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

Quick comment on Talent Dilution

Expand Messages
  • Dean Oliver <deano@rawbw.com>
    I think Bob is missing the point -- people do have questions about the value of stats today vs yesteryear. Assuming they are the same avoids the real
    Message 1 of 2 , Dec 19, 2002
    • 0 Attachment
      I think Bob is missing the point -- people do have questions about
      the value of stats today vs yesteryear. Assuming they are the same
      avoids the real question. And that is why so many statheads get
      ignored by the real coaches out there. They miss the point. The
      game IS different now than before. The athletes are trained
      differently. The styles are very different. Saying that collecting
      modern stats back in the '50's would adequately represent that to the
      point that you'd trust a cross-generational simulation is naive.
      There is no way to calibrate such a thing until someone develops a
      real time machine, at which point, I don't think the first thing I'd
      do is have the 69 win Lakers play the 72 win Bulls.

      Cross-generational competition is great bar talk, a hell of a lot of
      fun, but how you spin it determines a lot about how people conceive
      of the result. I think it all is interesting in some way
      scientifically, too. But saying that the answer is known as soon as
      the stats are known (even probabilistically known) -- that ends a fun
      discussion and is just close-minded.

      Maybe Bob has a real time simulator that says who will win the NBA
      title in every season 5 years down the road and in how many games.
      If he does that, I'll start believing.

      Gotta go. Have great Holidays. I'm back in a couple weeks (with
      maybe occasional viewings).

    • bchaikin@aol.com
      I think Bob is missing the point -- people do have questions about the value of stats today vs yesteryear. no - i fully understand this - of course there are
      Message 2 of 2 , Dec 20, 2002
      • 0 Attachment

        I think Bob is missing the point -- people do have questions about the value of stats today vs yesteryear. 

        no - i fully understand this - of course there are questions as to how to compare eras. that's why we are discussing this. this does not mean we can't try to simulate games between eras. the question is how do you start the process? alot of readers here are saying its not possible, or if so that i am not taking into account "talent" levels or some other factors yet made clear. am i correct in that the general assumption that most readers here have is that they are assuming that todays player is "much better' than 30-40-50 years ago? better basketball players? or better athletes? or just how are they "better"?.......

        i am simply trying to get people to see the "other side" of this discussion. pretty much all of the responses i have read concerning the computer simulation of games between teams from the various decades is that i am not taking into account some "nebulous" entity for correlating players of yesteryear to the players of today. but while a number of readers are saying this, they are not saying what this or these "factors" are, what they should be, how to correlate them, or how to calculate them. they are simply saying things are not being taken into consideration, while still others are simply saying it can't be done so don't even try, i.e. "....i wouldn't trust the simulator..."....

        Assuming they are the same avoids the real question.  And that is why so many statheads get ignored by the real coaches out there.  They miss the point. 

        oh really? is that so? stat heads get ignored by the "real" coaches out there? is this your opinion or is this some publicly disseminated fact? i know (knew, rather) of coaches (coaching staffs) in the nba that used stats and stats analysis to no end, and others that avoided them like the plague. where are you getting your information from? or, better yet, which of these nba coaches do you consider the "real" coaches? they'll all "real" to me....

        The game IS different now than before. 

        fine, for the sake of this discussion i'll take your word for it - for now. but only if you explain to me HOW it is different. scores of games the past few years are similar to the mid-to-late 1950s. here are some numbers:

                   FG%  FT%  PTS/G  FGA/G  FTA/G  REB/G  PF/G  TMS
        54-55   .385   .738     93.1      86.6     35.9      56.1    24.8     8
        55-56   .387   .745     99.0      91.4     38.0      60.1    26.4     8
        56-57   .380   .751     99.6      94.6     36.9      62.4    25.1     8     
        57-58   .383   .746    106.6    101.9     38.3     71.7     25.5     8
        58-59   .395   .756    108.2    102.3     36.3     70.0     26.0     8
        59-60   .410   .735    115.3    108.7     35.8     73.5     25.6    8
        60-61   .415   .731    118.1    109.4     37.5     73.3     26.8    8
        61-62   .426   .728    118.8    107.7     37.1     71.4     25.9    9
        62-63   .441   .727    115.3    101.2     35.9     66.7     26.0    9 
        63-64   .433   .722    111.0      99.1    35.0      65.9     26.8     9  
        64-65   .426   .721    110.6      99.8    35.6      67.3     25.9    9

        84-85   .491   .764    110.8      89.1    29.4      43.5     24.9   23
        85-86   .487   .756    110.2      88.6    30.3      43.6     25.2   23 

        97-98   .450   .737      95.6     79.7     26.3      41.5     22.4   29
        98-99   .437   .728      91.6     78.2     25.8      41.7     22.2   29
        99-00   .449   .750      97.5     82.1     25.3      42.9     23.3   29
        00-01   .443   .748      94.8     80.6     24.9      42.5     22.3   29
        01-02   .445   .752      95.5     81.3     23.8      42.4     21.2   29

        take a good long look at these "facts" (not "assumptions"). scores of games today are very similar to the mid-1950s. granted the avg league overall FG% in 98-99 is .05 better than the 55-56 season, but guess what - its .05 worse than the mid-1980s! i don't see or hear anyone telling me i'm not correctly cross-correlating from today's game compared to the mid 1980s. from what i'm reading everyone is 'assuming" the game today compared to the mid-1980s is basically the same...

        The athletes are trained differently. 

        absolutely - very few players lifted weights in the mid 1950s, or trained year round. but many had "physical" jobs in the off-season to "stay in shape"...

        i'm guessing that many of the readers here are saying that the players of the 1950s couldn't play up to the players of today. well then reverse this logic - put players of today back into the mid-1950s, when no one was lifting weights (teams didn't have weight lifting rooms in their arenas), or training year round. no one had saunas, or trainers baby-ing the players, when travel was by train mostly, put any player of today back in those times and see how they would play differently than they play today. when comparing teams of different eras, you shouldn't simply say that the players of yesteryear couldn't play in today's game, also you should ask yourself how would today's player have played back then, when the conditions for being professional athletes were tougher - less pay, harder travel, probably poorer nutrition (from what i've read)...

        The styles are very different. 

        fine - prove, or tell me what you mean, by this statement. tell me how the "styles" were/are different. what exactly do you me by "style"? the stats above don't show me a ton of difference, any more different than today's game versus the mid-1980s when scores were 10%-12% higher, and shooting was that .05 better than say 98-99. so explain to me what actually was different. what style was used in the mid 1980s versus today?...

        Saying that collecting modern stats back in the '50's would adequately represent that to the point that you'd trust a cross-generational simulation is naive. 

        is this someone's "professional" opinion as one who analyses nba stats? that's its naive? MW defines "naive" as "marked by unaffected simplicity" and also as "deficient in worldy wisdom or informed judgement". right! so take whatever evidence you can get, and that's the stats (and video tape if possible). what else is there to look at? where else would you start? you start simple and go from there. plus if you had all the stats they keep today kept back then, the process would be even easier...

        would it exactly represent that to the point you'd trust a cross-generational simulation? probably not. could it adequately represent it? certainly, and can do so for teams from the late 1970s to today because the stats were kept and are used in the simulation. tell me, where would you start? you seem to say no you can't to everything here, you can't do this, or you can't do that, but if giving the problem of doing so, would you just say it can't be done, its impossible. fine - you take that attitude and others like myself will attempt to do so. getting all the stats would be the first thing i would do, and go from there (now if only i could get my hands on old game tapes)....

        There is no way to calibrate such a thing until someone develops a real time machine, at which point, I don't think the first thing I'd do is have the 69 win Lakers play the 72 win Bulls. 

        says who? you? that there is no way? MW defines "calibrate" as "to standardize by determining the deviation from a standard so as to ascertain the proper correction factors". you mean with your sophisticated statistical analyses for pro hoops you can't come up with standards for each season and then attempt to equlibrate them with some type of correction factors or coefficients? if you do i'll be happy to try implemeting them into the sim and see how that affects the outcomes. i'd be happy to get your assistance....

        tell me, is the runs created type formulas for baseball simply hogwash? they seem to correlate across generations quite well, but then again baseball has complete stats for almost a century while basketball does not. kinda makes the task a bit easier, don't it?...

        Cross-generational competition is great bar talk, a hell of a lot of fun, but how you spin it determines a lot about how people conceive of the result.  I think it all is interesting in some way scientifically, too.  But saying that the answer is known as soon as the stats are known (even probabilistically known) -- that ends a fun
        discussion and is just close-minded.

        aahhhh...so if someone "scientifically" creates a basketball simulation based on statistical analysis (which is what i thought this discussion group was all about - using the analysis of basketball stats to further understand the game of pro hoops - and isn't a simulation based on stats research simply an extension of the statistical research?), it takes the "fun" away and is "close-minded"?? in attempting to simulate basketball on the computer for teams from various decades all you've said so far is that it can't be done, and given your reasons as to why. so tell me now, who is the "close minded" party here?...

        Maybe Bob has a real time simulator that says who will win the NBA title in every season 5 years down the road and in how many games.  If he does that, I'll start believing.

        real nice - but again in previous postings i've mentioned the sim is best used for "prediction" using current stats during a season - for that season, and the final stats in the off-season for the next season. i never said it could be used as a crystal ball for years down the road - you did. but then again, using your quote, "...how you spin it determines a lot about how people conceive it..."....

        One of these days, I'll have to go in and do a Stephen Jay Gould-style standard deviation look at the eras, but just eyeballing the stats, here's what I came up for with the top 50 players in minutes each season:

        Year FG%*  StDev    StDev/FG%
        5455 38.694% 0.034438 8.90%
        8687 50.612% 0.037028 7.32%
        *unweighted average

        The standard deviation as a percentage of FG% was a LOT higher in 1954-55,
        a key indicator that the level of competition was lower then than it is

        it's my understanding that on average the larger the sample population size the less skewed the standard deviation (which provides a precise measure of the amount of variation in a group), and as the square root of the sample population size is the denominator with the numerator being the square root of the summation of the avg FG% minus all the values in the set, then by definition on average you stand a much better chance of having the standard deviation skewed by a smaller sample population size, and in the mid-1950s there were 8 teams (<100 players) and in the 1980s and 1990s 3-4 times as many teams and players. thus the sample population size is 3-4 times as large...

        Possible? I know that the game hadn't taken to the streets yet, but the
        lack of any minorities is a telling factor, 

        i wholeheartedly agree how one would question not all the best players were in the nba in the early 1950s. thus you then have to look at what happened when the pro teams played a team like the globetrotters (or any other barnstorming teams composed of minorities or groups of any races/ethnicities not allowed to play in the nba). if these teams consistently beat the nba teams, and did so by wide margins (like the globetrotters historical beat the washington generals) i'd say you were absolutely right, that the nba did not have the best players or players of equal talent to the best players wherever they were. but from everything i've read, when nba teams played teams like the globetrotters or other barnstorming teams the competition was of equal value...

        as were the locations of the teams: New York, Philadelphia, and Boston as larger markets, Milwaukee and Minnesota as mid-sized to small ones, and Fort Wayne, Indiana, Rochester, and Syracuse, New York as tiny ones. There are no Fort Waynes in today's NBA, and there weren't any in 1986 either. Given the appallingly small level of revenue that must have been generated (1954-55 wasn't long after
        the time that the Northeastern teams used to bring the Globetrotters in after league games were over in order to bring in more cash), it's not unfeasible at all to say that more potential pro basketballers were choosing other fields, not just other pro sports but higher-paying jobs such as accounting and banking.

        fine, as i stated in a previous posting, do you know of any examples of players who were good/great college players in the 1950s, that pro coaches and players who made it to the pros said would have made the pros and possible been good or even great? i mentioned bob kurland, a 7 footer who in the mid-to-late 1940s was considered a great college player and a rival to mikan, but ended up getting a job and playing on that company's team in aau games i believe (phillips petroleum?)......you know of any other examples, or just making this assumption?....

        i understand its not unfeasible, this is true, but lets try to gauge what the ratio was. do readers think that for every player in the nba there was another who could easily have been there but opted for getting a different job other than the nba? or 5 for every 1 nba player, or 1 for every 5 nba players? that would certainly help determine if the best basketball players were in the nba. but again from what i've read in books like bob peterson's cages to jumpshots, salzberg's from set shot to slam dunk, and others, was that by the mid-to-late 1950s the best players were in the nba. have you read or somehow know differently? i'd like to know....

        Again, just because the players represented the top players of 1954-55 doesn't mean that their statistics are directly correlatable,

        and it doesn't mean that they do not correlate.....i'm saying that we can use the stats to make the correlation as there isn't a whole lot of other evidence available. if you want to simulate era versus era, what would you use? are you saying that the stats are not able to correlate. if so why? let's hear some reasons...

        any more than the top rock musicians of today's San Jose are just as good as the
        top rock musicians of San Francisco in the 1970s.

        > the only thing preventing solid simulation of 1950s teams versus say 1990s
        > team for pro basketball is the lack of complete statistics for players of
        > that earlier time.

        ...and the much larger difference in talent...

        define "much larger difference in talent". what do you mean when you say this? this is a very general statement - are you saying flat out that today's players are "better basketball players" than earlier years, or just "better athletes"?  if so, where is the cut off? what year or years did players first compare to today's players? when did players first become as good as today's players? when arizin was around? nat clifton? harry gallatin? schayes? pettit? cousy? russell? chamberlain? west? robertson? or not until bird, magic, and jordan? lets hear where this cutoff starts and why.......i'd be very interested in hearing/reading this....and knowing your reasons why...

        mickey lolich and john kruk were good if not great baseball players - i wouldn't call either one much of an "athlete". randy white - the basketball player drafted by dallas in the late 1980s - i remember looked like a karl malone clone, all buff and looking like adonis, a potential decathalon athlete wannabee, but he didn't turn out to be much of a pro basketball player...

        The 1960s are an entirely different period. If you said that the 1954 Minneapolis Lakers or Syracuse Nationals *would* beat the Memphis Grizzlies every time out, I'd say you're crazy.

        other than by creating a basketball simulation and refining it to re-create real life results and stats, how else would you know? this is what i am saying....

        The 1884 St. Louis Maroons were a professional baseball team in a league regarded by some, at least, as one of the major ones; they won nearly 70% of their games that year. Do you believe that they would defeat the Tampa Bay Devil Rays in a 7-game series? I don't.

        you are thinking of course of bring the maroons up to the present day to play the devils? what if you took the entire tampa bay team, and made them play under the conditions baseball was played in in the late 1800s - then yes i thnk very very possibly the maroons would win. the mits they used (if they even used mits back then) would be nothing like today's mits, more like winter gloves. the equipment would be more "primitive" compared to today's equipment, no?...how would the tampa bay devils rays play under those conditions?...

        Yes, I can see the axe that you have to grind. Looks like I'm going to have to work that standard deviation thing from the onset of the NBA to see how rusty said woodworking tool is...

        what axe might that be? what agenda are you referring to? by all means do your standard deviation throughout history. the numbers will be more statistical evidence that can be used for a cross-generations simulation...

        by the
        > mid-to-late 1950s and early 1960s, i certainly don't see any  problem with
        > those players playing against today's players and being  competitive...

        Well with Wilt, Russell, Baylor, etc. you'll get no argument.  Sure they could play today. But you'll get plenty of argument about the level of play of the players below that level, in the 1950s compared to the 1970s or later

        why could they play but not pettit, arizin, schayes, cousy, nate clifton? what's so special about those three players that wasn't special about the others i mentioned? what criteria are you using for this decision making?....

        again - where is the cut off? what year or years did players first compare to today's players? when did players first become as good as today's players or when could they first compete - in your opinion? lets hear not only where you think this cutoff starts but why.......

        We've already had discussions (maybe it was in APBR rather than here) about the stories of how Russell drove former scoring champ Neil Johnston out of the league.  And how, good as Cousy was, his ballhandling skills just don't compare with today's PGs -- I think it was Cousy himself who noted that he simply didn't have the
        ambidexterity, the ability to use his left hand and go to his left, that today's players have.

        not sure i understand the point you are trying to make here. i missed that discussion. was the jeest of it that russell was a much better player than johnston? johnston was 1st team all nba center for 4 straight years, from 52-53 to 55-56, and 2nd team in 56-57, russell's 1st year in the league. he ws out of the league a year later, but i don't know why. he was a better scorer than russell ever was, but russell was not only a better defensive center than johnston, but possibly the best defensive center ever. and for fyi - russell drove alot of other players and coaches, maybe not out of the league, but to drink anyway. he was dominant for quite some time on defense...

        again cousy's remarks may be true, but he was the premier ball handler of his time. very few players used both hands dribbling back then. you think that means they couldn't play basketball? you ever see jerry west play? he was all-pro for a decade, and everyone knew he wouldn't/couldn't "go left" and rarely dirbbled with his left hand (i've seen enough tape of him to know this is true). didn't seem to stop him tho. again - it was just the way the game was back then - the "style" as some of you may call it....

        are you saying that if you placed players with excellent dribbling skills from today back into the 1950s they'd dominate simply because they could dribble with both hands? back then a premium was put on shooting ability and team play, passing and movement, not today's isolation plays and grand dunking - alot of today's players may very well likely not have played back then (speculation of course) if they couldn't consistently hit a jumper or set shot, or moved and gotten open without the ball, because back then that was what was required by coaches....

        In other words, BobC's simulation doesn't account for changes in talent levels. 

        this your opinion? or a factual statement? have you even used the simulation yourself so that you can then quantitatively make a statement one way or the other?...

        and again how are you defining talent? what does talent mean? the ability to hit a jump shot? dribbling? the ability to play within a team concept? the ability to run all day and play all game every game? or just high jumping and dunking?...

        It *assumes* that they are equal in different years.

        you have a better "assumption"? if so what is it, and how are you determining it?....

        Which is fine for comparing 1998 to 1999; is probably okay for comparing 1977 to 1984; but gets dicey beyond that.  Even a 7 year jump is risky if the 7 years are 1956 to 1963; BobC himself in an earlier article noted that (quoting from his message):

        so then is the 15-20 year period of the early 1980s to today not okay? because its a much longer time period? was the game vastly different in the late 1970s to early 1980s than today? how about the late 1960s/early 1970s compared to today? is it ok to say the past 30 years are equally comparable? the past 40 years? the past 50 years? where is your cutoff - and why? what's your answer and what evidence are you using?...

        BobC disagrees with the claim that the overall quality level was higher in 1963 than it was in 1956, but even he agrees there is some effect on the overall talent distribution when the NBA experiences an influx of talent.

        an influx of really good to great players - yes - but not just an influx of players. each year the nba drafts players (remember the draft used to be 10-12 rounds long, not just 2) and the overall talent/quality level didn't really change much from year to year.  but for example when the ABA folded and the NBA took in some of its players, it took in only the very best, those who were good/great in the ABA and were good/great college players a few years earlier. does this mean that we cannot correctly simulate on the computer games between teams from before the 76-77 season to teams in the 76-77 season and later? what in your research or opinion was the difference between the 75-76 and 76-77 season, versus the 1956 to 1963 seasons? can i or can i not correctly simulate games between teams in 75-76 versus 76-77 in your opinion, and between 1956 and 1963, and why or why not? remember in 76-77 almost half the all-star team was former ABA players and 4 of the 10 1st and 2nd all-pro teams was former ABA players...

        my point earlier was that if the league has only 8 teams and you bring in 5-7 superstars in a few years time (chamberlain, west, robertson, baylor, russell) the effect is much bigger because the amount of total players is smaller. with the 29 teams of today you'd need 3-4 times as many superstars to enter the league in the same short time period to have the same effect, or don't you agree with that?...

        BobC likes to claim that he uses "facts" while the rest of us are making "assumptions", but to baldly plop statistics from one era into another era, without any translation or deflation, is making a huge, and for periods of, say 20 or 30 years, almost certainly untenable assumption.

        i too make assumptions but try to back them up with stats or analysis - that's the whole point of this discussion group - no? the above stats list with game scores, FG%, etc are facts - care to dispute them? how do you explain them? they show me that statistically speaking any differences between the game back then (1950s) and today is small. examples - scoring in the mid-50s is similar to todays game, same game, 4 quarters with 12 minutes a quarter. players shot as worse in 98-99 compared to the mid-1980s as players in the mid-to-late 1950s shot compared to 98-99? care to dispute this? what do you make of this? players in 62-63 shot better than in 98-99, what does that mean? did players in 62-63 only shoot from 15 feet and in? i don't think so - the game was a faster paced game and they assuredly got more layups...

        do you have any evidence to the contrary that the stats above don't show this - i'd like to read it. and what translation or deflation? what do you mean by "...without any translation...."? and why only deflation and not inflation? back then players got more rebounds than they do today, or don't you take this into account? you are only assuming deflation - why? again put player of today back in the 1950s and early 1960s to play the game the way it was played back then and under the conditions the game was played back then don't you think they'd get more rebounds on average? the game was a faster paced game in the early 1960s - teams averaged close to 120 pts/g in the early 1960s. wouldn't you have to "inflate" their rebounding to match that?..


        bob chaikin

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