Tweaking the Championship Equation,2011 Contenders rev1.1 and Young Stars

Yesterday, I came up with the Championship Equation:

  • Win 52 or more games (Houston is an aberration that can be explained in one word: Hakeem)
  • Have two star points (either >2 Stars, > Star + Superstar or > 2 Superstars)

Do this and win the title. It was good.It wasn’t perfect . Therefore,today, I want to do some tweaks and refinements to the equation, talk about some of the special cases (and how that impacts some of the contenders) and spend some time talking about Young Stars (>.200 WP48) at 21 Years old or Younger.

Basics (skip if familiar)

This article uses Wins Produced and WP48 [Wins Produced per 48 minutes] to evaluate player’s performance.* This measure uses three key components to evaluate a player:

  • The player’s per minute box score statistics
  • The player’s team’s per minute box score statistics
  • The average performance at the player’s position (PG, SG, SF, PF or C)

A full explanation can be found here. To give a general scale, an average player has a WP48 score of 0.100. The very best players in the league usually have a WP48 over 0.300. To put this in perspective; an average player who plays a full season at 40 minutes a game would generate around 6.83 wins for their team.  In contrast, a player posting a 0.300 WP48 would generate about 20.5 wins at 40 minutes a game over an 82 game season.

I may also talk about the half-baked notion and Wins over replacement Player (WORP).

Refining the Numbers

For the initial version of the exercise I looked at the following for each championship team:

  • Wins in the regular season
  • Qty of Players in the Playoff Top 6 in Minutes who were Stars (.200-.300 Wp48) in the Previous Regular Season, that Regular Season and the Playoffs
  • Qty of Players in the Playoff Top 6 in Minutes who were Superstars (>.300 Wp48) in the Previous Regular Season, that Regular Season and the Playoffs
  • Star Points (1 for a Star, 2 for a Superstar)

This made for some interesting close calls (Miami in 2006, Houston in 95). I decided therefore to round the WP48 for each player to the nearest .01 (so .295 becomes .300) and double check the position adjustment. With that done the data looks as follows:


The data set looks cleaner and only Houston in 1994 and 1994 present glaring exceptions (but we’ll get to those).

Exceptions, Corrollaries and Addendums:

Let’s go case by case on these. Remember that I’m focused on actions that lead to immediate turnarounds.

  • The first is the Lakers in 1980 (and the Celtics in 1981). They drafted some guys named Magic Johnson and Larry Bird and went from an also ran (47 wins, 1 Star (Norm Nixon), 1 Superstar (Kareem) for the Lakers, 29 wins, 1 Borderline Superstar (Cedric Maxwell)) to a Dinasty (60 wins,2 Superstars for the lakers, 61 Wins , 2 superstars for the Celtics). The lesson that while typically a draft pick won’t yield immediate results  is that if you have a team with 1 superstar on your roster finding a superstar in the draft can yield immediate results. Just don’t hold your breath for this happening. Well call this the The Magical Legendary Exception.
  • Bad players (<-.01WP48) in the playoff rotation of championship teams (top 6  in minutes played).  The list for these is as follows:

11 Guys on 9 teams. The rule of thumb here is that you need two star points for every playoff sink hole. We’ll call this the Mr. Eva Longoria Rule.

  • Houston in 1994. They did not quite have a superstar (Hakeem clocked in at .280). However they had 4 other players in their playoff rotation come in between .128 and .159 (Thorpe,Horry,Kenny Smith and Sam Cassell). So it’s  possible to win if your superstar is a little off if you have  4 other guys who can step up and you get some ridiculous three point shots and the best player in the league decides to retire in his prime. We’ll The Impossible Dream Scenario.
  • Houston in 1995. They only won 47 games in the regular season but they went off and got another star (Drexler) to complement Hakeem at mid-season. So you can trade yourself into contention (but it’s not very likely unless Chris Wallace or Kevin McHale is prominently involved). We’ll call this  The Trade exemption.
  • Players can step it up in the playoffs and they can also step it down. Star/Superstar depth is really important. The last 4 champions have featured at least 4 guys at greater than >.150 WP48 in the regular season and it not always the same people who come thru (Posey for the Celts in 2008 is a good example). The one superstar approach is a much riskier proposition it’s always better to have the depth than to not have it and regret it (see Posey and the Celts in 2009). We’ll call this Posey-Horry’s Law.

Now that we’ve refined the math and laid out some additional ground rules, let’s look at the stars and the contenders for 2011.

Stars and Superstars in 2010 Take 2:

Compiling the List of stars and superstars in 2010 is still a simple process. We find every player who played more than 400 minutes last year (sorry Yao) and posted a WP48 of greater than .200. The difference now is that I’m going to use Prof. Berri’s numbers (which have slightly better position adjustments) and I’m also going to round to the nearest .01 WP48 (. Using these numbers there were 39 players who met that criteria in 2010 and 12  were superstars (and worth 2 star points) and 27 were stars (and worth one star point).


The list rings truer now to me. The biggest question mark here is Oden but that’s actually extremely appropriate.  The two under-21s on this list (Batum and Blair) project out as future stars (and maybe even supertars). Why do I say this? Let’s look at a list of star Players (>.200 WP48) at 21 Yold or younger with at least min 800 minutes played.


I don’t know about you, but if I’m a GM with a player on that list, I Get down on my knees and thank the appropriate deity profusely.

We respect all denominations on this blog

2011 Favorites/Contenders/Pretenders rev 1.1:

If we look at every team’s current roster again with our new numbers and tally their projected wins and the number of star points we can divide the league into those teams that have a shot at the title and those who don’t. For this version, I will remind you that I’m using the Build 0.1 projections and not the updated final projection (see the Bulls example). For this version, I am dividing teams in 4 groups:

  1. The Clear Favorites those teams with more than two Star points and projected to win 52 or more
  2. The Contenders those teams with two Star points and projected to win 52 or more. These teams will need some luck to go their way.
  3. The Pretenders with Talent. Teams with the stars but not the wins (Sorry CP3)
  4. The Pretenders. Teams without the star points or the wins


So now we see four clear favorites, four contenders, six teams with the stars but not the wins and 16 pretenders. The Heat, Blazers, Lakers, Spurs project as the first elite group (and that really makes a lot of sense given their rosters). Of this group, the Heat seem poised in the East.

The Lakers look to have a tough road ahead out west but not from the expected challenger (OKC) but from the Spurs and the Blazers.

The  Mavs, Celts,  Bulls and Magic form the second group of contenders with the Eastern teams having more realistic shots given the distribution of the elite teams (sorry Mr. Cuban).

The pretenders with stars all could have a chance but they need to find some surprise wins (via the draft, free agency or trades), given their track records however, most of these teams are more likely to get fleeced than strike it rich. Golden State will move up in wins when I do the final build, so I’m making a judgement call and calling them a contender.

Of the pretenders, the Nuggets, Thunder and Suns are the most interesting. They don’t currently have the stars on their rosters but a couple of factors come into play:

  • Trading non-star Melo for one or two stars based on his reputation could make Denver the Fifth member of the Favorites group (granted it would be in the crowded west).
  • Josh Childress is a strange case. I can’t find a comparable example of a star going of and playing in Europe in the middle of his prime. Phoenix with Childress sneaks into the contender group.
  • The Thunder are the most fascinating. Durant barely missed the superstar cut. Their five best guys look like this : This is strangely reminiscent of the 94 Rockets. I like this team and think they’re a Year away but I’m willing to entertain them as a contender.

So the final tally is as follows:

4 Favorites, 7 Contenders and one possible contender pending the Melo deal. Again adjust your wagers accordingly but I would say the best values are the Blazers, Spurs and Warriors.

Written by:

65 Comments

  1. tfrab
    9/30/2010
    Reply

    speaking of san antonio: what about tiago splitter? which value of WP48 do you expect from him?

    • 9/30/2010
      Reply

      I think he’ll be at starter level (>.100 WP48). My concern is that he’ll keep Blair off the court but I trust Pop to do the right thing.

  2. Evanz
    9/30/2010
    Reply

    The Warriors should have two star points, right? Lee and Biedrins. They did yesterday, anyway.

    • 9/30/2010
      Reply

      Yes, they should. Damm excel :-). I’ll fix it. That puts the Warriors squarely as a contender.

  3. Xon
    9/30/2010
    Reply

    Why did Josh Smith plummet from .299 in the earlier version to .237 in this version? Just curious, not a passionate Atlanta fan because Atlanta fans are only passionate for the sweet, sweet release of death.

  4. Chicago Tim
    9/30/2010
    Reply

    I guess the point of your updated final projection for the Bulls is that, after factoring in minute allocation, expected individual improvements of young players, and benefits from playing in position with better teammates, the Bulls deserve a bump up to “favorite.” Although in the East, that’s a relative term, since they still have to face the Heat.

    But even before factoring in all these other matters, there’s also the fact that Boozer and Noah both fell just short of superstar status, while Deng and Korver are solidly above average, Watson and Gibson above average, and Rose average with great potential. As the exception for Houston shows, the arbitrary cut-off for stars and superstars risks ignoring the value of players who are solidly above average, solidly above star status, or (in rare cases) solidly above superstar status.

    • 9/30/2010
      Reply

      Much like the Thunder, I’d like their odds better in 2012 than 2011. The 4 Favorites are monster teams. The Heat and Lakers go without saying. Portland can’t have bad luck every year. Pop won’t waste Blair.

      • Chicago Tim
        10/1/2010
        Reply

        What? Are you backing off your 63 win prediction for the Bulls in 2011? Or are you saying it could be even better in 2012? Are you predicting more than 63 wins for the Blazers, Spurs, and Lakers? Or are the standards different because they can reach the finals without playing the Heat? If they are all 63-win teams, why wouldn’t the Bulls be considered a “monster” team, even if they can’t get past the Heat?

        • 10/1/2010
          Reply

          Not backing off at all. For the Playoff, you need at least two star points. They have less points and so the risk with them is greater than with the four favorites (Also, i think i’d like the over/under on wins more than the win it all bet). I think 2012 allows for a third guy to step into he star role for the Bulls (or for Noah to go Superstar).

          As for the win numbers for everyone else? You’ll have to wait for a bit.

          • Chicago Tim
            10/1/2010
            Reply

            Fair enough. I think there is a real chance for improvement across the board, but we’ll see. I hope there’s no lockout in 2011-12.

  5. some dude
    9/30/2010
    Reply

    the Warriors a contender? All because of Biedrens and Lee? Come on now…

    Where did Boston project based on this last season, because you have em as a contender and not a favorite (out of curiosity).

    Portland as a favorite doesn’t smell right. People are sleeping on them, but so long as Nate McMillen has that team playing at a snail’s pace, I don’t see it. And with Oden/Camby it’s more like 1 star than 2 because they won’t play much together (and I don’t think Oden’s minutes will be up much at all until the end of the season).

    Spurs are an interesting case. Splitter is huge, but injuries + age is a major concern there. Who knows when Manu or Timmy hits a wall? Blair is solid, but he doesn’t play because of his defense and Pop demands more from him. Let’s see if he learned some over the summer.

    Subjectively, I’d swap Boston and Portland as the favorites.

    Phx should also be nowhere on this list, Childress or not. That team might end up being the worst rebounding team in NBA history (surpassing last season’s Warriors). They’re going to get pounded on the glass big time and have no way of dealing with it without a trade. This is one situation where WP48 doesn’t matter because they have no power forwards on the team. It’s like a team of Deron-Cp3-Kidd-Rose-Collison. Sure, their WP48s will look awesome, but those 5 PGs would get obliterated (I’m being extreme to make a point). Losing Amar’e and Lou will be big for Phx. 8th seed would be a good year.

    I also think any melo deal will not bring back enough for Denver to be a contender right now (which means I don’t believe someone like Noah is going there).

    And finally, Clips are almost last on your list. What’s Griffin’s projection? I feel like they’ve gotta be better than where you got them right now.

  6. Evanz
    9/30/2010
    Reply

    some dude,
    I’m sure we don’t have to explain to you the difference between a sufficient and necessary condition. Right?

  7. some dude
    9/30/2010
    Reply

    Nah, Evanz. I was merely giving personal opinions on the topic of contenders as well as some criticisms here. I know the difference.

    I should say Arturo is doing good work. Numbers are numbers.

    • 9/30/2010
      Reply

      sd,
      Thanks for the kind words. Keep it up. The questions and comments inspire the analysis. And no matter the basement the calculations are being done in, numbers are numbers :-)

  8. 9/30/2010
    Reply

    SD,
    As I said before, I will be redoing the projections before the season begins (and rookies will be included). All your concerned will be addressed. That said, You gotta let me save something for the future :-)
    But in short:
    Warriors are very interesting but very definitely second tier (it’ll be interesting to see Lee on a good team).
    Last I checked, Boston still has Rondo and Garnett on the squad, a lot of veteran good pieces, and a ridiculous amount of size and Rasheed’s contract. But they’re still old and brittle.
    For Portland, I believe in Camby and I now believe in Nick Batum. Oden is really a bonus piece.
    Need to crunch phoenix fully but their roster doesn’t have a lot of dead weight on it.

    The current strong melo rumor (Iggy) would have Denver meet the necessary conditions to be considered a contender.

    The Clipps are better (I haven’t included rookies yet)

    • some dude
      10/1/2010
      Reply

      okay, cool. I was just wondering about the rookies because the Clips spot shocked me.

      As I posted in berri’s site, don’t believe the rumors.

      Warriors are a good team, eh! I still say they finish at 35 or below. Can’t wait to see this one play out.

      True about Boston, but I think it’s safe to say their regular season doesn’t matter. If healthy in April, they’re a favorite, IMO.

      Portland – well, I don’t believe in Camby or Oden’s knees. But I do believe in Batum enough. Hey, shouldn’t we expect Camby to decline badly, soon? and he’s pretty injury prone, too. How many minutes did you give him? I’d ballpark it at 30mpg for 67 games.

      Phx – I don’t care how goo their pieces are, how will they rebound? have you seen their lineup?

      Lopex < 16%
      Childress < 15%
      Turk < 16%
      warrick < 14%
      Frye < 17.5%

      Amar'e and Lou were around 20%. Grant Hill is about 16% but should decline at his age. Now, I assume those players will raise their rebound %s a bit for obvious reasons, but not enough to make up even half the disparity.

      This was the 2nd worst defensive rebounding team last year and they lost their only 2 good rebounders (and calling Amar'e good is…well let's move on).

      Now, Phx gave up 107 pp 100 poss last season. giving up more offensive rebounds and losing their shot blocker in Amar'e will hurt that (god, I just praised Amar'e's D?). Fry's defense is really horrible and so will Turk be as a PF.

      hedo and warrick are also well below Amar'e and Lou's OReb rate, so take away extra shots from that offense. So while Childress and Turk and Warrick offensively during a play can replace what Amar'e + Lou do, they can't get the extra shots for the team and they will give up more shots.

      Expect Phx to struggle to be above .500 next year.

  9. some dude
    10/1/2010
    Reply

    One more clarification on Phx. They were as good as they were last year because of having one of the best offenses ever. I don’t expect that to be recreated (ie I expect across the board drops in WP48). No way Frye and the team keeps that 3pt% so high.

    With a modest drop in offense and a drop in defense, expect next season to look similar to the Raptors. High scoring, no D, barely around .500 all year. they might fare a bit better than the Raps, however, because they won’t quit, but the results won’t be that much different.

    You cannot win in the NBA if you do not rebound. Heck, it’s even important in compiling WP48! What I’m saying is this:

    You need players with high WP48, but HOW you accumulate your wins matter and you cannot win if everyone does the same thing. It has to be a balanced approach. So some players need to rebound to get a high WP48, others to score efficiently, others to pass and not TO the ball. You can’t win with 6 players who can shoot efficiently but not rebound.

    • Evanz
      10/1/2010
      Reply

      “So some players need to rebound to get a high WP48, others to score efficiently, others to pass and not TO the ball. You can’t win with 6 players who can shoot efficiently but not rebound.”

      I don’t think that can happen, because a big guy will be penalized for not rebounding. Can you think of a big man with high WP48 (let’s say > 0.2) that *didn’t* rebound?

      • some dude
        10/1/2010
        Reply

        Well, that’s sort of my point. Big guys don’t have the ball as much by the very nature of the game, so to have a >.2 WP48 they MUST be rebounders. Phx doesn’t have rebounding bigs and also don’t have >.2 bigs that I’m aware of.

        In fact, do they even have a .15 big who played 30+mpg? Has any team ever won a title without that?

        What I was basically saying is, you need to have a balanced team which means a mix of bigs, smalls, and wings. Phx is basically like the Warriors last year. Lots of wings and smalls, but no bigs.

        • Evanz
          10/1/2010
          Reply

          Agreed, but I’m not sure why you’re focusing on Phoenix, anyway. They only have 1 star point, and are not listed as contenders by Arturo’s system.

        • 10/1/2010
          Reply

          88 LA and 97 Bulls. But both had a .140 guy playing the 4 or 5 (Rodman & Worthy). I think you found a a third necessary condition (and one that gets rid of the Suns).

    • 10/1/2010
      Reply

      Phx will be a great case study. I’m following the straight WP model and Ty over at Courtside Analyst is doing his Marginal Wins Score per 48 (which includes defense). I’m dying to find out what the weakness/strengths of the models are. Yet another reason to be excited by the upcoming season.

  10. Evanz
    10/1/2010
    Reply

    I would be very interested to look at the other side to this equation, that is, the teams that lost in the Finals or Semi-Finals. Did they have fewer star points? What was the biggest upset of all time according to star points? What was the most number of star points ever for a team that *didn’t* make the playoffs?

  11. Evanz
    10/1/2010
    Reply

    I think one change that I would make to this equation is that star points can only be awarded to starters or maybe the sixth man, as per Arturo’s Half-Baked theory. Let’s face it, Matt Barnes, as good as he is when he comes off the bench, is not going to get more than 20 mpg. He’s going to be the 7th or 8th guy off the bench after Blake and Odom.

    • 10/1/2010
      Reply

      Except that you’d be betting on Phil playing the wrong guys. I’m not taking that bet.Ithink come playoff time the best 6 Lakers that are available will get the playing time. Freaking Phil Jackson :-)

      • Evanz
        10/1/2010
        Reply

        true enough, but doesn’t Odom come off the bench, too? You’ve got Artest, Odom, and Barnes at SF. There’s a few minutes for Barnes at SG, but not many. I just wonder how will he get the minutes…

        • 10/1/2010
          Reply

          He’s Bynum/Kobe insurance in a way. If Bynum’s out : Blake,Kobe,Barnes,Odom,Gasol (Artest off the bench). If Kobe’s out: Blake,Barnes,Odom,Gasol Bynum (Artest off the bench). I suspect Barnes will wind up the 6th man in the playoffs and Artest 7th.

          • Evanz
            10/1/2010
            Reply

            Well, there’s a simple way to test this. Did any of the star points on the championship teams belong to role players (i.e. 6th men)? I’m guessing that the answer is either a) No or b) Yes, but only on teams that had star points to spare (say, >4).

            • 10/1/2010
              Reply

              Cooper in 82, Walton in 86, BJ Armstrong in 91, Posey in 08.

              Cooper and Posey are the most significant. Did I mention that I really hated when the Celts let him go?

              • Evanz
                10/1/2010

                Wasn’t Cooper a starter? He had the 5th most minutes that season.

              • Evanz
                10/1/2010

                nm, Rambis had the bulk of the starts. Clearly, though, by having the 5th most minutes (~2200 MP) on the team (over 1000 more than Rambis), Cooper was more than a 6th man.

              • 10/1/2010

                Agreed. He got minutes (same as Posey). I think Phil will find Barnes the minutes if he’s the best option .George Karl would totally bugger it but not Phil (because he’s Phil dammit).

              • Evanz
                10/1/2010

                yeah, and not that Phil would care about such things, but Barnes will definitely be a fan favorite in SoCal where he went to school.

        • some dude
          10/1/2010
          Reply

          1. Bynum is the better player when healthy (don’t dare quote WP48. Odom’s WP48 isn’t accumulated consistently. One game he’s amazing, another he’s Morrison).

          2. Gasol’s much better as a PF than as a C.

          Fisher-Kobe-Artest-Gasol Bynum was a MUCH better unit that Fisher-Kobe-Artest-Odom-Gasol during the regular season by the numbers. playoffs switched, but bynum was on one leg.

  12. some dude
    10/1/2010
    Reply

    Arturo, when am I going to get thos eWP48 numbers I was hoping to get?

    Paul Pierce Orlando series WP48 and Lakers series WP48
    Ron Artest Boston series WP48
    Matt Barnes Boston series WP48

    I’d love to see em!

    • 10/1/2010
      Reply

      sd,
      It have to manually crunch those and I’m actually working. I could run it by Andres and see if he could run a script.

  13. Evanz
    10/1/2010
    Reply

    Arturo, you’re going to have to bump up the Warriors again. I just realized you didn’t include Dorell Wright in your star list.

    Dorell had a 0.198 WP48 last season (rounded up to 0.200, right?) in 1496 minutes. That gives the Warriors 3 star points!

    • 10/1/2010
      Reply

      Prof. Berri has him at .179 WP48. Sorry :-). Still a contender though (And the best bet on the board)

      • Evanz
        10/1/2010
        Reply

        Oh, I was getting the stats from Andres’ website. I guess he has it wrong.

        • 10/1/2010
          Reply

          Hey now Evan, me wrong? :)
          I use an algorithm that works as follows. Ask Yahoo what the players are listed as, if the positions aren’t divied up correctly then readjust based on height and bmi (for the record the height + bmi stuff is all Dr. Berri’s idea and numbers), Dr. Berri adjusts the positions by hand. As such there can be some discrepancies for players on teams with lots of tweeners. This can be true of any team by the way. If you have two Forwards , you can change the evaluation of them by listing switching which you list at PF or SF.

          • Evanz
            10/1/2010
            Reply

            You have Wright at 40:60 SG:SF. If Berri has him at 100% SF, that would make his WP48 come out lower than your calculation, I suppose.

  14. some dude
    10/1/2010
    Reply

    doesn’t 82games already categorize them by the correct position?

    Arturo – no worries. I don’t have a way of doing it, so I was just hoping we could get those numbers. I think they give one of us a case. :D

    • 10/1/2010
      Reply

      Some dude,
      Yahoo works pretty well. 82 games doesn’t have them listed exactly (so total numbers add up 100%). In the future game by game data (which you also want above I see :) ) is the ideal way to go.

  15. some dude
    10/1/2010
    Reply

    but doesn’t yahoo mislabel them? For instance, don’t they have Bargnani as a Center where 82games has Bosh as the center and Bargnani as the PF?

  16. Neal Frazier
    10/2/2010
    Reply

    I am wondering if there is not some floor of combined WP48 for the top 6 guys that may matter more than the number of star points – this may explain the houston phenomenon – so I am wondering what the lowest sum of WP48 for the top 6 guys from the champs from the last 30 odd years is and if that metric does a better job of narrowing the field of contenders than the star points system? Ultimately, are there fewer teams with the necessary sumWP48 than there are teams with 2+ star points and 52+ wins produced?

  17. A simple contender criteria
    10/7/2010
    Reply

    Every title winner in the last 12 years has been top 5 on a simple stat, eFG% differential in the regular season of their win. The last Bulls title winner clearly wasn’t and one season the title wining Rockets were 6th; but, other than that, this criteria holds up for the last 20 years at least.

    Last season the Lakers were only 11th. The season before they were 5th best. I think the league GMs are probably over-rating this season’s Lakers chances. 2/3rds picked them to win it all. Vegas on the other hand is picking the Heat. Don’t know how important a weight they give to eFG% differential directly or implicitly but they the title winner almost always comes from their top 4 withthe 03-04 Pistons being the only exception in the last 7 seasons and they were tied for 6th pick.

    You could say this is in line with the Wins Produced model’s emphasis on shooting efficiency and making the team defensive adjustment. But you could also say simply using eFG% differential on its own is pretty good.

    • 10/7/2010
      Reply

      Good point. The trick here’s that we’re trying to proactively look ahead and predict the criteria for winning given the offseason changes. efg% and points and efficiency differential are somewhat more volatile to predict. This is meant as an ongoing exercise/discussion going forward.

  18. A simple contender criteria
    10/7/2010
    Reply

    “If” predicting own team efg% from player past stats is volatile, “then” there would seem to need to be significant usage, context, chemistry and coaching issues in play, right? That would go against the grain of a view or assumption that some models are based on. I imagine it could be even more difficult to predict defensive eFG% as context, chemistry and coaching issues impact those results too. But I haven’t looked at this data enough yet to go further.

    I wonder how good estimated net eFG% based on last season (or recent history for a longer period) and adjusted for off-season changes would identify the top 5 contenders. It won’t be as good as when you know that regular season’s results but it might still be pretty good. Net eFG% is not the whole source of greatness but it probably / usually the biggest part.

  19. A simple contender criteria
    10/7/2010
    Reply

    3 of the 4 conference finalists were top 5 last regular season on net eFG%. Boston and Orlando haven’t changed much and are likely to stay in that group. The Cavs and the Suns did change out key stars but maybe the Suns and their system and coaching can substitute and maintain. Might be worth a more detailed look after the off-season changes especially for them. And the Heat. Heat were actually 7th on net eFG% last season. I imagine they expect to do far better on offense but will the defense slip and by how much?

    Which approaches would do the best to try to project defense or specifically shot defense? None are likely to be reliably very precise. If shot defense is rolled up within a totally team based defensive adjustment it may not provide that much help in projecting individual shot defense impact in a new context. But perhaps using individual player shot defense data & impact analysis and giving some substantial weight to past coaching success on defense in some manner could get you reasonably close.

    • 10/7/2010
      Reply

      The defense adjustment by individual rather than team is one of the things we’re looking at as we go on. Coaches in general are not a significant impact on player performance (with some notable exceptions) but they can kill you on minute allocation.

      I’d be interested to see if you had a model in mind.

  20. A simple contender criteria
    10/7/2010
    Reply

    Ty’s use of counterpart match-up defensive data is one approach but I probably wouldn’t use it entirely. Half counterpart / half team level defensive data would be one compromise, better I think than all one or the other. In the past I’ve used the hoopnumbers Defensive Adjusted +/- and the factor split of that. No longer publicly available but it can be replicated by some. The shot defense impact was isolated by itself and the standard errors were pretty low. But since their is still a modest amount of uncertainty I’ve thought about just translating those ratings into a 5 piece range rather than using and claiming as exact. Even if you are off by one on the rating you are still pretty close and probably better than treating all teammates exactly the same or mostly the same in other models or pieces of models. Few big minute players would be off by more than one ratings point.

  21. A simple contender criteria
    10/7/2010
    Reply

    Players who change teams could be estimate by using a half weight on the individual defensive data, one quarter on the old team defensive data, and one quarter on the new team data.

    If the rest of the model covers everything but shot defense already, the team defensive adjustment could / in my mind should use just the shot defense factor and not all of defense for the adjustment as that creates some repetition.

  22. A simple contender criteria
    10/7/2010
    Reply

    You could translate any rating into a 5, 7 or 10 piece range to recognized that you only have the power to roughly estimate this part of the game.

    • 10/7/2010
      Reply

      I like your ideas but I would argue for a simpler model to start into which you could build and test vs historical data and then add additional complexity and test for increased predictive value and then iterate over time. If you make it too complex in one go it becomes impossible to separate the chaff from the value.

      My take on models is start simple and add complexity only if it can be proven to add value for the effort.

  23. A simple contender criteria
    10/8/2010
    Reply

    Start simple and build more complexity into it gradually is a good general approach but I think the simple ingredients of a model based on the boxscore in various flavors have already largely done step A of that plan and it is now time for step B- more complexity to get more accurate individual value assessment with regards to shot defense which is probably at least 20% of all value.

    • 10/8/2010
      Reply

      I did some early post on a defense inclusive model (as I agree that shot defense is the principal source of error). The math isn’t overly complicated, it’s the data gathering. I started the build with 82games opponent data but I didn’t quite trust it. We are currently working on getting that data set together though (I’m sure teams have this but they’re not sharing).

Leave a Reply

Your email address will not be published. Required fields are marked *