Filed under:

# Player Aging Patterns Over Time, Part 2

Last week, in part 1, I looked at how things like peak age, the percentage of full-time players who were a certain age and how many "old" players were among the best hitters (decade by decade) has changed over time. It seemed that the number of old players has been growing while the peak age might have crept up a little. Also, the percentage of full-time players who are very young (ages 20-24) has gotten pretty low. Here in part 2, I look at how well players have been able to maintain their performance as they age. Are players getting better at keeping their performance levels up?

I basically compared "old" performance levels to "young" performance levels. "Old" meant over 35 and "young" meant under 36. To compile this data, I used the Lee Sinins Complete Baseball Encyclopedia. The data included players who had at least 4,000 plate appearances (PAs) before age 36 and at least 1,000 over the age of 35. There were 188 players in all.

The first test was to take the old offensive winning percentage (OWP) as a ratio of the young OWP. OWP is a stat created by Bill James and it is supposed to tell us what winning percentage a team would have if it had nine identical players in its lineup and gave up an average number of runs. Since I used the Sinins database, OWP is park adjusted. So if a player had a .500 OWP over age 35 and a .600 OWP before that, his ratio would be .833. The correlation between this ratio and the year a player was born was just .113. So the old-to-young ratio is rising over time, which suggests that players are better able to maintain their performance levels. But the effect is small. The graph below shows the relationship.

The vertical axis shows the ratio of the old OWP to the young OWP (so if it is above 1, it means the guy actually did better when he was older) while time or birth year is on the horizontal axis. It sure looks like there is not much of an effect. Players born more recently are only doing a slightly better job of maintaining their performance levels as they age. I also used the year of a player's debut in the major leagues instead of birth year and the correlation was just about the same.

Since OWP does not take playing time into account (a guy could have a very high old OWP and maybe just barely made 1,000 PAs), I also took a look at RCAA. Here is what Lee Sinins says about that: "RCAA--Runs created above average. This is my own creation. It's the difference between a player's RC total and the total for an average player who used the same amount of his team's outs. A negative RCAA indicates a below average player in this category." It is also park adjusted. But I converted it into "runs above replacement level" or RAR. A player can have a negative RCAA but still have value by being better than the potential replacement. For example, a player's young RCAA might have been 200 while his old could be -50. And since I am doing a ratio of old to young performance level here, it would not make sense to have negatives.

I divided each player's PAs (both old and young) by 700 (a full season of PAs). Then multiplied that by 20, with the idea that an RCAA of -20 was the replacement level (for example, if a player had 2,100 old PAs, that is three seasons and therefore 60 runs-if his old RCAA was -30, his RAR would be +30). The same thing was done for his younger years. If a guy had +60 young RAR and +30 old RAR, his "old%" would be 33.33% (30/90 since his career total RAR was 90). Even with this adjustment, some guys still had a negative score (they were not necessarily worse than replacement level players overall because they might have been good fielders-I am just measuring hitting here). Anyone who was still negative was taken out. That left 168 players. The correlation between birth year and his old RAR as a percentage of the total was -.09. That means players born later are getting a slightly lower % of their career value when they are older than the guys born earlier. I also tried -30 RCAA as the replacement level. In that case the correlation was -.137.

I expected the correlations to be higher and positive in all cases. But since they are low here, it is possible that players are not aging any better than they used to. That is, we can't expect an old player to stay closer to his younger performance levels better than players did so 50 or 100 years ago.

If players are not maintaining there performance any better as they age than they used to, there may be more guys who reach 1,000 PAs at the older age. The numbers below list how many players born in the decade starting with the year given that had 4,000 young PAs and 1,000 young PAs.

1850    5
1860    11
1870    9
1880    10
1890    13
1900    11
1910    10
1920    8
1930    17
1940    29
1950    29
1960    36

Certainly more players are staying in the game as they age. But the recent increase is probably partly due to there being so many more teams and games than there used to be.