This is a very short study of an issue that arose in the context of the Barry Bonds discussion. I was wondering what the relevant aging rate for Barry Bonds is.
I decided to measure the declines in value in terms of Average Season Score, mostly because that’s what I can do with the data I have organized the way that I have it.
Bonds has had Season Scores the last three years of 38, 305 and 287, creating an Established Performance Level of 267 (Established Performance Level being here defined as .60 times the most recent season score, plus .30 times the previous season, plus .10 times the season before that.) I took all players in history (non pitchers) who had Established Performance Levels between 200 and 335, intending to create a group with an average performance level near Bonds.
I miscalculated a little bit. . .obvious reason. Since there are many more players in the bottom of that range than in the top, the average Established Performance Level of the players in the group was not 267, as I had intended, but 252. Anyway, there were 3,757 players in the group, whose average performance in their most recent season was this:
G | AB | R | H | 2B | 3B | HR | RBI | BB | SO | SB | Avg | OBA | Slg | OPS |
140 | 531 | 89 | 159 | 29 | 6 | 17 | 84 | 60 | 62 | 14 | .299 | .374 | .474 | .848 |
I then sorted those players by age, and looked at their losses in Average Season Score over the next two seasons. For example, the 27-year-old players in the group (of whom there were 380) had an Average Established Performance Level of 249. The next season they averaged 211 (down 15%), and the season after that 197 (down 21%). These numbers are entered in the chart below as –15 and –21.
Age | +1 | +2 | | | Age | +1 | +2 |
20 | +26 | +50 | | | 31 | -22 | -35 |
21 | +32 | +32 | | | 32 | -24 | -36 |
22 | +9 | +12 | | | 33 | -23 | -37 |
23 | +6 | +4 | | | 34 | -24 | -42 |
24 | -4 | -3 | | | 35 | -29 | -45 |
25 | -2 | -6 | | | 36 | -36 | -58 |
26 | -9 | -17 | | | 37 | -42 | -58 |
27 | -15 | -21 | | | 38 | -32 | -56 |
28 | -16 | -23 | | | 39 | -40 | -68 |
29 | -15 | -23 | | | 40 | -43 | -74 |
30 | -17 | -29 | | | Over 40 | -36 | -76 |
Players over 40 who performed in this range typically lost 76% of their value over the next two seasons. Much of this evaporation was caused by the retirement of many of the players. However, players over age 38 who didn’t retire almost uniformly took a tremendous step backward in performance over the next two seasons.
It is also interesting to note that the older players in this study resembled Bonds in that they tended to have higher walks and power, but less playing time than younger players. The 27-year-old players in this study had an on-base percentage of .359, slugging percentage of .452, OPS of .810, but averaged 142 games played, 541 at bats. The 40+ players averaged 120 games, 403 at bats, .424 on base percentage, .496 slugging, .920 OPS. Of course, the 40+ players in the study were very small in number, but the 38- to 40-year olds, who were more numerous, showed the same pattern.
Of course, we then get into the issue of what is “value”? It may be that the players I am comparing are of comparable Season Scores, but not truly of comparable value. The older players in the group, one could argue, had less playing time but more “value” because they were much better hitters.
Could be. It could also be that they merely had a different shape to their value, which I couldn’t really measure because the data base I am using doesn’t include defense. Very likely. . well, certainly. . .the older players in the study had much less defensive value than the younger players. But it may well be that their marginal value as hitters was much higher.
I then focused on the issue of players who had “old players skills” vs. “young players skills”. For this portion of the study I eliminated the players who had played before 1920, reducing the total number of players in the study from 3,757 to 3,059. I sorted those players based on what I called “old bases” and “young bases”. Young bases were defined as
2 * doubles, plus
5 * triples, plus
3 * stolen bases
and “old bases” were defined as
3 * home runs, plus walks.
The weights were set so that old bases and young bases more or less balanced for the group as a whole. The five “youngest” players in the group were:
Player | YEAR | G | AB | R | H | 2B | 3B | HR | RBI | BB | SO | SB | CS | Avg | AGE | OBA | SPct | OPS |
Willie Wilson | 1980 | 161 | 705 | 133 | 230 | 28 | 15 | 3 | 49 | 28 | 81 | 79 | 10 | .326 | 24 | .357 | .421 | .778 |
Edd Roush | 1924 | 121 | 483 | 67 | 168 | 23 | 21 | 3 | 72 | 22 | 11 | 17 | 13 | .348 | 31 | .376 | .501 | .877 |
Lou Brock | 1974 | 153 | 635 | 105 | 194 | 25 | 7 | 3 | 48 | 61 | 88 | 118 | 33 | .306 | 35 | .368 | .381 | .749 |
Sam Rice | 1920 | 153 | 624 | 83 | 211 | 29 | 9 | 3 | 80 | 39 | 23 | 63 | 30 | .338 | 30 | .381 | .428 | .809 |
Lou Brock | 1968 | 159 | 660 | 92 | 184 | 46 | 14 | 6 | 51 | 46 | 124 | 62 | 12 | .279 | 29 | .328 | .418 | .746 |
While the six “oldest” players were:
Player | YEAR | G | AB | R | H | 2B | 3B | HR | RBI | BB | SO | SB | CS | Avg | AGE | OBA | SPct | OPS |
Darrell Evans | 1985 | 151 | 505 | 81 | 125 | 17 | 0 | 40 | 94 | 85 | 85 | 0 | 4 | .248 | 38 | .356 | .519 | .875 |
Hank Aaron | 1972 | 129 | 449 | 75 | 119 | 10 | 0 | 34 | 77 | 92 | 55 | 4 | 0 | .265 | 38 | .390 | .514 | .904 |
Mark McGwire | 1995 | 104 | 317 | 75 | 87 | 13 | 0 | 39 | 90 | 88 | 77 | 1 | 1 | .274 | 31 | .441 | .685 | 1.125 |
Jason Giambi | 2005 | 139 | 417 | 74 | 113 | 14 | 0 | 32 | 87 | 108 | 109 | 0 | 0 | .271 | 34 | .440 | .535 | .975 |
Harmon Killebrew | 1964 | 158 | 577 | 95 | 156 | 11 | 1 | 49 | 111 | 93 | 135 | 0 | 0 | .270 | 28 | .377 | .548 | .924 |
Mark McGwire | 2001 | 97 | 299 | 48 | 56 | 4 | 0 | 29 | 64 | 56 | 118 | 0 | 0 | .187 | 37 | .316 | .492 | .808 |
I was trying to form a focus group of players who were like Bonds, in that all they really did was walk and hit homers. I then divided the players into ten groups, based on their “young bases” vs. their “old bases”. These are the performance norms for the ten groups:
Player | G | AB | R | H | 2B | 3B | HR | RBI | BB | SO | SB | CS | Avg | OBA | SPct | OPS |
Youngest | 143 | 569 | 95 | 181 | 33 | 11 | 8 | 74 | 47 | 45 | 27 | 9 | .319 | .374 | .457 | .831 |
2nd Youngest | 145 | 566 | 92 | 176 | 33 | 8 | 11 | 78 | 55 | 53 | 18 | 6 | .312 | .375 | .459 | .834 |
3rd Group | 143 | 551 | 91 | 168 | 33 | 7 | 14 | 81 | 57 | 58 | 14 | 6 | .305 | .373 | .469 | .843 |
4th Group | 143 | 548 | 90 | 164 | 32 | 6 | 17 | 84 | 58 | 68 | 13 | 5 | .299 | .369 | .473 | .842 |
5th Group | 144 | 545 | 87 | 163 | 32 | 5 | 19 | 87 | 61 | 71 | 9 | 4 | .299 | .370 | .482 | .852 |
6th Group | 145 | 541 | 86 | 159 | 30 | 4 | 22 | 90 | 63 | 77 | 8 | 4 | .293 | .368 | .486 | .854 |
7th Group | 143 | 530 | 84 | 154 | 30 | 3 | 23 | 90 | 64 | 81 | 6 | 3 | .291 | .370 | .492 | .862 |
3rd Oldest | 142 | 519 | 83 | 148 | 27 | 3 | 25 | 92 | 67 | 84 | 5 | 3 | .285 | .369 | .493 | .861 |
2nd Oldest | 144 | 521 | 84 | 147 | 26 | 2 | 28 | 94 | 74 | 90 | 4 | 3 | .282 | .373 | .502 | .875 |
Oldest Group | 139 | 486 | 80 | 133 | 20 | 1 | 31 | 93 | 80 | 93 | 2 | 2 | .274 | .378 | .511 | .890 |
And these are the value retention rates over the next two years:
Group | E Sc | Sc +1 | Sc +2 | |
Youngest | 250 | 226 | 206 | 90% | 82% |
2nd Youngest | 246 | 210 | 191 | 85% | 78% |
3rd Group | 252 | 219 | 191 | 87% | 76% |
4th Group | 249 | 217 | 197 | 87% | 79% |
5th Group | 251 | 215 | 192 | 86% | 77% |
6th Group | 252 | 214 | 203 | 85% | 80% |
7th Group | 253 | 200 | 186 | 79% | 74% |
3rd Oldest | 250 | 204 | 183 | 82% | 73% |
2nd Oldest | 260 | 217 | 180 | 84% | 69% |
Oldest Group | 252 | 203 | 179 | 80% | 71% |
The players with “young skills” tended to retain somewhat more of their value over the next two seasons, although it is unclear to what extent they did this because of their skills, and to what extent they retained more of their value because they were actually younger. The player with the “youngest” skills—about 300 of them—averaged 28.3 years of age, whereas the players with the oldest skills averaged 30.5.
The “oldest group” in the study above, however, still had much “younger” skills than Bonds. Bonds’ “skill profile” is in the oldest 10% of the oldest 10%--the oldest 1%. There are 30 players in this percentile, whose average performance is this:
G | AB | R | H | 2B | 3B | HR | RBI | BB | SO | SB | CS | Avg | OBA | SPct | OPS |
134 | 441 | 77 | 116 | 15 | 0 | 35 | 91 | 90 | 100 | 1 | 1 | .263 | .392 | .538 | .930 |
These players retained, on average, 67% of their value in the following season, and 54% in the season following that. Bonds’, of course, is even more extreme than this 1% “most extreme”.
Bonds is such an extreme and atypical player that it is difficult to model or predict his career based on anyone else. It is difficult to truly set aside the issues of Bonds being a troubled and troublesome person. It remains my belief, however, that
a) as a player ages, his skills tend to build, but narrow,
b) the last skills remaining tend to be power and control of the strike zone,
c) when a player loses his speed, when he loses everything except his power and his control of the strike zone, there generally isn’t very much left of his career.