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

Re: DeanO's Olympic interpretations

Expand Messages
  • schtevie2003
    ... Well, I would like to c that any subsequent exchange on the topic not be dispeptic, and I do have doubts that nerves (if repaired) could remain unfrayed.
    Message 1 of 48 , Sep 6 11:14 AM
      --- In APBR_analysis@yahoogroups.com, "Mike G" <msg_53@h...> wrote:
      > --- In APBR_analysis@yahoogroups.com, "schtevie2003" <schtevie@h...>
      > wrote:
      > > Maybe this is a good idea and maybe it isn't.....
      >
      > I'll bet you're still wondering.

      Well, I would like to "c" that any subsequent exchange on the topic not be dispeptic, and I
      do have doubts that nerves (if repaired) could remain unfrayed. Regardless, there is a
      larger issue of what topics are of general interest in this site. Everybody comes with their
      own perspectives, intellectual backgrounds, and prior beliefs. It is my impression that if
      group preferences were polled, empirical topics relating to NBA "history" of the kind to
      which I was referring would be noted as representing less of a group interest than current
      evaluations of players.

      I think that this is unfortunate, if for no other reason that when one evaluates players it is
      vitally important to put these evaluations in context. Much discussion is given to how
      players relate to their team concepts and how this may or may not translate across teams.
      Little to no discussion is given to how to adjust for statistics across years - even adjacent
      ones. I think the baseline assumption is that things change little over any given year and
      therefore can be overlooked with little consequence. That may or may not be the case in
      some years, but if it isn't an 800 pound gorilla, the issue is an ape of some heft.


      > >
      > >..the average team of
      > > the late 80s and 90s .. were dramatically better than the best
      > teams
      > > of the late 60s....>
      > > ... wasn't terribly
      > > interesting to many,...
      > > ...the rapidly increasing ability of the rest of the world as
      > evidence that rapid
      > > improvements in competitive ability were by no means out of the
      > question. ...
      > >...if 30 points can be gained in
      > > 10 years these areas, why would NBA history be immune from
      > significant average
      > > improvements over even longer spans of time?
      >
      > Some evidence suggests such a phenomenon is possible or even
      > likely. But certainty would require something we don't have -- a
      > time machine.

      OK, I am inspired to restate the argument below, as I firmly believe that imaginary
      technology or religiosity is not required to gain great insight.

      > Around the time you originally posted your theory, I was posting
      > corroborating evidence (based on diminishing player-minutes from
      > year to year.) There wasn't a total absence of interest from this
      > group.

      Yes, that discussion was most interesting, but also rapidly fell be the wayside. That would
      be interesting to reintroduce, if for no other reason that there have been several notable
      additions to the list of participants in this forum since that time period, and their
      perspectives would be most welcome.

      > Floating a selection of stats (and calling it "evidence" of a
      > phenomenon) just doesn't sway many people, up in here. If you think
      > offensive efficiency is equivalent to better basketball, I tend to
      > agree, for the most part.

      OK, let me begin by restating the argument of yesteryear. And please, anyone who has
      never weighed in on the topic, feel more than welcome to contribute.

      [I begin with a definition: When I refer to a "possession", I refer to what I call a "common
      possession" (DeanO has another name) which is the common number of times that
      competing teams possess the ball in a given game (and by addition, the "average" team in
      an NBA season). In rough form, this is calculated by subtracting offensive rebounds from
      conventionally defined "ball possessions". Besides being "correct", this definition is very
      useful for the purposes of intertemporal (heck, inter anything) comparisons.]

      OK. Next, let's consider the statistic of Points per Game (PPG), and more precisely, let's
      consider the average value of this statistic for a given NBA year. This yearly statistic can
      furthermore be decomposed into the product of two statistics, Points per Possession
      (PPPos) and Possessions per Game (PosPG).

      The topic I am interested in addressing is what, if anything, can changes in the yearly
      average of PosPG say about the way the NBA game is/has been played. In particular, can
      changes in PosPG "identify" whether the NBA game has improved over time. (And by this I
      do not mean aesthetic improvement, rather an improvement in the sense of improvements
      in efficiency - the greater ability, over time, to score or restrict scoring.)

      Now, let my begin by agreeing with where Bob Chaikin leaves off. There is nothing that
      one can say by looking at PPG or PPPos (alone) to address the issue in question. If PPG (or
      PPPos) increases, one could say either that offenses have improved or that defenses have
      worsened (and vice versa); one cannot identify progress with just these statistics. The
      question is: how does consideration of PosPG offer any hope for illumination?

      Here's how. But first we need to consider what constitutes PosPG. In any given game
      (and, similarly, over all games for a season) there are different kinds of possessions, each
      of which take up certain amounts of timem, and PosPG is their weighted average. For the
      sake of clarity in analysis (and doing absolutely no violence to reality) I break up PosPG
      into the weighted average of two types of possessions: fast-breaks and half-court set
      offenses. These categories perhaps don't need explanation, but I offer the following
      definition. Fast-breaks are all possessions which begin with the offensive team having at
      least a man advantage as they take the ball down the court and are characterized by this
      team trying to take scoring advantage of this numerical advantage. Half-court sets are all
      others. About this division of all possessions into two types, I make the following
      assumption that I think would be disagreeable to no one. Fast breaks, on average, yield
      more PPPos and take up less time that half-court sets (which corresponds to a higher
      PosPG). Accepting this division and definition, we move on.

      For the purpose of illustrating the argument I wish to make, let's now suppose the
      following two-year NBA "trend": in Year 1, the NBA average for PosPG is some value, and in
      Year 2, the PosPG average is less. This is to say that there are on average fewer
      possessions per game in Year 2 than Year 1; that is, we are assuming that the "pace of the
      game" has slowed.

      Furthermore, for the sake of analysis, let's impose the economist's condition of "ceteris
      paribus" which is to say that "all else is equal" in other aspects of the comparison. In the
      instance, we assume that the age and skill structure of the hypothetical NBA is identical in
      the two years in question. Furthermore, let's assume that there are no tactical or strategic
      innovations that would influence scoring or the pace of the game.

      What can we infer from this analytical set-up?

      Well, straight away, we know that up to two things could have happened. Either the
      average mix of plays in the NBA shifted away from fast-breaks to half-court sets, or that
      more time was spent, on average, in each of these two general offensive categories, or
      both. Assume that it was the first, that there were fewer fast breaks and more set
      offenses, what would this imply? The direct implication is that aggregate offensive
      productivity (PPPos) should be expected to fall, as fast breaks yield more points on
      average than do set offenses. Now assume that it was the second, that offensive
      possessions of both types took longer. What would this imply? Well, the bottom line is
      that there is no reason to expect on theoretical grounds that PPPos would increase and if
      anything, one would expect that PPPos would decrease. Consider the two cases. First, if
      the average time of a fast-break were to increase (essentially an increase in the time it
      took for a shot to be released) it would imply that defenders would have more time to get
      back and defend, in turn implying an expectation of decreased PPPos. Second, if the
      average time of a half-court set increased, this would mean that offender were being
      forced to wait a little longer to get an equivalent shot as in the preceding year. So if
      anything, one might expect the PPPos to decrease.

      The bottom line then is that a slowing of the pace of offense/the game (decreases in
      PosPG) given an NBA of constant skill/age and "technology", implies only one thing as
      regarding offensive efficiency (PPPos). PPPos should decrease, or in other words, the
      average defense should improve relative to the average offense in realizing that there are
      gains to be had in preventing fast breaks.

      But what then are we to make of the very consistent trends, over roughly a quarter
      century, where the pace of the game DRAMATICALLY SLOWED but offensive efficiency
      INCREASED? There needs to be some theoretical explanation, and there is. By inference,
      absent any other offsetting explanation (partial or full), what must be the case is that the
      "technology" of the game improved. In particular, what must have happened is that team's
      in the run and gun 50s and 60s were shooting too quickly and forgoing the better shot
      opportunities that patience would have earned them. And it is in this general sense, that
      team's in the late 80s were so dramatically better than their 60s counterparts; they were
      more patient.

      I say that this is what must have been the case, because there is no compelling evidence
      that there were consistent rule changes that lead to the near monotonic trends in
      question. Also, there is no evidence that the average NBA player of a later era was a
      dramatically better shooter (defined as better able to make shots, location by location)
      than his counterpart of a previous era. And though I would argue that there is compelling
      evidence that the modern NBA player is a much better athlete (which might have some
      influence on these trends, but it isn't clear that the gain would have been a net plus for the
      offense rather than the defense) the theoretical argument above has nothing to do with
      this impression.

      Well, that is a rehashing of the argument, hopefully more clearly stated than the last time
      around (in any case, slightly differently so). I look forward to all comments, especially
      from those who are relatively new to the group, which would address the logic of the
      theoretical argument (I don't think that there is any substantive disagreement over the
      empirical record of the NBA averages themselves).

      And should the argument hold (as I am quite sure it does) it would introduce an
      explanatory variable that could possibly account for the "rise of the foreigner" in
      international basketball. In particular, rapidly converging on an optimal game pace may
      be an important explanation of the rapid improvement of the international game, in
      addition to an increase in skill, athleticism, etc.



      > Whether that makes '90s NBA teams more competitive than '60s teams
      > depends on the game environment. This was decisively illustrated by
      > these Olympics. Argentina would not do well as an NBA team. USA
      > (this bunch) did not do well as an Olympic team.

      Just as a final note, my argument about modern teams being more competitive has
      nothing to do with the differing game environments in a very important sense. I believe I
      have generally identified the superiority of an average patient team over an average run
      and gun team. Patience was an option available in every era and in every historical
      environment (unless it wasn't, and in which case I would like to hear an argument as to
      why) so I have to disgree with the above formulation. As to how Argentina would do as an
      NBA team, it is by no means clear to me that they would stink, just as it is not clear to me
      that if the USA team, knowing all we know after the fact, would not still be the gold medal
      favorite, were the Olympics to start today.

      That said, I take your point, and I look forward to all the constructive commentary on what
      I think is a very important topic in whatever it is that we generally like to talk about.
    • schtevie2003
      ... Out with emotion, in with on point arguments. ... I am not sure which reasoning you are characterizing. But I don t think we are disagreeing here, at
      Message 48 of 48 , Sep 20 1:03 PM
        --- In APBR_analysis@yahoogroups.com, "carlos12155" <carlosmanuel@b...> wrote:
        > I'm reluctant to post in this topic, because for some reason it seems
        > too emotionally loaded, but here I go. Schtevie, I think that your
        > reasoning has some flaws and will try to point them out.

        Out with emotion, in with on point arguments.

        > A)The notion that other things being equal a faster pace should mean
        > improved offensive efficiency only holds true if the faster pace is
        > the product of more fast breaks and the offensive efficiency in half
        > court sets remains the same. If the faster pace is the result of
        > shooting earlier in half court sets, there is no reason offensive
        > efficiency should improve. A faster or slower pace per se does not
        > tell us anything about offensive efficiency.

        I am not sure which reasoning you are characterizing. But I don't think we are disagreeing
        here, at least not completely. I am saying that we know that average time per possession
        increased. This could have come about either by relatively fewer fast breaks in the mix -
        which implies a decrease in offensive productivity, as fast breaks are "better breaks" - or
        by more time spent in half-court sets - which should not imply an increase in offensive
        productivity, if anything, perhaps a decrease.

        You write "if the faster pace is the result of shooting earlier in half court sets [now me: this
        is equivalent to going backwards in time] there is no reason offensive efficiency should
        improve". This is my point exactly, going back in time, the offenses were worse.

        Hence, I disagree with your last sentence. It is precisely because (in going backwards) that
        the pace quickened (and despite teams presumably having the benefit of a higher
        proportion of fast breaks) and their productivity dropped which implies that they were
        playing sub-optimally (i.e. they could have done better than they did, just by playing more
        under control and waiting for a better shot.)

        > B)You discard too easily the possibility of another explanation. Maybe
        > another factor made the offensive efficiency of half court sets
        > improve and as a result fast breaks became less neccesary. These
        > alternative explanations need to be discussed and shown to be false.

        The attractiveness of the argument I am making is based on the fact that as a result of the
        basic theory and facts, we know that the game changed in a certain direction (towards
        greater "efficiency"). And, we know what this means in terms of turnover and scoring
        percentages, overall they increase - by "definition". Furthermore, if nothing else changed,
        it is necessarily true that the fact that the game slowed during this time period implies
        that things could have been done better previously, but weren't. As to other explanations,
        I invite anyone to suggest them, and I have identified the categories where they are likely
        to be (differing rule changes/interpretations, changes in populations, etc.) And I agree
        that they should be discussed, but they need not be shown to be false for the game
        improvement argument to hold. What needs to identified are the relative contributions of
        various factors. (So, at the end of the day, one might say that the actual gain in offensive
        efficiency from the 60s (say) to the 80s (say) was x% due to refs deciding that palming the
        basketball was no longer a turnover, y% due to improved shooting ability, and z% due to
        an increase in athleticism, etc.) Right now, however, I am trying to establish the ceteris
        paribus argument that fast play was hasty play, causing many, many, forgone points.

        > C)You assume that the offensive team controls the pace when in fact
        > the defensive and offensive team control it. Now, it's true that the
        > correlation between a slowing pace and improved offense suggests that
        > the offensive team was causing the change, but it doesn't follow that
        > improved shot selection was neccesarily the cause.

        I actually do believe that the offensive team primarily controls the pace in this instance
        because it is the plausible story. (The alternative is an odd story that the offenses were led
        to their improvements by defenses playing better! This story would be that the defenses
        were eliminating fast breaks and playing harder defense, obliging teams to protect the ball
        more and be more patient, with great unexpected gains in productivity.) But for my
        argument to play out, it depends not at all on the assumption of offenses consciously
        slowing down play or being induced superior performance by better defense. (In which
        instance, the superiority of the moderns versus the old school is all the greater still.)

        And you are correct that improved shot selection is not neccesarily the cause of increased
        productivity. But, there is no strong anecdotal evidence that players became better
        shooters (and I don't think that free throw evidence would support it) so, by elimination,
        that leaves shot selection (and better ball control/fewer turnovers).

        > D)You assume that the scoring skill of 60s players was comparable with
        > today's players on the basis of their similar free throw shooting. It
        > seems to me a bit too simple.

        Yes it is simple, but one can argue in terms of broad categories. To repeat: if productivity
        increased, either there were more shots available (i.e. fewer turnovers) or they were better
        shooters (but not better free throwers?) or they got better shots (call it shot selection.)
        That is all possible categories, no?


        > Now, I agree with the general idea that 60s teams probably played too
        > fast, but before stating it as a fact we should try to prove it.

        I have given a proof (again, barring no "out of bound" explanation like changes in rules or
        their interpretation, asymmetric effects of improved or worsened athleticism, a deeper or
        shallower talent pool, etc.) Fleshing out the story of what happened and when is not proof,
        but confirmation.

        A
        > couple of ideas are, 1) Tracking the offensive efficiency of
        > individual players who played during the time period. If there was a
        > "coaches driven" change, their off. eff. should show an evolution
        > quite different from today's players.

        The reason I say this is not proof but confirmation is that we know that if the league
        average moved in one way, it will be reflected in averaged player averages. I think that the
        more interesting story is to try to explain why it moved in the way it did, and my strong
        prior is that league-wide strategy changes were based on emulating teams perceived as
        "successful". As I have noted, my guess is that perceptions of success related (and
        probably still relate) more to trying to "be like Mike, or the Celtics dynasty, or the Bad
        Boys, or whatever" rather than "be like the team that maximizes offensive efficiency".


        2) Compare data with foreign
        > leagues (where we can assume there is less scoring skill) and see
        > where the difference in talent shows up. If we could get data about a
        > long period of time, it could be interesting to see whether we can
        > detect statistically the improvement by international players that has
        > been seen.

        Sounds good to me.

        > Carlos
        >
        >
        > --- In APBR_analysis@yahoogroups.com, "schtevie2003" <schtevie@h...>
        > wrote:
        > > --- In APBR_analysis@yahoogroups.com, bchaikin@a... wrote:
        >
        > > What exactly is your trouble with the notion that slowing game pace
        > and inc=
        > > reasing
        > > offensive productivity directly imply that teams in the base time
        > period we=
        > > re not exhibiting
        > > optimal ball control and shot selection? The idea couldn't be more
        > simple.=
        > > If you have
        > > trouble with this piece of theory, state it. If you think there
        > were offse=
        > > tting factors
        > > otherwise explaining the empirical phenomenon, identify them.
        > Otherwise, s=
        > > tate that you
        > > don't (in a civil fashion.)
        > >
        > > Finally, perhaps, you can expound further on the general issue of
        > posting p=
        > > ropriety and
        > > identify what are legitimate and illegitimate topics of
        > conversations in th=
        > > e "stats group". In
        > > this learned discussion, be sure to include at least some mention of
        > the re=
        > > lative virutes of
        > > simulations and formal statistical analysis; that should be interesting.
        > >
        > >
        > >
        > > > bob chaikin
        > > > bchaikin@b...
      Your message has been successfully submitted and would be delivered to recipients shortly.