Just to add, we see this kind of effect everytime there is a drastic change in playing personnel. The same type of thing happens when the DH was introduced. Similar, but not to the same degree, the same thing happens in every expansion year.

But like I said, half a percent here, half a percent there, and change same of my assumptions to something equally reasonable (like the age prime group I've mentioned) and when you start compounding everything, a doubling of talent from 1920s to 1990s becomes only a 30% increase. And from 1946 onwards, there is no change in talent. In this case, Ruth remains one of the greatest hitters ever. Though Ted Williams might edge him out.

I accidentally submitted the same message twice, can't delete the first one!



[edited by Hoiles on February 8th, 2001 at 01:52 AM]

Tangotiger, while your efforts here are certainly admirable, I don't exactly know what is supposed to be concluded from this study. Are we supposed to assume that a mid-20's Babe Ruth, with identical background, training, etc., would have a hard time sticking in the majors now?

Also, there should be better comparison years than 45-46. In between these years, the regular players (like Ted Williams, Joe DiMaggio, and Bob Feller) came back to the NL after the "replacement" players of the war. Surely the league was of higher calibre in 1946, which meant that players who were in the majors in '45 would have likely performed worse in '46.

"Surely the league was of higher calibre in 1946, which meant that players who were in the majors in '45 would have likely performed worse in '46. "

THIS IS EXACTLY WHAT I SAID. I said that those guys who played in BOTH 45/46 had their stats drop by 5%. This is now at least the 3rd time I've said this.

As for better years than 45/46, I did it for EVERY SINGLE PAIR OF YEARS from 1919 to 1999. 45/46 just happens to be the best example. I then chain all these comparisons.

For example, if a player has a LWR of .640 in 1945 and .610 in 1946, and then we take a player from 1946 with a LWR of .640, and in 1947 he had a .660, AND IF WE TAKE A LARGE ENOUGH, and good enough, SAMPLE, then we can conclude that the difference between 1945 and 19FORTY-SEVEN is: .640 (1945) = .610 (1946), .640 (1946) = .650 (1947), therefore .640 (1945) = about .620 (1947).

As for what we are supposed to conclude:
1 - we have an excellent methodology to determine the change in talent and playing conditions for the whole history of baseball
2 - if we can control as many variables as possible, our accuracy will go from somewhere below passable to somewhere above very good
3 - the largest variable to control is the pool of players (i am in the middle of refining this); if we lower the prime age class from 28 to 26, then the methodology will show that BABE RUTH REMAINS ONE OF THE GREATEST HITTERS EVER. i am now trying to establish exactly what is the prime age class. stay tuned.

I think it's basically Impossible to quantify the relative quality of an era's athletes vs. that of another one's. How would you do it? I admire yer effort of course-but it doesn't quite work for me.

Did I read it right-the level of play increases by 1% a year, more or less? Does that mean that a guy like, say Harold Baines is playing in a league where they average player is 20% better than when he broke in in 1980? Does that mean Baines is 20% better than he was in '80 or what? I am sure that isn't what yer getting at, but that is the problem I have with it. Last I checked in '99-Baines didn't do that well last year of course-he knocked in over 100 and hit .312 or so. Somehow or other I don't think he's overmatched, even pushing 40.

Certainly I agree that the average player is better than they were back in '80 or '55 or '25, we would sure Hope so-but I wonder if it's not something that would affect the greatest players all that much, of which Ruth most certainly is one.

My fave of all time is Yaz-and while he certainly hit the weights and etc. worked his butt off year in, year out-he was but 5-11" 185 lbs., more or less-and that is what-Paul Moliter or Gary Sheffield in size? Not even looking Bill Dickey, Cobb or Ruth or Gehrig in the eyes quite.

Do you think, therefore, that a guy from '61(Yaz being one) would have trouble playing today? Logic would make you say 'Yeah', but when you remember that Moliter, who faced Yaz for 6 years, wound up leading the AL in hits in '96-....then the wheels start to fall off the wagon.

At least for me.

But keep up the work-it does make for interesting reading, and fodder for fights too.
__________________
"Manny, Castillo and Hideo-Makes ya feel Good, doesn't it?!"

I've established the peak age at somewhere slighty above 26.5 years of age (I calculate age to the day, as of July 1. It doesn't really matter how you calculate age, as long as you calculate it the same way for everyone). This was done by comparing each of the pairs of years, adding up the changes. The 26 to 27 age class was the most stable.

Re-running the methodology using the prime age class as 24 to 29, here are the results:
- the NL talent level has remained VERY constant. Over the last 70 years, the talent level has increased by 10% total.
- the AL talent level was poor relative to the NL up until 1946. After that point, the talent level of both league increased by about the same level. Specifically, the talent level of the AL 1945 was about 70% of what it is today. By 1949, it was about 87% of today's level. The NL, on the other hand, had a talent level of 82% of what it is today in 1945. The year after, 1946, and continuing onwards, it jumped to 90%, which is where it was from 1928 to 1943.

If anyone knows what happened in the AL RELATIVE TO THE NL right after the war, I'd be curious to know.

In any case, working the numbers for 1927 AL, and we see that all the offensive numbers RELATIVE TO THE OUTS must be multiplied by about two-thirds. This is what it gives us for Babe (numbers in parenthesis are his real-life numbers)
AB - 568 (476)
H - 153 (192)
HR - 48 (60)
BB - 110 (138)
SLG - 584 (772)
OBA - 387 (487)
This is what Babe would have hit if plucked from 1927, and magically brought to today.

There is one additional way that I can increase my pool of players, but it would require an additional step. (I believe Voros does this as well.) If, for example, players in the 22.5 to 23.5 age class, as a group, perform at 93% of their peak level, then what we COULD do is to increase all their stats by about 7% to bring them into their peak class. We can do this for every age class, and would allow us to sustantially increase our player pool. This would force us to accept however that we know what the aging pattern is at every age group. I can estimate pretty well what those numbers are, but it would introduce another variable. And as we've seen, if we are off by half a percent in our estimation of the aging pattern, then compounding those errors over time will give us some potentially large error.

For the moment, I will stick with the 24-29 age class.

T-squared:

I am wondering if the probelms may not be in your math. While the ability to "hit" a baseball is probably normally distributed in the general population, the "position players" which are about 15 per team amount to 15*16=240 in 1927 and 15*30=450 in 1999. These 240 and 450 players represent the extreme right end of the normal curve. If you isolate on them as "the population", you will end up with a "log normal" distribution (also called the Pareto distribution). In the log normal distribution, the Mean will be significantly higher than the Mode, being dragged up there by the extreme values at the far right and not having extreme values at the fat left to balance this as in the normal distribution. This is evidenced by the fact(as earlier pointed out, that you have a league mean BA of say .260 and guys hitting .370, but nobody much below .210. This will be true for most of the measures of "hitting effectiveness (BA, OBP, SLG, OPS, HR/PA, BB/PA, etc). Rerun your data and assume a log normal distribution and I would bet your numbers arent showing as much improvement.

quote:

---

originally posted by tangotiger
AB - 568 (476)
H - 153 (192)
HR - 48 (60)
BB - 110 (138)
SLG - 584 (772)
OBA - 387 (487)
This is what Babe would have hit if plucked from 1927, and magically brought to today.

---

You both bring up interesting points.

Last time I touched a stats book was in 1990. Please give me step-by-step procedures.

As for breaking players into classes, I was thinking about that as well. My problem is that my pool of players is already pretty shrunken by using players only in the 24 to 29 age class. Breaking those guys into a high-end, middle, and low-end group might add some volatility. Before I do that, I am thinking of introducing an age adjustment factor, as I alluded to earlier. By doing that, my population will more than double. Then, I'll be in a position of separating into classes of players.

I did do this once a looooooooooong time ago, and Edgar Martinez came out as a BETTER hitter than Junior. What happens is that Edgar started late, and he is hitting far better for his age than the typical hitter does. The aging pattern for Edgar is not apparent, and so by applying an age factor, we are increasing his peak talent level at more than his actual peak talent. (I hope that made sense. Basically, he holds onto to his talent more than the average player, even the average superstar, and so I'm not even sure if it makes sense to apply an aging pattern to Edgar that would be the same to say Reggie Jackson.)

admin options:
>open / close topic
>move thread
>delete topic
>edit topic