Page 4 of 5

PostPosted: Tue Mar 16, 2004 10:53 pm
by reiser
wkelly91 wrote:I know this is a small sample but.......Marcus Giles .349 2nd half 8-o


I agree.

In defense of subjectivity here, this isn't voting patterns, M1 or the weather we are measuring, it's baseball players. quite often, and the Giles case is excellent, we have to rely on a limited sample. while this certainly contains flaws, it's all we have. take Renteria's breakout 2nd half from 2002-a good leading indicator of his 2003 campaign. Similiarly, like it or not, over the past 3 years Chavez is a .258 hitter Pre-All Star, and a .310 hitter Post-break.

more to the point, you certainly would want to weigh a player who had some sort of statistical collapse post-all star break. Where'd Baldelli's AVG go after the break? How about V. Wells power?

PostPosted: Tue Mar 16, 2004 10:59 pm
by GotowarMissAgnes
reiser wrote:
wkelly91 wrote:I know this is a small sample but.......Marcus Giles .349 2nd half 8-o


I agree.

In defense of subjectivity here, this isn't voting patterns, M1 or the weather we are measuring, it's baseball players. quite often, and the Giles case is excellent, we have to rely on a limited sample. while this certainly contains flaws, it's all we have. take Renteria's breakout 2nd half from 2002-a good leading indicator of his 2003 campaign. Similiarly, like it or not, over the past 3 years Chavez is a .258 hitter Pre-All Star, and a .310 hitter Post-break.

more to the point, you certainly would want to weigh a player who had some sort of statistical collapse post-all star break. Where'd Baldelli's AVG go after the break? How about V. Wells power?


Voting patterns, like baseball statistics are generated by humans and, with sufficient samples, are reasonably predictable.

As I pointed out, minor league stats are just as predictive as major league stats, so our data on Giles is not limited.

The question is whether or not you weigh this information. It's how much weight you put on it compared to other information. And a good rule is, the smaller the sample size you are looking at, the less weight you should put on it.

Law of Large Numbers
Regression to the Mean.

Learn 'em or get Burned by 'em.

PostPosted: Tue Mar 16, 2004 11:40 pm
by reiser
GotowarMissAgnes wrote:
There is so much random variation in player's performance that it is really dangerous to make decisions based on these data.

The question is whether or not you weigh this information. It's how much weight you put on it compared to other information. And a good rule is, the smaller the sample size you are looking at, the less weight you should put on it.
Law of Large Numbers
Regression to the Mean.

I do agree with these points, but want to emphasize that a small sample shouldn't necessarily be discarded.

oh, and Chavez is a better 2nd half hitter :-b

PostPosted: Tue Mar 16, 2004 11:44 pm
by BigLebowski
LoDuca a great first half catcher.
Beltre a pretty good second half pickup after he sucks in the first half again.

PostPosted: Wed Mar 17, 2004 12:25 am
by El Jefe
My slow starter/fast finisher is Adrian Beltre 2003 Pre-AllStar stats were: .225AVG, 6HR, 32 RBI...Post-AllStar: .257AVG, 17HR, 47RBI.
My fast start/slow finish guy is my main man Jimmy Edmonds. He seems to be a MVP candidate every season 'til July comes around. In 2003 his 1st half was .303AVG, 28HR, 67 RBI...2nd half he was .214AVG, 11HR, 22RBI.

PostPosted: Wed Mar 17, 2004 12:27 am
by El Jefe
BigLebowski wrote:Beltre a pretty good second half pickup after he sucks in the first half again.


Dude...You stole my thunder while I was looking up stats. Good call though. ;-D

PostPosted: Tue Apr 27, 2004 7:03 am
by GotowarMissAgnes
I just had to resurrect this thread so that I could hear more comments about how Jim Thome, Adrian Beltre, Bagwell, Magglio, Wade Miller, Marcus Giles are off to such slow starts, while Hillenbrand and Delgado are off to such fast starts.

Of the 21 guys mentioned in the thread (if I counted right), 8 are running counter to the predictions, 12 are having the predicted slow or fast start and one (Mike Lowell) was predicted to have both a fast and a slow start.

Hot/cold = random variation.

PostPosted: Tue Apr 27, 2004 7:45 am
by mfe
best thread on here

carlos lee for late bloomer

PostPosted: Tue Apr 27, 2004 8:53 am
by stumpak
V. Wells, in his 2 full-time years, has done considerably better in the 2nd half.

LoDuca is an example of a guy that falls apart in the 2nd half every year.

PostPosted: Tue Apr 27, 2004 1:07 pm
by GotowarMissAgnes
Hmm, let's try this again.

HOT START/COLD START = RANDOM VARIATION

If you want to predict which guys will have hot starts or cold starts you will do just as well flipping a coin as looking at their past performance.