Degree of Variance

[SmedIndy]

Intuitively, I think the variance SHOULD be narrowing given the increase in population, especially compared to the early days of baseball and before integration.

[poorme]

Originally posted by SmedIndy
Intuitively, I think the variance SHOULD be narrowing given the increase in population, especially compared to the early days of baseball and before integration.

aha. but it isn't. why not? good question. more teams? bigger rosters? what is going on?

[SmedIndy]

Originally posted by poorme


aha. but it isn't. why not? good question. more teams? bigger rosters? what is going on?

Are you looking at a long-term trend, or a one to three year anomaly?

There's 125 years + of data to sort through. We may be just getting a weird spike or something.

Or something may be going on....

[pwdennis]

I think you are getting a weird spike - looking at stats from before 1920, you get some huge disparities between the batting champion and the league average. I haven't done any analysis of the matter, though

[poorme]

Originally posted by pwdennis
I think you are getting a weird spike - looking at stats from before 1920, you get some huge disparities between the batting champion and the league average. I haven't done any analysis of the matter, though

well, the work I did was meant to compare the mid 60s to 2001. I'm pretty confident that if you chose any year around that same time frame, you'll get the same results. I don't know what would happen if you looked at 1919. Maybe I'll run the numbers if I get a chance in the next few days. I'm probably going to get fired anyway, so that will free up some of my time...

[poorme]

Taken from Lee's newsletter today:

1) Pirates LF Brian Giles returned to the lineup, after missing 3 games with a strained calf.

Giles is closing in on 2500 career PA with the Pirates (he has 2488). When he reaches that mark, the .240 gap between him and his league average will set the team record--



code:-------------------------------------------------------------------------------- DIFF PLAYER LEAGUE
1 Brian Giles .240 1.025 .785
2 Ralph Kiner .229 .971 .742
3 Honus Wagner .189 .862 .674
4 Willie Stargell .175 .889 .713
5 Barry Bonds .165 .883 .718
6 Arky Vaughan .152 .887 .735
7 Paul Waner .142 .896 .754
8 Fred Clarke .128 .797 .669
9 Dave Parker .126 .848 .721
10 Elmer Smith .125 .881 .756 --------------------------------------------------------------------------------


For purposes of this list, Giles does benefit from the fact that, as a 31 year old, he doesn't have a decline period to bring his figure down. Nevertheless, when I have the sabermetric baseball encyclopedia run the same list through the age of 31, Giles still ranks 2nd--


code:-------------------------------------------------------------------------------- DIFF PLAYER LEAGUE
1 Honus Wagner .248 .930 .682
2 Brian Giles .240 1.025 .785
3 Ralph Kiner .229 .971 .742
4 Paul Waner .170 .932 .762
5 Barry Bonds .165 .883 .718
6 Willie Stargell .157 .866 .710
7 Arky Vaughan .152 .887 .735
8 Dave Parker .142 .864 .721
9 Elmer Smith .125 .881 .756
10 George Grantham .124 .901 .777

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX


now, part of what I was looking into was trying to better compare players of different eras. since the run distributions of the eras are different, I don't think it is accurate to make these comparisons. yes all eras have an "average" runs created. but a player in the 99th percentile in 1964 will not have as many runs created above average as the 99th percentile player from 2001. you can't compare the numbers to show the latter is a better ballplayer than the former.

[SmedIndy]

Originally posted by poorme
Taken from Lee's newsletter today:

now, part of what I was looking into was trying to better compare players of different eras. since the run distributions of the eras are different, I don't think it is accurate to make these comparisons. yes all eras have an "average" runs created. but a player in the 99th percentile in 1964 will not have as many runs created above average as the 99th percentile player from 2001. you can't compare the numbers to show the latter is a better ballplayer than the former.

But....what is the % gap between the high and the median. I'd almost rather use the median as a baseline of players over a PA or IP limit and see what the spread is there.

[poorme]

Originally posted by SmedIndy


But....what is the % gap between the high and the median. I'd almost rather use the median as a baseline of players over a PA or IP limit and see what the spread is there.

i'm confused by what you're saying. mathematically, the median is 50%, so the 99% player would be 49% over the median.

for my analysis up above, I used starters with over 100 innings and hitters with over 350 AB's. the logic is you don't want to give equal weight to a starter with 5 starts and a 12.63 ERA.

I think a problem with your analysis lies in defining the bottom dwellars. Who are you comparing to the mean? Rey Ordonez and Jamie Navarro? Are those "replacement level" players? Are those the worst guys in the Majors?

I think the real worst players in the majors probably really are guys with only a few dozen at bats or innings pitched. These players don't need 200-400 games in the major leagues to prove they can't play at that level.

It's like managers. The elite few on the top of their game will always be a lot further from the mean than the ones on the bottom because the worst players will, necessarily, be unable to hold their jobs for long.

I believe players today are probably a little bit better - on average - than players of previous generations. However, I think the older players dominated their game to a much greater extent because baseball was structured differently. (Not because they were better players than the stars of today, necessarily.)

As has been said before...the top players of most any era would probably excel in any other era.

An example. No one has won a triple crown since 1967 despite the fact that we've had three men in a three year period break the all-time single season home run record. Despite the fact that Todd Helton and Larry Walker play in Coors Field. Despite .300-30-100 seasons by half the all-stars in the league practically. Some of it certainly is due to the fact that there are more players competing for the honor, but the greats of today vary from the average player of today to a lesser extent than in the past.

Run your numbers again for the entire history of the majors and see what patterns develop. Taking the 1960s out of context and comparing it to today is less meaningful unless you know where it stands in context to the decades around it.

[poorme]

thanks for the input bigklu. you have a good point about the bottom dwellers - that's why I looked at only everday player types -- bench guys were totally eliminated from the data set.

funny, i think today's stars dominate more than those in yesteryear - how else do you explain the bonds/sosa/mcgwire/pedro/schilling type stats?

I suspect that the lower strata of pitching is worse today than it was --- or maybe the steriod effect has really thrown things out of kilter.

I picked the 60s because it seemed obvious there were some different dynamics going on. Maybe in the coming weeks I'll have the time to look at it more.

[SmedIndy]

Originally posted by poorme


i'm confused by what you're saying. mathematically, the median is 50%, so the 99% player would be 49% over the median.

for my analysis up above, I used starters with over 100 innings and hitters with over 350 AB's. the logic is you don't want to give equal weight to a starter with 5 starts and a 12.63 ERA.

actually, the median is the midpoint.

What I meant was how far above the median the 99% percentile was in terms of the data. So if the median was 5 RCAA or 800 OPS, where did the 99% fall in? The 80% percentile? etc.

[poorme]

Originally posted by SmedIndy

What I meant was how far above the median the 99% percentile was in terms of the data. So if the median was 5 RCAA or 800 OPS, where did the 99% fall in? The 80% percentile? etc.

aha. let me see if I remember how to find that...

[Ytown Tribe fan]

The guy was in 674 games and pitched over 1040 innings, and how many wild pitches do you suppose he threw in all that time? He threw four. Four wild pitches in 1043 innings.

He walked 162 guys, and half of those were intentional.

Dan Quisenberry was the greatest control pitcher in major league history and the best closer in baseball for six years, and was a great guy.

Bill James has a really nice writeup of Quiz in the new HBA. A lot of his poems have been published in Aethlon and are good reading, funny, deep and relevant.

Quiz was a great pitcher for a fine team. The only guy who ever owned him was Al Oliver. Even by relief standards, his ERA was great -- he has one of the top ERA+'s of all time, behind only Pedro and Lefty Grove, and tied with Walter Johnson and Roger Clemens and Smokey Joe Wood and Hoyt Wilhelm.

If he had just pitched a little longer and more innings, he'd be in the HoF.

[SmedIndy]

That may take care of outliers at both ends, which is what you want, I think.

[Skip]

Originally posted by SmedIndy
actually, the median is the midpoint.

What I meant was how far above the median the 99% percentile was in terms of the data. So if the median was 5 RCAA or 800 OPS, where did the 99% fall in? The 80% percentile? etc.

Yes. Good Smed, good good Smed . The median is a better measure of central-ness in cases like this than the mean.