Announcement

Collapse
No announcement yet.

Raptors 2014-15 Projected Wins

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Raptors 2014-15 Projected Wins

    Warning: This post contains math and basically just uses stats.

    Hi all.

    Thought I'd take a quick run at trying to see where this team projects going into this year using a few statistical methods. The primary rotation seems to be pretty set at this point, and any third string PG or backup C won't skew these numbers too much, I don't think.

    For all of these methods, I need to make assumptions on the minutes distribution. To do this, I first set an ideal rotation based on my personal preferences (which I think are reasonable). See below.

    PG: Kyle Lowry (30), Greivis Vasquez (18)
    SG: DeMar DeRozan (30), Vasquez (6), Louis Williams (12)
    SF: Terrence Ross (30), DeRozan (6), James Johnson (12)
    PF: Amir Johnson (18), Patrick Patterson (30)
    C: Jonas Valanciunas (32), Amir (10), Chuck Hayes (6)

    Leaving the following minutes breakdown:

    Lowry: 30
    DD: 36
    Ross: 30
    Amir: 28
    JV: 32
    Vasquez: 24
    Patterson: 30
    Williams: 12
    JJ: 12
    Hayes: 6

    But that is a perfect scenario. The reality is, games will be missed due to injury, and our backups and third stringers will see an increased role. So the next step was estimating the number of games to be missed. This is a complete crapshoot of course. But to get a good guess, I took the average number of games played over the past 3 seasons (yes, correcting for the 66 game season in 2011-12) and used that as the number of games played for the primary rotation. Then I assigned the backup minutes accordingly to fill in the gaps (assigned entirely to 3rd stringers, to be conservative). The result can be seen below.



    The third strings like Fields and Hansbrough end up with a lot of minutes due to my assigning them all the injury minutes. I built this using WS at first, and this was the most conservative approach (and the simplest) so I stuck with it. You'll be able to see the impact in the win totals in each section. BeBe and Bruno I just used for cleaning up minutes from rounding elsewhere - they contribute 0 wins in all the models (except xRAPM, where they contribute average wins).

    ----

    Win Shares:

    So, using last year's WS/48 numbers, and the total minutes I approximated above, we can get a WS projection for this coming season.



    Total WS predicted: 53 wins. 53-29 record. Very nice.

    -----

    Wins Produced:

    Again, using last year's WP/48 numbers, and the total minutes I approximated above, we can get a WP projection for this coming season.



    Total WP predicted: 55 wins. 55-27 record. Even nicer.

    -----

    xRAPM:

    A little more complicated this one. First, we grab all the offensive and defensive xRAPM numbers for each player (that's an adjusted plus-minus in terms of impact on the team regardless of teammates, opponents, team effects). They are presented in a points per 100 possessions impact. So a net impact on offense, and one on defense. We then apply to these the pace the Raptors played at, so get each player's impact over the length of a game (48 minutes). The Raptors played at a 91.5 pace last year. So that suppresses the numbers somewhat. Then you take that points/48 minutes number, and add the impact to the league average points per game (101). That gives each player an ORTG and DRTG based on xRAPM. Then we can apply pythagorean wins to predict the wins generated by each player per game.



    Total RAPM wins predicted: 47 wins. 47-35 record. Not as nice as the other two. But still a very nice low water mark.

    -----

    Final prediction: I'll just use a simple average of the three systems. That means 51-31. 50 wins here we come.
    Last edited by DanH; Tue Jul 15, 2014, 07:37 PM.
    twitter.com/dhackett1565

  • #2
    Wow this is great man, and tbh it's a pretty objective and statistically supported method.

    How accurate are WS and WP generally in predicting records?

    Comment


    • #3
      50-52 wins is what I'm going with.

      Comment


      • #4
        It's also interesting to see that DeRozan is actually expected to produce more wins than Lowry (slightly). Obviously he plays more, but that is still very interesting.

        Comment


        • #5
          imanshumpert wrote: View Post
          Wow this is great man, and tbh it's a pretty objective and statistically supported method.

          How accurate are WS and WP generally in predicting records?
          I don't have a link to any specific studies, and to do one league wide would take some time.

          But I recall doing a comparison of several teams using their prior year's numbers (and actual minutes) and most came within 2 or 3 wins, with a few outliers having greater differences (last year's Raptors team is a great example - WS underpredicted their win total by 7 wins).
          twitter.com/dhackett1565

          Comment


          • #6
            imanshumpert wrote: View Post
            It's also interesting to see that DeRozan is actually expected to produce more wins than Lowry (slightly). Obviously he plays more, but that is still very interesting.
            Yeah, it is primarily because of the minutes he plays, as he is not great in any of the per minute stats (he is solid in WS, about average in the other 2).
            twitter.com/dhackett1565

            Comment


            • #7
              DanH wrote: View Post
              Yeah, it is primarily because of the minutes he plays, as he is not great in any of the per minute stats (he is solid in WS, about average in the other 2).
              Yeah it's a bit like the Iverson effect I guess. Not great per 48 numbers, but the ability to log heavy minutes while playing almost every game and doing a good chunk of the scoring for the team. Obviously DD is nowhere near Iverson's usage, but the concept is similar I suppose.

              Comment


              • #8
                I like 55 wins
                For still frame photograph of me reading the DeRozan thread please refer to my avatar

                Comment


                • #9
                  I'd be happy with 49 wins.. that's progress and beats our previous record. I can see 50 wins but the fate of the team falls on Lowry's shoulders again. I hope he plays like he did last year and is injury free.

                  Comment


                  • #10
                    I love it! I think Williams and JJ get a little bit more minutes but other than that well done.
                    I relish negativity and disappointment. It is not healthy. Somebody buy me a pony.

                    Comment


                    • #11
                      72-10 book it
                      For still frame photograph of me reading the DeRozan thread please refer to my avatar

                      Comment


                      • #12
                        Great post. This is what a good thread looks like.

                        Comment


                        • #13
                          I think 45-48 is a pretty reasonable mark. The east is more balanced this year (still as bad), so it may mean a few losses here and there.

                          Comment


                          • #14
                            I like 51 wins.
                            The name's Bond, James Bond.

                            Comment


                            • #15
                              55-57 if everything goes well, I predict 55

                              Sent from my HTC One using Tapatalk

                              Comment

                              Working...
                              X