Do players jump from team to team more often now, because of free agency, than they did before free agency?
In early May I read somewhere a debate on this issue in which some of my old research was sited. X and Y were debating the issue; X said that free agency has led to players changing teams more often than they used to, and Y responded that Bill James showed that this was untrue. Uh oh.
Well, yes, I did study that issue, but it’s been a while. It’s been fifteen years or more since I did research on the issue, and, since those kind of studies can only look backward in time, that pushes it back 25 years at least. The free agency era is less than 35 years old. The research is pretty dated, and. . how good was it to begin with?
As an aside, we do hear media and fan comment now that the players don’t stay with the same team now the way they did years ago—but then, I very clearly remember hearing the same comment several times 45 years ago. As we age, our perception of time changes, so that things around us seem to me moving more rapidly. When you were in the fourth grade, the period between the start of the fall semester and Christmas seemed like an eternity. At my age, it seems like a weekend. Our minds proportion time compared to the span of our memories, so that six years to a 60-year-old seems about the same as one year to a 10-year-old. This creates the illusion, as we age, that the world is losing permanence.
I decided that I should do some good, thorough research about this subject. The way that I studied this issue before was to pick a moment as a starting point, and then look at how many players remained where they were x years later. In other words, let’s take the regulars of 1970; 24 times 8 is 192 regulars. How many of those 192 regulars were still with the same team one year later? Two years later? Three years later? What is the “decay rate”? How does the decay rate from 1970 compare to that from 1960, or 1950?
That isn’t a bad way to study the issue, but there are some problems with it. It focuses only on regulars (and regular pitchers), ignoring part-time players and ignoring players who may have been regulars in other seasons but weren’t in the base season. It produces results so close to 100%, for the first two years, that differences may not be meaningful or apparent, and so close to zero after eight or ten years that one has the same problem. It is subject to fluctuation due to changes in the game like expansion (which impacted the data for the 1960s), World War II (which impacted the data for the 1940s) and even the DH Rule (the 1970s). It is hard to find a starting point from which to measure the decay rate.
There must be a better way to measure it.
OK, here’s what I came up with. Suppose that we look at all players in history who have played 1,000 games, and we ask this question: When this player played in his 1,000th major league game, how many major league teams had he played for? What is the figure for players who played their 1,000th game in the 1920s, the 1930s, the 1940s? What about the 1,500th game, or the 2,000th?
That’s a lot more work, but it is a better way to study the issue. If players change teams more rapidly than they did in the past, then, by the time they have played 1,000 games, they have to have played for more teams, right? I don’t see how the study can fail. I started with a list of all players in history who have played 1,000 games, 1,500 games or 2,000 games, and the year in which the player passed that marker. By my count there are, through 2008:
217 players who have played 2,000 games,
606 players who have played 1,500 games, and
1,378 players who have played 1,000 games.
Then I looked up each of those players in the Encyclopedia, manually, and counted how many teams he had played for at the moment of his 1,000th, 1,500th and 2,000th games.
I wish I had studied computer programming.
Anyway, let’s use 1,000 games as our “base number” here, since there are far more players at that level, and thus we get a truer read more rapidly.
The first player to get to 1,000 games played in his career, I was surprised to learn, was not Cap Anson. It was actually a fairly obscure player named John Morrill. I started the counts in 1876. Morrill, Anson and Paul Hines played almost the same number of games every year from 1876 on, and were neck-and-neck in terms of career totals. However, because of an injury to Anson in 1879, Morrill was a little bit ahead, and he was the first player to get to 1,000 games in his career.
Barely. Morrill, Anson, Hines, Jim O’Rouke and Ezra Sutton all crossed the 1,000-game barrier in 1887. We’ll consider the first “decade” here to be the 19th century, since there were no players who played in 1,000 games prior to 1887. . our first “decade” is a little over a decade, 1887-1899.
By the end of the 19th century 93 players had played in 1,000 games in their careers. Of those 93 players:
8 had played for only one team at the time of their 1,000th game,
16 had played for two teams,
24 had played for three teams,
18 had played for four teams,
15 had played for five teams,
4 had played for six teams,
5 had played for seven teams,
1 had played for eight teams,
1 had played for nine teams, and
1, a middle infielder named Pop Smith, had played for ten different teams.
Pop Smith’s 10 teams in 1,000 games remains the major league record—in fact, until the last few years there had not been another player reaching nine. Greg Myers and John Mabry, in recent years, played for nine teams in their first 1,000 games, and someone else did in the 19th century, but no one did that in the 20th century.
I did not count playing for another organization in the minor leagues. Jeff Bagwell is a one-team player (the Astros) even though he was drafted and signed by a different team. Also, I was counting the number of team changes, not the number of teams involved, so if a player left a team and then went back to them, it counts as another team. Harold Baines went from the White Sox to the Rangers to the A’s to the Orioles to the White Sox to the Orioles to the Indians to the Orioles to the White Sox; that’s nine teams, although there are only five franchises involved. He played for the White Sox and the Orioles three times each.
Also, if a player stayed with a franchise when the franchise moved. . ..no team change. One can see it either way. We are looking at the issue of stability, as perceived by the fans. If Hank Aaron plays for one franchise in three different cities, is that perceived by the fans as his being with the same team all those years, or not? I decided to count it as all the same team, but certainly there is another kind of “location instability” that could be measured there.
I did not include pitchers in the study, even if they played 1,000 games in their careers.
Anyway, the 93 players from the 19th century had played for an average of 3.71 teams at the time of their 1,000th major league game. That’s 270 games per team (1000 divided by 3.71). This is a chart summary of the 19th century data:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1876-1899
|
93
|
8
|
9%
|
3.71
|
270
|
Now let’s compare that to the data from the three following decades:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1876-1899
|
93
|
8
|
9%
|
3.71
|
270
|
1900-1909
|
56
|
5
|
9%
|
3.48
|
287
|
1910-1919
|
86
|
35
|
41%
|
2.26
|
442
|
1920-1929
|
86
|
27
|
31%
|
2.53
|
395
|
The 1920s were not quite as stable as the 1910s, but in general the trend line here is toward more roster stability—toward players staying longer with their first team.
We romanticize one-team players. “In my day,” says Old Joe Blowhard, “if you were any good, you stayed with the team that signed you. You lived in that city. You married a girl from that city, raised your kids there, bought a business there and stayed there after your playing career. Those people knew you, and you became a part of that community. That was the way it was.” Of course, that was never exactly the way it was; neither Babe Ruth nor Ty Cobb nor Honus Wagner nor Cy Young nor Willie Mays nor Hank Aaron was a one-team player. But my idea was that, while one-team players are so rare that it is difficult to measure the frequency of them reliably, we could get at the same issue by looking at the precursors to that. To play his entire career with one team, the player must be on his first team when he plays his 1,000th game. He must be there when he plays his 1,500th game, and his 2,000th. When the number of precursors increases, we can assume the number of one-team players in increasing, and vice versa.
Adding now the 1930s and 1940s:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1876-1899
|
93
|
8
|
9%
|
3.71
|
270
|
1900-1909
|
56
|
5
|
9%
|
3.48
|
287
|
1910-1919
|
86
|
35
|
41%
|
2.26
|
442
|
1920-1929
|
86
|
27
|
31%
|
2.53
|
395
|
1930-1939
|
85
|
30
|
35%
|
2.14
|
467
|
1940-1949
|
70
|
29
|
41%
|
2.23
|
448
|
The length of time players stayed with a team took a great leap forward about 1910, and then flattened out. Let’s add the 1950s and 1960s:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1876-1899
|
93
|
8
|
9%
|
3.71
|
270
|
1900-1909
|
56
|
5
|
9%
|
3.48
|
287
|
1910-1919
|
86
|
35
|
41%
|
2.26
|
442
|
1920-1929
|
86
|
27
|
31%
|
2.53
|
395
|
1930-1939
|
85
|
30
|
35%
|
2.14
|
467
|
1940-1949
|
70
|
29
|
41%
|
2.23
|
448
|
1950-1959
|
104
|
42
|
40%
|
2.29
|
437
|
1960-1969
|
116
|
42
|
36%
|
2.38
|
420
|
There may have been a slight downturn in the 1960s, probably attributable to side-effects of expansion, but the data is essentially stable. We now enter the free agent era, which began in the mid-1970s:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1876-1899
|
93
|
8
|
9%
|
3.71
|
270
|
1900-1909
|
56
|
5
|
9%
|
3.48
|
287
|
1910-1919
|
86
|
35
|
41%
|
2.26
|
442
|
1920-1929
|
86
|
27
|
31%
|
2.53
|
395
|
1930-1939
|
85
|
30
|
35%
|
2.14
|
467
|
1940-1949
|
70
|
29
|
41%
|
2.23
|
448
|
1950-1959
|
104
|
42
|
40%
|
2.29
|
437
|
1960-1969
|
116
|
42
|
36%
|
2.38
|
420
|
1970-1979
|
159
|
57
|
36%
|
2.24
|
446
|
The game average for the 1970s is the third-highest of all time, a hair behind the 1940s and some small distance behind the 1930s. Studying data from the 1970s, I thus concluded—correctly, I believe—that the 1970s rosters were not more unstable than previous generations. Now the 1980s:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1970-1979
|
159
|
57
|
36%
|
2.24
|
446
|
1980-1989
|
176
|
57
|
32%
|
2.47
|
405
|
Beginning immediately with the onset of free agency, the number of games per team began to go down. This trend continued in the 1990s:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1970-1979
|
159
|
57
|
36%
|
2.24
|
446
|
1980-1989
|
176
|
57
|
32%
|
2.47
|
405
|
1990-1999
|
175
|
40
|
23%
|
2.78
|
360
|
And it has continued in the 21st century:
Decade
|
1,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1970-1979
|
159
|
57
|
36%
|
2.24
|
446
|
1980-1989
|
176
|
57
|
32%
|
2.47
|
405
|
1990-1999
|
175
|
40
|
23%
|
2.78
|
360
|
2000-2008
|
172
|
31
|
18%
|
3.18
|
314
|
That seems like fairly compelling data, but, to be on the safe side, let’s look at the data for players playing 1,500 games:
Decade
|
1,500 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1876-1899
|
20
|
3
|
15%
|
4.50
|
333
|
1900-1909
|
32
|
1
|
3%
|
4.72
|
318
|
1910-1919
|
23
|
7
|
30%
|
2.17
|
691
|
1920-1929
|
42
|
8
|
19%
|
2.60
|
577
|
1930-1939
|
36
|
14
|
39%
|
2.28
|
658
|
1940-1949
|
31
|
7
|
23%
|
2.71
|
554
|
1950-1959
|
38
|
19
|
50%
|
2.18
|
688
|
1960-1969
|
51
|
15
|
29%
|
2.45
|
612
|
1970-1979
|
64
|
25
|
39%
|
2.53
|
593
|
|
|
|
|
|
|
1980-1989
|
99
|
17
|
17%
|
2.99
|
502
|
1990-1999
|
76
|
15
|
20%
|
3.07
|
489
|
2000-2008
|
94
|
11
|
12%
|
3.85
|
390
|
One might ask why the Games-per-team average is so much higher here than it was before. The reason is that those players who play 1,500 games in their careers are much, much better players than those whose careers end in the 1,000-1,499 game range. As they are better players, their teams are more inclined to keep them, and so the average games per team is much higher.
In any case, here again we see that there has been a quite dramatic decrease in the length of time that players stay with a team, beginning with the onset of the free agent era. The percentage of players who were still with their first major league team at the 1500-game mark has dropped, over the last three decades, from 39% to 12%--the lowest it has been in 100 years.
The numbers of 2000-game players are so small that, looked at by decade, we would have unstable numbers. Let’s group them into three-decade packages:
Time Period
|
2,000 game Players
|
One-Team Players
|
One-Team Percentage
|
Average Teams
|
Games Per Team
|
1876-1919
|
18
|
2
|
11%
|
3.83
|
522
|
1920-1949
|
34
|
10
|
29%
|
2.62
|
763
|
1950-1979
|
50
|
20
|
40%
|
2.14
|
935
|
1980-2008
|
115
|
21
|
18%
|
3.37
|
593
|
First of all, we can see here that there has been a stunning increase, in the last 30 years, in the number of players have long careers. But the central data is consistent with the data from the 1000-game and 1500-game studies.
In short, there is no question whatsoever that the rate at which players move from team to team has in fact increased, and increased quite dramatically, during the free agent era. The data could not be any more clear or any more definitive.
The length of time that a moderately talented player typically spends with one team has decreased, since free agency began, from 446 games to 314 games.
The number of games that a more talented player typically spends with one team has decreased from 593 to 390.
The number of games that the most talented player typically spends with one team has decreased, over the last 30 years, from 935 games to 593.
Another new article tomorrow: Ellis, Richie and the Duke.