RE: [APBR_analysis] Re: SABR/Sports Econ update
- -----Original Message-----
From: schtevie2003 [mailto:schtevie@...]
Sent: Wednesday, July 30, 2003 1:35 PM
--- In APBR_analysis@yahoogroups.com, "Michael Tamada"
>> drafting young players, baseball is in a much much toughersituation
>> than basketball, and the draft is much more of a crapshoot inbaseball.
>> Mike Piazza was what, a 63rd round draft choice? Conversely[...]
>> Drafting basketball players is much easier. How muchstatistical
>> digging did Milwaukee need to do when Lew Alcindor was asenior and
>> they won the coin flip? Except for LaRue Martin, a number one[...]
>> That's not to say that drafting NBA players is easy, just easierthan
>> baseball. Sam Bowie or Michael Jordan? The choice wasnon-trivial
>> when Portland faced it. But, as DeanO says, quantify that. Ithink
>> we're a long way from being able to provide big help thePortlands
>> and other teams facing tough draft decisions. I'm sure thatDeanO
>> and other consultants provide some help. But not big help.I should clarify what I said here: obviously all NBA teams
get big help from their scouting efforts including the
statistical reports. When I say that they won't get "big
help", I mean: what are the ways in which we can improve
the current state of statistical reporting and modelling,
and how much would it help teams when they draft players?
Here's where sabrmetrics has made much progress over the
past say 20 years and has and will I think provide big
help to MLB. Whereas although there's plenty of work
to be done improving basketball's quantitative reporting,
I don't think these improved quantitative techniques will
cause a big change in how teams draft. Alcindor and
Olajuwon would still be #1 picks, and a GM doesn't need
to have a statistical consultant to tell him that.
>I have a rather different perspective on this. It seems to meYou and I are looking at different aspects of the word "harder".
>much harder to draft in basketball, if only because the stakes of
>a single decision are much higher. This is true in at least two
>senses. First, first round draft picks in basketball are expected
>to have a much greater influence on team competitiveness than
>their baseball counterparts - due solely to the natures of the
>games. Thus, errors are correspondingly more costly. And
>second, the labor market is much freer in baseball, thus the draft
>itself is a less important institution for franchise success.
Not right or wrong, just different.
I think we can re-state your point as this: the stakes are
higher in the NBA draft. And one important implication is
that any little edge that a team can get, such as by using a
new-fangled statistical technique, is more useful and important
in basketball than baseball.
True enough. But I think that that effect is overriden by
this one: stats are less useful in evaluating basketball
players than they are in evaluating baseball players. A ream
of computer printouts on a LeBron James or Michael Jordan
(or a Nick Collison; one of DeanO's useful and important
points that he's mentioned in the past is that the superstars
are easy to rate; it's the middle tier that's harder) can
tell us a little about a player, but they are relatively less
useful in basketball than the good ol' Mark I eyeball. So
one can see a player's decision making. His athletic ability
(tools). Put his statistics into the context of his teammates,
his coach, the team's strategies and use of the player, and
the quality of the opponents. And then talk with coaches etc.
about his personality and psychological profile.
Example: there's really solid evidence that in baseball, on-base
percentage is a highly important stat, one that even now is
under-used and under-reported (many newspapers now report OBP,
but it's always secondary to BA). Moreover, the sabrmeticians
claim (and I find this believable) that a players' college and
minor league OBP statistics can be used to project his likely
future ML OBP.
I don't foresee an analogous sabrmetric revolution in basketball.
Will someone someday come up with evidence that rebounds are a
lot more important than we've thought? I don't think so, because
we already know that rebounds are a pretty important stat.
And more importantly, will someone be able to come up with a
statistical formula which allows us to better predict which
college players will turn into the monster rebounders?
Second example: sabrmetrics' biggest bugaboo was and continues
to be defense. Very hard to measure well. Fortunately for them,
defense is pretty clearly a distant third in importance to offense
But in basketball, defense is half, (maybe even more) of the game,
if one folds defensive and offensive rebounds into the defensive
and offensive aspects of the game.
And defense in basketball is at least as hard to measure as it is
in baseball. It's hard to quantify (although people do measure
"stops" and DeanO has even created defensive box scores), and
EXTREMELY team-dependent (and opponent-dependent).
Kenny Anderson e.g. has a reputation as a poor defensive player.
Probably deservedly so. But, eyewitness accounts from his
final year in Boston, with Jim O'Brien, suggest that O'Brien
was able to get Anderson to put some effort into defense;
probably not enough to turn him into a solid individual
defender, but one who was willing to play the team defense game,
and help the Celtics be a decent defensive team. Certainly
the team defense stats say that the Celts had a pretty good
team defense (.425 FG% allowed, tied for 2nd best in the
NBA in 2002). And Anderson was a part of that. Not that he
could be the linchpin of a good team defense, but he was at
least going along with the program, according to sportswriters.
As DeanO says, quantify that. This change in Anderson's
behavior was pretty important, but how do we measure much
less predict it, quantitatively?
- ----- Original Message -----From: Gary CollardSent: Thursday, August 07, 2003 12:13 PMSubject: Re: [APBR_analysis] SABR/Sports Econ updateJim Armstrong wrote:
> On Mon, Aug 04, 2003 at 04:00:38PM -0500, Gary Collard wrote:
> > I'm not sure why that was so controversial. The concept of market size in
> > the NFL is pretty much meaningless, since most league revenue is shared
> > equally. The reason that a Yankees in baseball have such an advantage is
> > that they have local TV revenues that are an order of magnitude or more
> > greater than most (all?) of the other teams and is significant compared to
> > national revenue, thus they can afford to have a payroll that is 60%
> > greater than any other team even before they pay the luxury tax as they do
> > in 2003. In the NFL, there is no local TV at all, and (over a period of
> > years, letting spikes in bonus payments wash out) little payroll deviation,
> Actually, if you look at the distribution of team player payrolls, the NFL
> and the NBA are quite comparable (see standard deviation in data below).
That is why I specifically said "over a period of years, letting spikes in
bonus payments wash out" in the case of the NFL. The one year payroll
numbers you listed are meaningless to my point, do you have the data to run
them for the last 5 years or more? That will tell you who has the "harder"
Gary CollardMaybe the coefficient of variation (SD / Mean * 100) is a more apt measure for comparing the variation of payrolls for different sports across seasons.Year NHL NFL NBA MLB
1994 28.3 8.7 15.2 26.6
1995 26.6 12.7 24.1 27.7
1996 43.3 11.9 21.9 31.4
1997 #N/A 15.3 28.9 33.0
1998 #N/A 12.1 27.0 37.4
1999 33.4 12.0 23.0 43.1
2000 37.4 13.8 23.6 38.3
2001 31.1 13.5 24.6 38.3
2002 33.0 18.1 20.6 36.6
2003 35.9 #N/A 24.0 38.9On this measure, NBA teams show less variation in payroll than baseball and hockey teams, but the NFL teams are more level than any of them.Data from Rodney Fort's excellent resource: http://users.pullman.com/rodfort/SportsBusiness/BizFrame.htmed