PDA

View Full Version : How the Midmajors Reel in Those Stars-interesting piece in today's NYT with a Duke



SFDukie
03-16-2010, 07:33 AM
Connection:

March 15, 2010
How the Midmajors Reel in Those Stars
By PETE THAMEL
Dave Telep evaluates high school basketball players for a living. While most fans watch the N.C.A.A. tournament to root for their alma mater or to follow their brackets, Telep sees a referendum on his professional opinion.

Too many times, Telep said, he will see a player like Patrick O’Bryant, a 7-footer who led Bradley on an unexpected run to the Round of 16 in 2006, and torture himself for not having identified him as a high-caliber prospect.

So Telep, the national recruiting director of Scout.com, decided last summer to execute a semi-scientific evaluation. Call it a study of Cinderella’s DNA: what makes up a great midmajor player and why those players were not recruited by traditional powers.

“You can only do my job for so long before you flip on the N.C.A.A. tournament, look at the guy from Bradley, shake your head and wonder, ‘How did that happen?’ ” Telep said. “The whole goal of this study was to lessen my angst during the N.C.A.A. tournament.”

Telep spends his summer canvassing the country watching games from Las Vegas to Orlando, with many Marriotts and airports in between. So he had his summer intern, a Duke sophomore named Drew Cannon, dig up background information on 32 players who made first-team all-league in conferences like the Atlantic 10, the Colonial Athletic Association and the Missouri Valley.

From Davidson’s Stephen Curry to Xavier’s B. J. Raymond to Old Dominion’s Gerald Lee, the study analyzed the players and why they were overlooked. Some results were obvious...

NYT:How the midmajors... (http://www.nytimes.com/2010/03/16/sports/basketball/16ncaa.html?hp.)

weezie
03-16-2010, 07:37 AM
Paging Drew Cannon! Perchance he posts here?

ice-9
03-16-2010, 08:05 AM
Hmmm...this study feels incomplete. It does a good job of explaining why a prospect ended up at a mid major than a high major (too skinny, too short, etc.), but it doesn't answer whether a school should recruit a tall but skinny player with no obvious position. How often does this player turn out to be a first team all-conference? How often a bust?

cspan37421
03-16-2010, 08:41 AM
I agree, jyuwono. It's sort of a fallacy of composition. You say that "midmajor stars had 2 of 3 characteristics desired/sought by major programs" and then you say "Player X has 2 of 3 such characteristics, therefore he'll be a midmajor star."

Richard Berg
03-16-2010, 10:41 AM
My guess is that the Duke statistics major understands this fallacy just fine, but the journalist does not. (or simply ran out of column-inches)

brevity
03-16-2010, 10:49 AM
Hmmm...this study feels incomplete. It does a good job of explaining why a prospect ended up at a mid major than a high major (too skinny, too short, etc.), but it doesn't answer whether a school should recruit a tall but skinny player with no obvious position. How often does this player turn out to be a first team all-conference? How often a bust?

I'm assuming it's just the news article that's incomplete.

Good luck to Mr. Cannon. How competitive is the field of sports statistics? On the one hand, there's a lot of number-crunching already out there. On the other hand, sports fans never tire of it, so there's always a demand. Maybe a pro analyst can stand out among the many unpaid amateur enthusiasts who dominate the arena.

The Houston Rockets have a recent track record of benefiting from statistical analysis. That Duke connection could come in handy...

patentgeek
03-16-2010, 05:47 PM
My guess is that the Duke statistics major understands this fallacy just fine, but the journalist does not. (or simply ran out of column-inches)

My guess is also that the Duke statistics major understands the incompleteness that has been mentioned, which is likely a result of the interpretation by the article's author or the lack of available column inches. As I understood the article, it would seem that the study revealed that a player with advanced skills/intangibles but lacking, say, height or build may well be a good gamble for a mid-major program (and would therefore be an attractive target for a mid-major), but may not be for a major program. Part of the "fallacy" that posters have mentioned is inherent in recruiting - some players, even those of the "can't miss" variety, sometimes just don't pan out. My guess is that the study mentioned in the article is intended to improve the odds for mid-major coaches by noting which combinations of traits/skills tend to work out better than others (but with no guarantees). And improving their ratio of hits to misses would have to be valuable.

patentgeek
07-15-2010, 08:29 AM
Looks like the same Duke student whose study on mid-major players was written up in the NYT in March is at it again.

http://www.basketballprospectus.com/article.php?articleid=1183

This article proposes a potentially improved technique for analyzing team strength/rankings using in-game win probability numbers. I'm not sure it's perfect (the author points out some potential flaws), but I thought the idea itself was interesting.

CrazieDUMB
07-15-2010, 08:52 AM
I love this stuff. Maybe a better way to hold off problem number 2 (about how the first half is weighted more) is to reset each chunk of time to 0-0. He could break down the game into 10 4-minute chunks, assign a win probability to each chunk as if it were extrapolated for the whole game, and average them all. Would that work?

sagegrouse
07-15-2010, 09:20 AM
I agree, jyuwono. It's sort of a fallacy of composition. You say that "midmajor stars had 2 of 3 characteristics desired/sought by major programs" and then you say "Player X has 2 of 3 such characteristics, therefore he'll be a midmajor star."

Actually, it's a "post hoc, propter hoc" fallacy: "after which,, therefore, on account of which." It addresses spurious correlations. :rolleyes: An example of "fallacy of composition" is UNC 2010 -- nine McD All-Americans just "have to" produce a top ten team. :eek:

sagegrouse
'"Fallacy of composition: a fallacy in which what is true of a part is, on that account, alleged to be also true of the whole." -- Paul A. Samuelson, Economics, Fifth Ed., 1961.'

mgtr
07-15-2010, 03:48 PM
I think you might want an "ergo" in the middle of your quote. :D

sagegrouse
07-15-2010, 07:51 PM
I think you might want an "ergo" in the middle of your quote. :D

Yep. This is one I should have looked up: "post hoc, ergo propter hoc" is what I meant.

sagegrouse

Jim3k
07-15-2010, 08:35 PM
And, don't mess with the Grouse when it comes to stats. He's more than just handy with understanding them. He's got 'complishments and sech. 'N 'sperience. 'N big letters after his real name. I 'spect he can give oversight to the dude's proposal with the best of them. (Nice to be the Grouse's classmate.)

dcdevil2009
07-15-2010, 10:01 PM
I'd like to see the opposite of this study: what traits do the guys that are overrated have in common. It's a bit selfish, but since Duke has the luxury of not having to take fliers on underrated guys, it would be helpful to know who in the top 100 might not pan out. Would you rather find a diamond in the rough or avoid some rough in your diamonds?

cspan37421
07-16-2010, 07:15 AM
Actually, it's a "post hoc, propter hoc" fallacy: "after which,, therefore, on account of which." It addresses spurious correlations. :rolleyes: An example of "fallacy of composition" is UNC 2010 -- nine McD All-Americans just "have to" produce a top ten team. :eek:

sagegrouse
'"Fallacy of composition: a fallacy in which what is true of a part is, on that account, alleged to be also true of the whole." -- Paul A. Samuelson, Economics, Fifth Ed., 1961.'

I agree with your quote and from Samuelson, and that is how I see it, though I believe that these distinctions can be a bit muddy at times. It's my understanding that the post hoc ergo propter hoc fallacy relates to a mistake in inferring causation from timing - namely, that if B follows A temporally, then A must have caused B. Isn't that correct?

I haven't read the original article in awhile. As I scan the thread it seems that someone went out and identified characteristics of mid-major stars, noting that they had at least 2 of 3 characteristics desired by major programs.

The fallacy then is that it it implied that if you found someone with 2 or 3 of those characteristics, they'd be a midmajor star. And it's a fallacy because not all such players become stars, but midmajor stars are a subset of such players.

Based on the description here:

http://www.nizkor.org/features/fallacies/composition.html

I still think it's the fallacy of composition, not post hoc. But one can look at these things a couple of ways. In terms of post hoc, B would be "becomes a midmajor star" and A is "has these characteristics". So if B happens after A, the fallacy is thinking A causes B. I guess the difference between the two is whether you're implying causation (such as in post hoc) or merely correlation (which would fit better with composition).

Fun to think about. I love learning about fallacies; they're everywhere, especially in politics and advertising!

sagegrouse
07-16-2010, 09:28 AM
I agree with your quote and from Samuelson, and that is how I see it, though I believe that these distinctions can be a bit muddy at times. It's my understanding that the post hoc ergo propter hoc fallacy relates to a mistake in inferring causation from timing - namely, that if B follows A temporally, then A must have caused B. Isn't that correct?

I haven't read the original article in awhile. As I scan the thread it seems that someone went out and identified characteristics of mid-major stars, noting that they had at least 2 of 3 characteristics desired by major programs.

The fallacy then is that it it implied that if you found someone with 2 or 3 of those characteristics, they'd be a midmajor star. And it's a fallacy because not all such players become stars, but midmajor stars are a subset of such players.

Based on the description here:

http://www.nizkor.org/features/fallacies/composition.html

I still think it's the fallacy of composition, not post hoc. But one can look at these things a couple of ways. In terms of post hoc, B would be "becomes a midmajor star" and A is "has these characteristics". So if B happens after A, the fallacy is thinking A causes B. I guess the difference between the two is whether you're implying causation (such as in post hoc) or merely correlation (which would fit better with composition).

Fun to think about. I love learning about fallacies; they're everywhere, especially in politics and advertising!

Dear CSPAN --

I took the extreme step and went back and read the article quoted by the OP. :rolleyes: Hmmm.... The article seems to have two points, not necessarily what author Pete Hamel intended: (1) major programs systematically overlook players that would have been productive at the highest levels of NCAA hoops; (2) it is possible, but not easy, to gain insights from looking at the attributes of these players in HS.

But truthfully, I have no idea what those "insights" are. Telep and Duke sophomore Drew Cannon did a thorough job of researching the background of players who made all league in the supposed mid-major conferences. They came up with some glimpses of truth: some excellent guards were small or slow; some forwards were too short or thin. But then what?

As a guy who spent a decade or two in social science research , I believe that Telep and Cannon made a good start but stopped short of what would be needed to make meaningful conclusions. What they were doing is looking for "market imperfections," where schools that were able to offer better coaching and competition, much more TV time and publicity, and a more luxurious setting for "busting your butt" overlooked prospects who would have been stars as high-major players.

There will be no simple theory of market imperfections, I predict. Anyway, looking at the attributes of players who who were mid-major stars is a reasonable start. This "data mining," while not scientific in any way, may suggest some patterns that may ultimately be useful.

Two other steps are needed. It is hard to see how you can draw any conclusions without looking at mid-major players who [I]WEREN'T stars. I mean, recruiting is an exercise in discriminating between "good players" and "mediocre players." How can you decide whom to pick if you only look at the attributes of good players? But then the problem is that you are dealing with low-grade ore, in that the 32 all-conference players are from a sample of well over 500 mid-major players (not to mention all the nonrecruited HS players). I suppose one could take only a random sample of the average players to make the research tractable.

The other useful approach is to essentially divide your sample in half, maybe even separate years, and test the conclusions from one half on the other. In essence, you analyze one-half of the data intensively to glean insights; then you take a complete independent set of data to see if these results hold up.

I would probably take a different approach and look intensively at a couple of categories:

(a) "Late bloomers" -- incredibly skilled players who, like the Currys, are physical immature in HS. How can you separate late bloomers from those who will remain 90-pound weaklings? Are there some physiological tests?

(b) "Physical freaks" -- amazing physcial specimens without the skills of a great basketball player. I might look for "small-motor" tests that would help tell how players could have the hand-eye coordination and reaction times required of a basketball player. FWIW, it seems that K experimented with a few of these folks back in the 1990's: Tony Moore, etc.

I did enjoy your reference on "fallacy of composition," because it had the exact example I used in ragging the UNC program about doing so poorly with so many good players:


"Every player on the team is a superstar and a great player, so the team is a great team." This is fallacious since the superstars might not be able to play together very well and hence they could be a lousy team. "

sagegrouse

cspan37421
07-16-2010, 09:47 AM
Dear CSPAN --

I took the extreme step and went back and read the article quoted by the OP. :rolleyes: Hmmm....
sagegrouse

No need to get snarky; I was on my way out the door and didn't have time to re-read it, but wanted to reply with an exploration of the distinction between the types of fallacies. Sure, I left myself open for criticism on that point, but it's not necessary to imply that laziness was the reason I didn't go back to read it before replying. That would be ... fallacious reasoning.

I'm generally in agreement with the rest of your post.

sagegrouse
07-16-2010, 10:00 AM
No need to get snarky; I was on my way out the door and didn't have time to re-read it, but wanted to reply with an exploration of the distinction between the types of fallacies. Sure, I left myself open for criticism on that point, but it's not necessary to imply that laziness was the reason I didn't go back to read it before replying. That would be ... fallacious reasoning.

I'm generally in agreement with the rest of your post.

Hi! No offense -- I was making fun of myself for not having read the article to begin with. All the best -- sagegrouse