Loading ...
Sorry, an error occurred while loading the content.

Some conceptual questions

Expand Messages
  • Dean Oliver
    1. Does it reflect better on a player if, when he plays best, his team loses or if, when he plays better, his team wins? 2. What is better for a team -- a
    Message 1 of 10 , Nov 2, 2001
    • 0 Attachment
      1. Does it reflect better on a player if, when he plays best, his
      team loses or if, when he plays better, his team wins?

      2. What is better for a team -- a player who scores a lot of points
      while shooting poorly (or turning the ball over a lot) or a player
      who scores few points while shooting very well?

      3. Does offense or defense "win championships"?

      4. What is the ideal defender?

      5. Do you want quickness or size?

      6. Does blowing out an opponent really reflect on a team's ability?

      7. What is the value of having one player who can draw double teams
      on improving the efficiency of teammates?

      These are some of the fundamental ones that I have grappled with
      historically. I'd like to hear some opinions.

      Dean Oliver
      Journal of Basketball Studies
    • Michael K. Tamada
      ... I think the latter, but one has to be very careful when trying to measure this, for two reasons: 1. Some players will get more playing time when the game
      Message 2 of 10 , Nov 2, 2001
      • 0 Attachment
        On Sat, 3 Nov 2001, Dean Oliver wrote:

        >
        > 1. Does it reflect better on a player if, when he plays best, his
        > team loses or if, when he plays better, his team wins?

        I think the latter, but one has to be very careful when trying to measure
        this, for two reasons:

        1. Some players will get more playing time when the game is a blow-out,
        which will skew their statistics. I remember noticing that when Wally
        Walker played for the Sonics, his best games (not just in terms of total
        minutes and points, but I believe shooting percentage and
        points-per-minute too) came when the Sonics blew out their opponent.
        Walker was a lousy NBA player, but he was fairly swift and could move
        without the ball, so he'd get plenty of points on fast break lay-ins
        against opponents who had lost their heart.

        So if you looked at Walkers' performances, and Sonic victories or
        point-differential, you'd see a strong correlation. But it wasn't as if
        Walker's performances were leading to Sonic victories. The causality went
        the OTHER way.


        2. A similar causality problem occurs even when we're not talking about
        obvious benchwarmers like Walker. We might try to identify key starters
        who are the "leverage players": the ones who, when they play well, the
        team wins, and when they play poorly, the team loses. If we found such
        players, their teams would probably want to do things to help them play
        well, and the opponents would try to design defenses (or offenses too)
        that stopped the leverage player.

        But all of that could easily be empty, useless strategizing. Unless we
        can somehow prove that the causality is: leverage player plays well -->
        team wins, then we may be strategizing about the symptoms, not the
        disease. If, say, Reggie Miller turns out to be a leverage player for the
        Pacers, does that mean that an opponent should double-team him and
        cause him to have a bad game? That may be a strategy as useless as curing
        a patient's fever by dunking the thermometer in ice water. The seeming
        leverage might be a statistical artifact, with the causality running the
        other way.


        So, on the whole I don't think think the stats can be used to show
        anything. Personal observation is probably better, because it allows us
        to decide who's really driving the action.

        An example, again from the Sonics of the late 1970s: Paul Silas was a
        player who usually seemed to play better when the Sonics were losing.
        Younger players might lose their heads or give up but the rugged veteran
        would keep chugging away.

        But such players wouldn't have value if their good play still resulted in
        losses. If Silas was indeed a player who played better when the Sonics
        were losing, he was valuable because his presence on the court meant the
        Sonics would have a better chance of making a comeback, instead of falling
        further behind and turning the game into a blowout.

        Often of course, Silas' heroics would go for naught and the game would
        still end up a Sonic loss. But the value was still there.


        All of which is I guess a long-winded way of agreeing with Stuart
        McKibbin's point: we want players who play well. Some players might play
        better in certain situations: able to lead a team to victory (Gus
        Williams comes to mind), or to rally the troops against an impending loss
        (Silas). But I don't know if such player can be identified statistically.


        > 2. What is better for a team -- a player who scores a lot of points
        > while shooting poorly (or turning the ball over a lot) or a player
        > who scores few points while shooting very well?

        Stuart McKibbin nails this answer too: it depends on the context. An
        offense of nothing but Bo Outlaws or the later Artis Gilmore would not be
        a good offensive team, despite these guys' 60% FG percentages. But if
        you've already got a Pete Maravich or a World B Free (or an Alan Iverson
        or a Larry Hughes) on your team, you don't want to be giving major minutes
        to another player of that type. But such a player can be of value if the
        team lacks offense: Philadelphia would be in bad shape without Iverson,
        and Free generated a lot of offense for the Clippers, Warriors, etc.

        If one already has a good team, then the answer I think is more likely to
        be the high percentage low scorer. Because presumably the team already
        has a good offense and enough scoring. So you don't want to add another
        gunner in there. (But the gunners who are already on the team may be
        providing valuable offense. I'm thinking of Vinnie Johnson and the late
        1980s Pistons here: those Pistons didn't need to add a World B Free to
        their roster. But that's not to say that VJ was expendable; he provided
        them with the point-scoring that they badly needed. But with him on the
        roster, the players they would want to add were not gunners but rather
        people who would make other contributions, e.g. James Edwards.)

        If you've got a bad team, then the answer is more likely to be that you
        want the big scorer. Because you probably need the points. Tony Campbell
        with the early Timberwolves, or of course Free with the Clips and
        Warriors. Those teams would've been even more wretched without their big
        gunner.

        > 3. Does offense or defense "win championships"?

        I'd like to explore this in a way similar to the way that some baseball
        obsververs have looked at pitchers vs. batters. Who has more influence on
        the outcome of the pitcher vs batter matchup? I think they make a pretty
        good observation when they note that batters have a huge range of
        productivity, from say Barry Bonds to Derek Bell. But pitchers range of
        what they give up is not nearly so large. Even when you compare say Pedro
        Martinez to say Bobby Jones: Jones might give up three times as many runs
        as Pedro but Bonds will produce a lot more than three times as many runs
        as D Bell.

        So even though some outcomes are clearly under the control of the pitcher
        (whether to issue an intentional walk for example) on the whole, it's the
        quality of the batter that matters more than the quality of the pitcher.

        Or so goes the argument. I'm not sure I believe it completely but it
        seems reasonable. (An alternative explanation is that batters simply have
        a larger degree of talent differential than pitchers do. However, this
        seems implausible. Although hitters at different positions do show
        large differences (catchers vs first basemen), even at any one position
        we observe large differences in hitting ability (Rey Ordonez vs A-Rod).)


        Interestingly, when we apply this to the 2000-01 regular season statistics
        for FG% and opponents' FG%, their standard deviations are the SAME: 1.5%
        (mean of 44.3%).

        However teams' offensive rebounding percentages show a larger standard
        deviation than their defensive rebounding percentages (or, equivalently,
        the standard deviation of their opponents' offensive rebounding
        percentages): 2.3% for off rebds, 1.4% for def rebds. (Average of 28.1%
        offensive rebounds, 71.7% defensive -- obviously a bit of rounding error
        in there.)

        Hardly conclusive, but it's a start. There's also of course the
        possibility that while offense and defense (or pitching and hitting) may
        be equally important during the regular season, playoff teams are
        different, and the winner may be more likely to be determined by the
        defense or the pitching.

        Also there's the possibility that offense and defense might have different
        standard deviations when we look at the SAME team across the season (as
        opposed to looking at DIFFERENT teams in the same season).

        > 4. What is the ideal defender?

        Very tough question to answer, in addition to the players that Stuart
        McKibbin mentioned, Rodman was a notable one-on-one defender. Payton too.
        Russell and Wilt of course, for team defense even more than one-on-one
        defense. Too broad a category I think to come up with a simple answer.

        > 5. Do you want quickness or size?

        Quickness in a PG, size in a C. Other than that, depends on context
        again. Some players have strengths and weaknesses wherein they benefit
        more from quickness, others from size. Players like Iverson and Brent
        Barry will probably become a lot less effective when they lose a step.
        Charles Oakley would be a lot less effective if he were 2 inches shorter
        and/or 30 pounds lighter.

        > 6. Does blowing out an opponent really reflect on a team's ability?

        A little, but I think the evidence here is pretty clear (or it was when
        someone else did some statistics several years ago): team's records in
        close games are of little importance. The winner of those is determined
        largely by luck (good teams will tend to win more than their share of
        course, but not always). Blowing out an opponent is a little better
        indicator. But the most important one: good teams do not get blown out.

        So blowing out an opponent is not a bad indicator (it's definitely better
        than looking at a teams W-L record in close games, overtime games, or the
        like). But the best indicator is not one of blowing out opponents, but
        rather NOT getting blown out YOURSELF. Early in the season this can be an
        even better indicator than a team's won-loss record: if a team is 5-2,
        but the 2 losses were by 20 points, look out. (Unless of course there are
        extenuating circumstances such as injuries.)

        A similar but not identical observation was made on a Sonics email list a
        couple of years ago. The Sonics early in the season had a won-loss record
        which was better than what was suggested by their "points scored" vs
        "points allowed" numbers. So either they were winning more than their
        fair share of close games (a lucky streak unlikely to be kept up
        throughout the season), or their games included an inordinate number
        of ones in which they got blown out (or, possibly, failed to include the
        normal number of games in which they blew out opponents), causing their
        overall PTs:Opp PTs ratio to drop.

        Either way, this suggested that the Sonics would not maintain their
        won-loss rate during the rest of the season, and sure enough, they did
        not.


        > 7. What is the value of having one player who can draw double teams
        > on improving the efficiency of teammates?

        I'd think pretty high, IF that player has good passing ability. (This may
        all change under the new defensive rules.) Jordan could draw double teams
        and in doing so make his teammates better (exhibit A: John Paxson and
        Steve Kerr). Elvin Hayes and Spencer Haywood could draw double teams, but
        they were black holes into which basketballs disappeared. Good players,
        legitimate all-pros in terms of individual talent, but they didn't make
        their teammates better to the extent that a Jordan, Kareem, etc. did.


        --MKT
      • Dean Oliver
        Time for me to get back to these... ... No easy answer to this one. What I have seen is that the best teams records do very strongly reflect their best
        Message 3 of 10 , Nov 18, 2001
        • 0 Attachment
          Time for me to get back to these...

          --- In APBR_analysis@y..., "Dean Oliver" <deano@t...> wrote:
          >
          > 1. Does it reflect better on a player if, when he plays best, his
          > team loses or if, when he plays better, his team wins?
          >

          No easy answer to this one. What I have seen is that the best teams'
          records do very strongly reflect their best player's percentage of
          good games. Specifically, I calculate the "winning %" of individuals
          as the % of time their offensive rating is higher than their
          defensive rating. And quite consistently, a team's record reflects
          the record of the player who uses the most possessions. It varies a
          little, but it seems to work.

          On instinct, I say it's better if a player plays best when his teams
          wins.

          > 2. What is better for a team -- a player who scores a lot of
          points
          > while shooting poorly (or turning the ball over a lot) or a player
          > who scores few points while shooting very well?
          >

          As others have mentioned, context may be important here, which is an
          important point. The value of players really can't be measured with
          one number. Rodman was highly valuable to the teams that had good
          scorers. He was an extreme liability on teams that didn't have good
          scorers even if he was putting up the same rebounding numbers.

          > 3. Does offense or defense "win championships"?
          >

          I did a study on whether it was more common for good offensive teams
          to win the title or good defensive teams. Generally, the good
          offensive teams did. The article was rather sarcastically looking at
          a method I have that adjusts team offensive and defensive ratings
          based on how much they "relax" -- something you can approximately
          measure. Well, if you say that teams only relax on defense, not on
          offense, it says that defense wins championships a little more
          frequently. The result was half-sarcastic because of teams like last
          year's Lakers. The Lakers really turned up their defense in the
          playoffs, something that started happening in the early '80's, I
          beliieve. Offense carries them, but to lock up the title, the Lakers
          really needed their defense to kick in, as it did. That may be what
          the phrase meant.

          > 4. What is the ideal defender?
          >

          I've gotten to the point where I don't think this has an answer
          either. Big teams seem to be better defensive teams in general, but
          that may be because it's the easiest physical thing to quantify.
          Quickness and good hands were the hallmark of the Bulls' defensive
          units. Good hands are extremely under-rated in basketball.

          > 5. Do you want quickness or size?
          >

          John Wooden always said quickness. I always felt that the Magic-led
          Lakers and the Bird-led Celtics were good examples of how size wins
          because they weren't that quick. The Lakers were fast, but not
          quick. The Bulls, however, show how quickness wins. However, they
          were also quite big. They weren't big in the middle, but they were
          big at the guard position.

          > 6. Does blowing out an opponent really reflect on a team's ability?
          >

          The ease with which you win does matter. But after a point, it
          really doesn't matter. I know the BCS computers account for this
          now. What I interestingly found once was that there might even be
          such a thing as an "ideal score". Winning by 30 may reflect worse on
          you than winning by 25. Weird result based upon a lot of theory.

          > 7. What is the value of having one player who can draw double
          teams
          > on improving the efficiency of teammates?

          Hmm, I think that's what Allen Iverson's absence may help us show.
          And he isn't the greatest passer. This is one fundamental thing that
          I cannot account for right now with my work. I do try, saying that
          assists are more valuable when they are to players who shoot better
          (in part, presumably, because their defender had to guard the guy who
          had the assist). But it's really hard to do. (JohnM -- this is not
          in the formulas you use because it adds a lot of complexity.)

          Dean Oliver
          Journal of Basketball Studies
        • Mike Goodman
          Good thing, some of us are not shy about responding to our own posts! ... a ... That surely holds true whether looking at a good team or a bad team. In
          Message 4 of 10 , Nov 19, 2001
          • 0 Attachment
            Good thing, some of us are not shy about responding to our own posts!

            >--- In APBR_analysis@y..., "Dean Oliver" <deano@r...> wrote:
            > Time for me to get back to these...
            >
            > --- In APBR_analysis@y..., "Dean Oliver" <deano@t...> wrote:
            > >
            > > 1. Does it reflect better on a player if, when he plays best,
            >is
            > > team loses or if, when he plays better, his team wins?
            >
            > No easy answer to this one. What I have seen is that the best
            >teams'
            > records do very strongly reflect their best player's percentage of
            > good games. Specifically, I calculate the "winning %" of
            >individuals
            > as the % of time their offensive rating is higher than their
            > defensive rating. And quite consistently, a team's record reflects
            > the record of the player who uses the most possessions. It varies
            a
            > little, but it seems to work.
            >
            > On instinct, I say it's better if a player plays best when his
            >teams
            > wins.

            That surely holds true whether looking at a good team or a bad team.
            In general.
            What about Derrick Coleman with Charlotte? There were some
            astonishing negative correlations between his production, and his
            team's success.
            I think he was their most productive player in '99-00 (per-minute).

            > > 2. What is better for a team -- a player who scores a lot of
            > points
            > > while shooting poorly (or turning the ball over a lot) or a
            player
            > > who scores few points while shooting very well?
            > >
            >
            > As others have mentioned, context may be important here, which is
            an
            > important point. The value of players really can't be measured
            with
            > one number. Rodman was highly valuable to the teams that had good
            > scorers. He was an extreme liability on teams that didn't have
            good
            > scorers even if he was putting up the same rebounding numbers.
            >

            Not only Iverson, but Dominique Wilkins missed more shots than he
            made, yet was quite valuable in doing so. It seemed to me, a good
            many of his misses were (like Ivy's) penetration moves (usually over
            rather than under), and lots of offensive rebounds ensued (Nique got
            a lot of his own misses).
            When was Rodman a liability, and on what teams?
            I thought he became a liability (by degrees) when his scoring
            evaporated, and then when his defense slipped. And of course the
            distractions necessary for his "genius" lifestyle.
            Aging (and attitude) eventually made Rodman a net negative.

            > > 3. Does offense or defense "win championships"?
            > >
            >
            > I did a study on whether it was more common for good offensive
            teams
            > to win the title or good defensive teams. Generally, the good
            > offensive teams did. The article was rather sarcastically looking
            at
            > a method I have that adjusts team offensive and defensive ratings
            > based on how much they "relax" -- something you can approximately
            > measure. Well, if you say that teams only relax on defense, not on
            > offense, it says that defense wins championships a little more
            > frequently. The result was half-sarcastic because of teams like
            last
            > year's Lakers. The Lakers really turned up their defense in the
            > playoffs, something that started happening in the early '80's, I
            > beliieve. Offense carries them, but to lock up the title, the
            Lakers
            > really needed their defense to kick in, as it did. That may be
            what
            > the phrase meant.
            >

            (I was informed, by an eyewitness, that the Lakers would be sweeping
            thru last year's playoffs, because of their "swagger". I was
            skeptical.)
            Championship runs require 'some' defense, but historically what ruins
            most playoff runs is lack of shooting.
            Shooting percents and assist rates show the most dramatic overall
            dropoff in postseason play, across all players' data. Partly this is
            just increased competition, but fatigue and pressure take their toll.
            Hitting the pressure shot, or even having the willingness to take the
            shot, is still what defines a player, in crunch time.

            > > 4. What is the ideal defender?
            > >
            >
            > I've gotten to the point where I don't think this has an answer
            > either. Big teams seem to be better defensive teams in general,
            but
            > that may be because it's the easiest physical thing to quantify.
            > Quickness and good hands were the hallmark of the Bulls' defensive
            > units. Good hands are extremely under-rated in basketball.
            >

            Quickness vs. size is fun to watch. Bulls were tough even with 5 mid-
            sized models on the floor. Opponents with a Muresan or a McIlvaine
            just couldn't use them at all.
            Good hands are IT. A player of any size who shows up with bad hands
            is pretty useless. A few players have good hands almost every day.
            I would rather come without my shot than without good hands.

            > > 5. Do you want quickness or size?
            > >
            >
            > John Wooden always said quickness. I always felt that the Magic-
            led
            > Lakers and the Bird-led Celtics were good examples of how size wins
            > because they weren't that quick. The Lakers were fast, but not
            > quick. The Bulls, however, show how quickness wins. However, they
            > were also quite big. They weren't big in the middle, but they were
            > big at the guard position.
            >

            Actually, the Bulls always had big bodies at their disposal -- just
            not skilled, big bodies.
            Sometimes quickness is not in the feet, or even in the hands, but in
            anticipation. Larry Bird was certainly not quick physically, but you
            would never know that, when he was already making the move before you
            could see the situation developing.

            > > 6. Does blowing out an opponent really reflect on a team's
            ability?
            > >
            > The ease with which you win does matter. But after a point, it
            > really doesn't matter. I know the BCS computers account for this
            > now. What I interestingly found once was that there might even be
            > such a thing as an "ideal score". Winning by 30 may reflect worse
            on
            > you than winning by 25. Weird result based upon a lot of theory.
            >

            This always depends on coaching. If the coach isn't blowout-minded,
            it might reflect depth. Or it might say more about your opponent.

            > > 7. What is the value of having one player who can draw double
            > teams
            > > on improving the efficiency of teammates?
            >
            > Hmm, I think that's what Allen Iverson's absence may help us show.
            > And he isn't the greatest passer. This is one fundamental thing
            that
            > I cannot account for right now with my work. I do try, saying that
            > assists are more valuable when they are to players who shoot better
            > (in part, presumably, because their defender had to guard the guy
            who
            > had the assist). But it's really hard to do. (JohnM -- this is
            not
            > in the formulas you use because it adds a lot of complexity.)
            >
            > Dean Oliver
            > Journal of Basketball Studies

            I can agree that some assists are more 'impressive' than others, and
            these usually do lead to a sure basket; but since the assist
            doesn't "count" (literally) unless the shot is made, you get the
            situation where, say, Kobe hits Shaq for the alleyoop dunk. Good
            pass, good result. In another interval, Shaq may hit 3 outside
            shooters with perfectly decent passes, and only 1 basket may result.
            So why look for a way to degrade Shaq's 3 nice passes, when he is
            already only getting credit for 1 measly assist?
            Oftentimes, a low-pct pass leads to a high-pct shot. And an easy
            pass leads to a lower-pct shot. Passes inside come with more
            turnovers, in other words.
            So, rather than over-analyze the compounding percentages, just
            add 'em up.
            Points is points!

            Mike Goodman
            cluttered desk
          • Dean Oliver
            ... show. ... that ... better ... and ... Here is the reason: Teams will cover that Shaq alley-oop more tightly than the pass to the jump shooter. It s a
            Message 5 of 10 , Nov 19, 2001
            • 0 Attachment
              > > > 7. What is the value of having one player who can draw double
              > > teams
              > > > on improving the efficiency of teammates?
              > >
              > > Hmm, I think that's what Allen Iverson's absence may help us
              show.
              > > And he isn't the greatest passer. This is one fundamental thing
              > that
              > > I cannot account for right now with my work. I do try, saying
              that
              > > assists are more valuable when they are to players who shoot
              better
              > > (in part, presumably, because their defender had to guard the guy
              > who
              > > had the assist). But it's really hard to do. (JohnM -- this is
              > not
              > > in the formulas you use because it adds a lot of complexity.)
              > >
              > > Dean Oliver
              > > Journal of Basketball Studies
              >
              > I can agree that some assists are more 'impressive' than others,
              and
              > these usually do lead to a sure basket; but since the assist
              > doesn't "count" (literally) unless the shot is made, you get the
              > situation where, say, Kobe hits Shaq for the alleyoop dunk. Good
              > pass, good result. In another interval, Shaq may hit 3 outside
              > shooters with perfectly decent passes, and only 1 basket may result.
              > So why look for a way to degrade Shaq's 3 nice passes, when he is
              > already only getting credit for 1 measly assist?

              Here is the reason: Teams will cover that Shaq alley-oop more
              tightly than the pass to the jump shooter. It's a heck of a lot
              easier to get a pass to an open jump shooter than to Shaq under the
              basket. The guy who gets it to Shaq there deserves more credit than
              getting it to a random jump shooter. My system is set up so that if
              Shaq keeps passing it out to jump shooters who keep missing, it says
              he is stupid in doing so because he could make his own shot at a
              higher rate.

              Here is the difference I calculate. On the alley-oop to a, say, 100%
              shooter, I give 50% credit to the shooter and to the assistant. On
              the random pass to the open jump shooter, who, in your example, makes
              1 of 3 shots, I give 16% to the assistant, 83% to the shooter. Most
              shooting is around 50%, so the assistant gets about 25% and the
              shooter gets about 75%. Basically, if I see that a point guard goes
              to a team and his teammates start shooting worse, the system accounts
              for that, weakening the value of his assists.

              No matter the percentage, this does avoid the double-counting that
              has been raised as a problem with tendex.

              > Oftentimes, a low-pct pass leads to a high-pct shot. And an easy
              > pass leads to a lower-pct shot. Passes inside come with more
              > turnovers, in other words.
              > So, rather than over-analyze the compounding percentages, just
              > add 'em up.
              > Points is points!

              The thing I wanted to recognize in accounting for assists was that
              some players really should pass more, especially those that shoot
              poorly. Those that shoot poorly will be covered for their shot
              less. Hence, defenders will be covering the outlets more, making the
              assists harder (Jason Kidd's low shooting percentage means he should
              pass more, so this refinement helps him, for example). It is a
              refinement that makes only small differences at the NBA level (almost
              never more than 5% in offensive ratings). It makes a bigger
              difference at lower levels of hoops where possessions are not roughly
              equal to a point and, more importantly, where the variations in
              talent on a team are much greater.
            • Mike Goodman
              I am becoming increasingly convinced that good statistics improves the game, while bad statistics can harm the way people play. ... Dean, I hope you aren t
              Message 6 of 10 , Nov 20, 2001
              • 0 Attachment
                I am becoming increasingly convinced that good statistics improves
                the game, while bad statistics can harm the way people play.

                --- In APBR_analysis@y..., "Dean Oliver" <deano@r...> wrote:
                >......It's a heck of a lot
                > easier to get a pass to an open jump shooter than to Shaq under the
                > basket. The guy who gets it to Shaq there deserves more credit
                >than
                > getting it to a random jump shooter.

                Dean, I hope you aren't advocating 'difficulty points'.
                In the 1980s, the Celtics did not outjump or outrun anyone. They won
                by making good, easy passes that led to good, easy shots. That, to
                me, is good basketball.
                Good, because it can be replicated by good coaching, unselfish
                attitudes, and slow/white guys.

                > My system is set up so that
                >if
                > Shaq keeps passing it out to jump shooters who keep missing, it
                >says
                > he is stupid in doing so because he could make his own shot at a
                > higher rate.

                It is not now, nor has it ever been, dumb to pass out of the double
                team. I believe this strongly enough, that I am tempted to say it is
                worth some double-counting in the assisted-basket area.
                Put into practice, I claim it is more valuable in crunch time, to
                have a bunch of good passers on the floor, than to have a go-to guy,
                with all the eggs in one basket.

                > Here is the difference I calculate. On the alley-oop to a, say,
                >100%
                > shooter, I give 50% credit to the shooter and to the assistant. On
                > the random pass to the open jump shooter, who, in your example,
                >makes
                > 1 of 3 shots, I give 16% to the assistant, 83% to the shooter.
                >Most
                > shooting is around 50%, so the assistant gets about 25% and the
                > shooter gets about 75%. Basically, if I see that a point guard
                >goes
                > to a team and his teammates start shooting worse, the system
                >accounts
                > for that, weakening the value of his assists.
                >
                > No matter the percentage, this does avoid the double-counting that
                > has been raised as a problem with tendex.
                >

                This graded-awarding idea makes good sense intuitively, but how can
                it be applied? How can one judge the shooting pct for an individual
                shot? In retrospect, it's either 100% or 0%.
                If you watch the game, you can count 'coulda-been' assists, 'blown'
                assists, or whatever. But isn't there a need for a consistent way to
                apply this, across past seasons, etc?

                > The thing I wanted to recognize in accounting for assists was that
                > some players really should pass more, especially those that shoot
                > poorly. Those that shoot poorly will be covered for their shot
                > less. Hence, defenders will be covering the outlets more, making
                the
                > assists harder (Jason Kidd's low shooting percentage means he
                should
                > pass more, so this refinement helps him, for example).

                Some low shooting pcts are the result of unwillingness to pass, i.e.,
                you can send 2-3 defenders after some guys, and they still will shoot.
                Others, like Kidd, are dared to shoot; he just needs to hit the
                shot. Likely, Kidd can't pass much more than he does.

                It is a
                > refinement that makes only small differences at the NBA level
                (almost
                > never more than 5% in offensive ratings). It makes a bigger
                > difference at lower levels of hoops where possessions are not
                roughly
                > equal to a point and, more importantly, where the variations in
                > talent on a team are much greater.

                Perhaps we can factor out assisted-points as follows: suppose the
                Utah Jazz offense consists of 50% assisted baskets, 30% unassisted
                baskets, and 20% free throws (unassisted, officially). Suppose
                further that John Stockton has 50% of the assists and 10% of the
                baskets; and Donyell Marshall has 20% of the hoops and 5% of the
                assists.
                Without doing any math, we can assume that the great majority of
                Stockton's hoops are unassisted, and the majority of Marshall's are
                assisted.
                By factoring out each player's assists, from the team assists on his
                own hoops (no one assists himself), we could achieve a fair reckoning
                of each player's assisted/unassisted hoop ratio.
                It may not be too effective for bench players, unless we figure an
                assist-per-minute thing.

                I admit to a strong prejudice for passing teams (as opposed to 1-on-1
                teams). Double-counting of assists has always haunted me, and I
                always shrugged it off by believing passing gets you thru more tough
                spots than the ability to go 1-on-1; so I have rewarded that
                statistically.
              • Dean Oliver
                ... won ... Not sure what difficulty points are, but I don t think so. If anything, I m advocating simplicity points . Rewarding the players who make the
                Message 7 of 10 , Nov 20, 2001
                • 0 Attachment
                  --- In APBR_analysis@y..., "Mike Goodman" <msg_53@h...> wrote:
                  > Dean, I hope you aren't advocating 'difficulty points'.
                  > In the 1980s, the Celtics did not outjump or outrun anyone. They
                  won
                  > by making good, easy passes that led to good, easy shots. That, to
                  > me, is good basketball.

                  Not sure what difficulty points are, but I don't think so. If
                  anything, I'm advocating "simplicity points". Rewarding the players
                  who make the good passes that lead to easy shots.

                  >
                  > It is not now, nor has it ever been, dumb to pass out of the double
                  > team. I believe this strongly enough, that I am tempted to say it
                  is
                  > worth some double-counting in the assisted-basket area.

                  That's where you are occasionally wrong. You're right in the NBA,
                  probably, but the Bulls are in the NBA, so I may have to think
                  twice. In HS ball, some players are good enough that they are more
                  likely to score out of a double-team than their teammates are wide-
                  open at the perimeter. It is an extreme case and, if HS is for
                  learning (not winning), then passing out of a double-team is still a
                  good thing.

                  >
                  > This graded-awarding idea makes good sense intuitively, but how can
                  > it be applied?

                  You do a pretty good job guessing at it below.

                  > Perhaps we can factor out assisted-points as follows: suppose the
                  > Utah Jazz offense consists of 50% assisted baskets, 30% unassisted
                  > baskets, and 20% free throws (unassisted, officially). Suppose
                  > further that John Stockton has 50% of the assists and 10% of the
                  > baskets; and Donyell Marshall has 20% of the hoops and 5% of the
                  > assists.
                  > Without doing any math, we can assume that the great majority of
                  > Stockton's hoops are unassisted, and the majority of Marshall's are
                  > assisted.
                  > By factoring out each player's assists, from the team assists on
                  his
                  > own hoops (no one assists himself), we could achieve a fair
                  reckoning
                  > of each player's assisted/unassisted hoop ratio.
                  > It may not be too effective for bench players, unless we figure an
                  > assist-per-minute thing.

                  This is remarkably close to what I do, even the adjustment you have
                  to do for bench players. You can estimate pretty well what
                  percentage of a player's shots are assisted. It's a very complicated
                  formula at its most refined. Stockton does have fewer of his made
                  shots assisted on. Overall, the Jazz had 71% of their fgm's assisted
                  on. My estimates of the percentages of each player's fgm's that were
                  assisted on:

                  Plyr %asstd asstd fgm
                  Malone, Karl 0.690 462
                  Marshall, Dony 0.793 339
                  Russell, Bryon 0.750 231
                  Stockton, John 0.479 157
                  Starks, John 0.735 201
                  Manning, Danny 0.811 200
                  Vaughn, Jacque 0.591 100
                  Polynice, Olden 0.816 168
                  Ostertag, Greg 0.796 111
                  Benoit, David 0.820 58
                  Lewis, Quincy 0.786 39
                  Stevenson, Desh 0.740 23
                  Crotty, John 0.616 14
                  Padgett, Scott 0.815 15

                  The sum of assisted fgm is 2118, 7 higher than the 2111 assists the
                  Jazz had. So it's a little bit off, but not so much that I worry
                  about it.

                  Dean Oliver
                  Journal of Basketball Studies
                • aussievamp2
                  ... hands ... day. ... I agree. I have a sort of theory that assist to turnover ratio defines talent to some degree for want of a better word, and maybe
                  Message 8 of 10 , Jul 4 10:57 PM
                  • 0 Attachment
                    --- In APBR_analysis@y..., "Mike Goodman" <msg_53@h...> wrote:
                    > Good hands are IT. A player of any size who shows up with bad
                    hands
                    > is pretty useless. A few players have good hands almost every
                    day.
                    > I would rather come without my shot than without good hands.
                    >


                    I agree. I have a sort of theory that assist to turnover ratio
                    defines 'talent' to some degree for want of a better word, and maybe
                    this is it (cf difference between NBA, college, NBL, WNBA, etc. etc.)

                    Of course, I could just be whacked out, any comments?
                  • HoopStudies
                    ... maybe ... etc.) ... Assist-to-turnover ratio reflects style of play, position, coaching, strategy -- lots more than just pure talent. Isolating talent is
                    Message 9 of 10 , Jul 8 4:54 PM
                    • 0 Attachment
                      --- In APBR_analysis@y..., "aussievamp2" <rnmscott@b...> wrote:
                      > --- In APBR_analysis@y..., "Mike Goodman" <msg_53@h...> wrote:
                      > > Good hands are IT. A player of any size who shows up with bad
                      > hands
                      > > is pretty useless. A few players have good hands almost every
                      > day.
                      > > I would rather come without my shot than without good hands.
                      > >
                      >
                      >
                      > I agree. I have a sort of theory that assist to turnover ratio
                      > defines 'talent' to some degree for want of a better word, and
                      maybe
                      > this is it (cf difference between NBA, college, NBL, WNBA, etc.
                      etc.)
                      >
                      > Of course, I could just be whacked out, any comments?

                      Assist-to-turnover ratio reflects style of play, position, coaching,
                      strategy -- lots more than just pure talent. Isolating talent is
                      hard.
                    • Richard Scott
                      Assist-to-turnover ratio reflects style of play, position, coaching, strategy -- lots more than just pure talent. Isolating talent is hard. -- Yeah, it is.
                      Message 10 of 10 , Jul 8 5:24 PM
                      • 0 Attachment
                        Assist-to-turnover ratio reflects style of play, position, coaching,
                        strategy -- lots more than just pure talent. Isolating talent is
                        hard.


                        -- Yeah, it is. It could be 'goodness of execution' factor as well as
                        talent?

                        e.g. college < WNBA < NBL/Euroleague < NBA etc.

                        of course depends on how the stats are kept too if there is 'assist
                        inflation' in the NBA.

                        Thanks,

                        Richard
                      Your message has been successfully submitted and would be delivered to recipients shortly.