked and I have the same result for your predictions. Thanks for the link.

      “Also, for Wins Produced, were you using their decimal predictions or did you round up?”
      I used a form in which author published it…

      “I calculated the rounded forms for WOW1 at 7.71 and WOW2 at 7.91.”

      … but as you can see it doesn’t change much and in this case rounded numbers were worse!!!

      “I think a prediction of 57.14 wins is pretty stupid and goes against what we’re doing. ”
      Why? It’s like hedging a bet between 57 and 58 wins. Also note that all Vegas’ lines have half-win and it doesn’t affect their accuracy.

    • thenbamodeller

      April 29, 2013 at 13:47

      If you report rounded numbers then you ought to report the mode. However, the mode is not the best estimator in this case, the mean is, so reporting the mean (decimal) form is obviously right.

  5. Leszczur

    April 30, 2013 at 13:23

    The funny thing is Wages of Wins forecasts seem to do considerably worse job than Hollinger.
    Surely it would take more than just one season of data to conclude anything of value about their methods, but I guess John has his small win after all.

    • wiLQ

      April 30, 2013 at 15:12

      But is it even a semi-fair comparison of methods? Maybe Hollinger is just better at predicting minutes or injuries… IMHO the more interesting question is how much did Hollinger use PER for his predictions?

      • Devin

        May 13, 2013 at 16:23

        Or trades, for that matter. Or, in the case of Wins Produced, position played, which in some cases could matter a great deal.

        • wiLQ

          May 13, 2013 at 16:36

          To be fair, Wins Produced made positions played matter more than any other tool so you kind of created your own additional problem with inability to predict them.

          But overall, there are so many variables at play here that without entire dataset responsible for predictions it’s hard to know what exactly went wrong and where: But you can investigate your own mistakes which I wish more people would do. Hopefully this series will help in some way.

  6. julienrodger

    May 24, 2013 at 07:03

    My predictions &

    They were made before the Harden trade so my HOU prediction is way off, but:

    SAS 61 (3)
    OKC 57 (3)
    LAL 57 (12)
    LAC 51 (5)
    MEM 49 (7)
    UTA 48 (5)
    DEN 47 (10)
    GSW 42 (5)
    POR 42 (9)
    DAL 41 (0)
    MIN 38 (7)
    SAC 27 (1)
    NOH 24 (3)
    PHX 22 (3)
    HOU 19 (26)

    MIA 67 (1)
    IND 53 (4)
    BOS 52 (11)
    NYK 46 (8)
    BKN 46 (3)
    CHI 45 (0)
    PHI 43 (9)
    ORL 41 (21)
    ATL 37 (7)
    DET 35 (6)
    WAS 33 (4)
    TOR 33 (1)
    CLE 32 (8)
    MIL 29 (9)
    CHA 28 (7)

    I calculate 6.6 which ranks pretty decently. If I eliminate OKC and HOU because of the Harden trade and calculate the other 28, I get 6.04.

    • wiLQ

      May 30, 2013 at 12:34

      I also have 6.6 for your simple average but in terms of RMSE your score was 8.6.
      And Rockets screwed almost everybody so you can easily improve your rank by removing them ;-)
      BTW, do you realize your predictions added up to 1245 wins?

  7. PrimetimeShabo

    June 12, 2013 at 09:10

    I picked the Spurs to beat the Heat one year in advance. I also picked OKC to beat Miami last year (one year in advance). In the last 10 years, I’ve picked at least one of the NBA championship teams correctly one year in advance. I always make my predictions for the upcoming year as soon as the last championship game ends.

    • PrimetimeShabo

      June 12, 2013 at 09:11

      Sorry, I meant I picked at least one team correctly during each of the last 10 NBA championships a year in advance. Often, I pick both correctly.

      • wiLQ

        June 12, 2013 at 14:16

        That’s great but that’s not what I did here, do you do win predictions too?
        Then let me know before next season what are they ;-)

  8. Mike Lu

    September 11, 2013 at 06:56

    Hi, I’m interested in submitting my 2013-2014 predictions to this tracker. Could you add me to the next iteration? My predictions are up already here:

    Alternately, if you’d like it to be e-mailed, I can certainly do that. Much appreciated!

    • wiLQ

      September 11, 2013 at 13:05

      Your predictions are already saved and ready to enter ;-) Thanks!
      BTW, if you make some changes down the road let me know before the season [not after every single correction, just final version]


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s