Value Pattern Similarity

August 12, 2021
                                              Age/Value Similarity Patterns
 

            I have invented not one but two new methods to measure Age/Value Similarity Scores.  This was in reference to the Joe Morgan discussion, but we’re kind of done with that; that’s mostly just where it came from.   What we are looking for here is a way to measure the extent to which two players value patterns, as they age, are like one another.

            There is actual value in having this, I think.  I did a report for a major league team last winter.  The team was wondering whether they should re-sign a free agent who had such a poor season that he had not much of a market anywhere else, and he wanted to come back to the team for not a lot of money, but they were wondering whether he would produce anything.  Among the things I looked at in preparing the report was players his age who had had similar value patterns.  I didn’t have a METHOD to do it; I just cobbled one together at the time.   I recommended that they should keep him, in part because some similar players had had bounceback seasons, and they kept him, and he’s having a decent year. 

 

            The first Value Pattern Similarity Index that I created started by asking "At what age did this player have his best season?  At what age did this player have his second-best season?  At what age did this player have his third-best season?, etc., until you have the entire career.   Babe Ruth’s line is:

 

First

Last

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Babe

Ruth

28

26

25

29

31

32

33

24

23

35

36

21

22

37

34

 

            Which means that Babe Ruth had his career high in Win Shares in 1923, which is the year that he hit .393, his second-highest Win Shares at age 26, then ages 25, 29, 31, 32, etc.   We only do the top 15 years. 

            No one else has exactly the same pattern of their best seasons as Ruth does.   Three other players (Steve Renko, Lenny Randle and Lil Stoner) also had their best seasons at age 28, their second-best seasons at age 26, their third-best seasons at age 25, and their fourth-best seasons at age 29, as Ruth did.   One of those players, Lenny Randle, had his fifth-best season at age 31, but then the pattern breaks, and Ruth has no more perfect comps. 

            Every difference in the pattern is a "penalty"; that is, a deduction from a "perfect match" score of 1,000.   For obvious reasons, the largest penalty is for a difference in the age of the #1 season, then second-largest is for a difference in the age of the #2 season, etc. 

            That system worked OK, but it seemed weird to have a system in which Lenny Randle might have the most similar value pattern to Babe Ruth.   He doesn’t, and he couldn’t, really, for two reasons.  One is, very few players actually play 15 seasons in the majors, thus are marked down as comps for Babe Ruth because the of holes in the chart.   Second, I added a wrinkle in the system to mark comparisons down based on differences in the Career Win Shares total.  

            That system concluded that the three players with the value patterns most similar to Babe Ruth were (1) Frank Robinson, (2) Stan Musial, and (3) Wade Boggs.   The fact that Ruth is an all-time great right fielder and Robinson is an all-time great right fielder is just kind of a coincidence, except not really.  Whenever you find players with one set of similarities, you will very often find that they have other similarities that you hadn’t focused on.  A player who was born in 1943 will tend to be similar to another player in 1944 or 1945.  A very tall left-handed pitcher will show up as similar to another very tall left-handed pitcher.  It happens because those things give shape and definition to a career in ways that you’re not always aware of. 

            Anyway, by this system, the players with age/value patterns most similar to Joe Morgan are (1) Rod Carew, (2) Jim Palmer and (3) Jeff Bagwell.  Don’t ask me why.

 

First

Last

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Joe

Morgan

31

29

28

30

32

21

33

27

38

23

25

26

36

22

39

Rod

Carew

31

28

29

30

27

26

32

23

34

21

25

36

33

37

22

Jim

Palmer

29

31

27

30

32

24

26

25

36

23

33

34

20

28

35

 

            It’s one of those things; I get it, but I don’t quite get it.   Babe Ruth and Lil Stoner, really?  When you reduce the best-seasons pattern to a series of numbers like this one can sort of see that there are similarities in the patterns, but there is no intuitive logic to it.  It doesn’t quite ring true somehow.   That’s why I decided to try it again, and came up with a different system.

            Before I should move on, though, I should add this.  As a part of that system, I came up with a "Uniqueness Score" for each player’s value pattern.   The players with the MOST unique value patterns in history were (1) Barry Bonds, and (2) Monte Ward.  Monte Ward was the 19th century guy who won 47 games as a pitcher when he was 19 years old, then spent most of his career as a shortstop, which you have to admit is a pretty odd value pattern.  Barry Bonds, I’m guessing you get it.  Bonds scored at 99 on a uniqueness scale, Ward at 97.   The average for players with 400 or more career Win Shares was 43, and Joe Morgan scored at 47, for what that’s worth.

 

            The other system is more intuitive.   This is a comparison of the Win Shares of Joe Morgan and Reggie Smith, through age 35:

First

Last

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

Joe

Morgan

30

19

26

2

24

24

29

39

40

37

44

37

30

17

18

Reggie

Smith

0

19

25

24

25

29

26

23

25

20

14

29

24

9

17

 

            We start out assuming, if their Win Share patterns were identical, that the similarity score would be 1000.   Morgan at age 21 had 30 Win Shares and Reggie Smith had none, so that’s a 30-point deduction, making the similarity 970.   

            They have the same number of Win Shares at age 22 (19 Win Shares for Morgan in 1966, 19 for Smith 1967), and they are separated by only one Win Share at age 23.  That leaves them at 969.   At age 24 Morgan was injured and basically missed the season, so that’s a 22-point penalty, leaving us at 947.  Penalties of 1, 5 and 3 points the next three seasons puts us at 938. 

            Then Morgan hits his best seasons, and the gap between them grows radically larger.  The similarity drops to 890 in three seasons.  Actually, I see now that Joe Morgan had one Win Share when he was 19, which I missed when constructing this chart.  Well, anyway, I was just using Reggie Smith to illustrate the process.  Reggie ranks as the #100 age/value comp for Morgan.  The Number One comp, again, was Rod Carew.    

            Rod Carew is the number one age/value pattern comp for Morgan in both systems, and Nap Lajoie is in the top seven age/value comps for Joe Morgan, in both systems.  Again, the fact that they’re all second basemen doesn’t directly influence the process; it is just sort of a coincidence.   Here is the age/value comp for Morgan, Carew and Nap Lajoie:

 

First

Last

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

Joe

Morgan

1

0

30

19

26

2

24

24

29

39

40

37

44

37

30

17

18

Rod

Carew

0

0

19

13

21

11

17

22

28

32

30

30

37

22

16

20

12

Nap

Lajoie

0

0

5

21

26

19

22

42

22

31

41

14

33

32

32

27

47

 

 

First

Last

36

37

38

39

40

41

Career

Joe

Morgan

21

14

29

19

12

0

512

Rod

Carew

17

16

9

12

0

0

384

Nap

Lajoie

14

22

23

7

12

4

496

 

            As you can see, Nap Lajoie through age 27 had only two seasons in which he had significantly more value than Lajoie (ages 24 and 26), and Morgan had one season in which he dramatically out-performed Lajoie at the same age (age 21).  Through age 27, Morgan had 155 Win Shares, and Lajoie had 157. 

            Rogers Hornsby had 502 career win shares, by the way—ten less than Morgan, six more than Lajoie.  That’s not a significant separator among the three of them.   Eddie Collins had 574 Win Shares, which is a fairly significant separator between Collins and the other guys.            Morgan had injury seasons, Lajoie had injury seasons, Hornsby had injury seasons. . .so what?

            Morgan and Lajoie actually have very similar age/value patterns.  The gentleman that I was discussing this with seemed to be saying that Morgan couldn’t be considered the greatest second baseman of all time because he didn’t hit his prime until age 28, and all of the greats had done much more before that age.  

            Well, not really.  Three points:

1)      Every player’s age/pattern is unique,

2)     Morgan’s age/pattern is unique but actually very normal, and

3)     Actually, it is fairly common for all-time greats to have their best seasons in the same age range.   

 

There are 55 players in baseball history who earned 400 or more Win Shares, a group that contains all of the candidates for the greatest-ever-at-his-position tag except the catchers.   Of those 55 players, 14 began their run of their three greatest seasons at age 29 or later.  Morgan began his best three-year run at age 29 (1973-1974-1975.  Morgan won the MVP Award in 1976, at age 32, but was credited with more Win Shares in 1973, at age 29.  WAR likes his 1976 season slightly better, 9.6 to 9.3.) 

Anyway, 14 of the 55 all-time greats started their best run of seasons as late as Morgan or later, 12 of those 14 later than Morgan. Willie McCovey started his best three-year run (1968-1969-1970) at age 30.  Rickey Henderson’s best three-year run started at age 30, Warren Spahn at 30, Craig Biggio at 30 and Mike Schmidt at 30.  Willie Mays started his best three-year run at age 32, Honus Wagner at age 32, and Nap Lajoie at age 33. 

(Off topic, but Willie Mays numbers at ages 32-33-34 may not jump forward from his career, because the numbers in the 1960s are not as big as the numbers in the 1950s, but the three seasons 1963-1964-1965 are, by WAR, not only the best three-year run of Mays career, but the three best seasons of Mays’ career, the only seasons in which he was over 9.0 (9.1, 9.0, 9.5).   The same is true of Al Kaline; his numbers in the 1960s don’t LOOK as good as his numbers in the 1950s, but there are actually more Wins in there in the 1960s numbers than the 1950s numbers.  1955-56-57 Kaline had 16.0 WAR and 77 Win Shares; in 1965-66-67 he had 16.4 WAR and 81 Win Shares.)

Directly comparing Mays and Morgan, Mays’ three best seasons were ages 32-33-34, and he had WAR of 9.1, 9.0 and 9.5.  Morgan from ages 28 to 32 had 9.3, 9.3, 8.6, 11.0 and 9.6.  Morgan shows as both better than Mays at his best—which I don’t necessarily agree with—and as younger than Mays at the time of his best seasons.

But I realized that I was probably not really understanding the gentleman’s point.  I understood him to be saying that Morgan could not be considered an all-time great because he wasn’t all that good up to age 27.  That would be a be a mystifying and apparently meaningless argument, because what difference does it make how old he was when he was great?   It doesn’t seem to make ANY difference. 

This puzzled the hell out of me, and I worried about it and worried about it, but I think now that that wasn’t exactly what he was trying to say.  I think his main point was more properly like "Morgan couldn’t be an all-time great because he just wasn’t that good through age 27."  It makes a difference if you change which words in the sentence you emphasize. 

            But how good is "that" good?  Morgan in Houston was not as great a player as he was in Cincinnati, that’s true.   Morgan was not as great as Mays, over the course of his career, but then, no other second baseman was, either.  But Morgan in Houston was still a tremendous player.  He was as good as Reggie Smith. 

            And the argument that Houston trading away Morgan somehow proves that he wasn’t that great a player. ..well, why doesn’t that apply to Babe Ruth?  Does Boston selling Babe Ruth, AFTER he had set the major league record for home runs for the first time, somehow prove that Ruth wasn’t that great?  The Phillies did not want to pay Nap Lajoie enough money to keep him.  Does this somehow prove that Lajoie was not great?  The Cardinals, Giants and Braves all traded/sold Hornsby because he was more trouble than he was worth.   At the exact same age at which Houston traded Morgan, the Red Sox traded Tris Speaker to the Indians for Sad Sam Jones and Fred Thomas.  So what?   At the same age (after the age 27 season, before 28), Connie Mack sold Eddie Collins to Chicago.   The Reds traded Frank Robinson for Milt Pappas.  So what? 

            Of the 55 players who have earned 400 or more Win Shares, at least half changed teams before they had their greatest seasons (Honus Wagner, Lajoie, Ruth, Bonds, Gary Sheffield, Sam Crawford, Miguel Cabrera, Morgan, Greg Maddux) or while they were still at the top of their game (Reggie Jackson, Hornsby, Jimmie Foxx, Manny Ramirez, Roger Clemens, Cy Young, Tris Speaker, Eddie Collins, Rickey Henderson, Alex Rodriguez, Pete Alexander, Paul Molitor, Dave Winfield, Fred Clarke and others.)   In the case of Hornsby, it is legitimate to use that as evidence of a lack of value, because teams very clearly got rid of Hornsby because they thought that his personality and behavior damaged the team.  In the case of Morgan or Ruth or Roger Clemens or David Ortiz or Bobby Abreu, it is just constructing a narrative to support a pre-conceived opinion.

            Thank you all for reading. 

 
 

COMMENTS (8 Comments, most recent shown first)

Brock Hanke
I am confused by the charts for players until the Morgan/R. Smith one, which lists by age, instead of just 1...15. You did say that you only counted Ruth's best seasons, but his chart has no Win Shares number larger than 37 (entry 14). Checking with Win Shares (still the most important book in my baseball collection), it has Ruth with seasons of 55, 53, and 51 WS, and four more seasons when he had exactly 45 WS. He can't possibly have 18 WS in defense, so I don't understand the numbers. I understand the concept, but not the actual charts.

Rallymonkey5 comments on Rogers Hornsby. I've lived in St. Louis all my life, and got seriously into baseball in 1954, when I was 6. At that time, the Hornsby Story in STL was what I now call the Hagiography Version. It goes like this: In mid-1925, Branch Rickey stopped being field manager of the Cards, and moved upstairs; Hornsby succeeded him as manager. Under Rickey, they had been a bad team. Hornsby turned them right around, and they played under him in 1925 just as well as they'd play the next year (all this is true). The next year, 1926, the Cards won with Hornsby at the helm, in spite of their best player, Hornsby himself, having a very bad year (only hit .316 with power, a miserable season - if you're Rogers Hornsby). This, too, is all true. It's also true that they acquired Pete Alexander halfway through 1926, but he wasn't that great, in the regular season.

Anyway, the Hagiography says that Hornsby and Rickey got into an ego match as to which of them deserved credit for 1926. Rickey said he'd put the team together; Hornsby said that they had amounted to nothing until he became manager. Both had a point, and AT THAT TIME, no one knew that Branch Rickey would become The Mahatma. Hornsby got traded.

Unfortunately, Hornsby wanted a manger job so he could prove that he was The Genius. And he'd been traded to the Giants, managed by John McGraw. Hornsby wasn't going to be the manager. He played great, but McGraw must have thought that he was the Breath of Hell. So, he got traded to the Boston Braves.

The 1928 Boston Braves were a very bad team whose owner tried to turn them into a good team by acquiring two STL superstars: Hornsby and George Sisler. He made Hornsby manager, because that was Hornsby's price. Well, Sisler's days of superstardom were long gone, and the team had nothing else, so they didn't win for Hornsby.

At that point, either Hornsby decided to end his quest for managerial stardom, or Joe McCarthy thought he could handle Rajah without making him manager. This worked great for a couple of years, and then Hornsby came down with injuries that destroyed the rest of his career.

All of this is very plausible, unless you've read Hornsby's autobiography, My War With Baseball, which I did not read until five years ago. In that book, Hornsby made it clear that the problem was between him and Sam Breadon, the owner, and was about money and exhibition games. At that point, five years ago, I realized that the Hagiography Version was dead.

This is the danger of speculating about Hornsby's career - you may end up with the Hagiography Version, because it is so plausible. But it isn't real. The records of the teams he played for are, essentially, irrelevant. Hornsby wasn't hard to deal with because he thought he was a material genius; he was hard to deal with because he was a jerk with an enormous ego, who got away with it because he was JUST THAT GOOD.

The "teams he left not getting worse" comment does have some justification. The 1927 Cards did not get worse than they had been in 1926 because 1) The player they got for Rogers, Frank Frisch, had a career year. In 1927, Frisch was just as good as Hornsby was in 1926, if not better, and 2) The Cards got a full year of Pete Alexander, instead of just half of one. That right there - Ol' Pete - was certainly enough of a boost to justify the three game gain the Cards made in 1927.

But all that is really irrelevant. Hornsby got traded because he was horrible to deal with, not because the teams he was playing for were unhappy to have his numbers. It took Joe McCarthy to deal with him productively.


5:07 AM Aug 25th
 
bjames
tigerlily
Thanks Bill. However, I think you've got an error in there on Mays' WAR totals. You state that his best three years according to WAR are 1963-65 when he scores at 9.1, 9.0 & 9.5 and that these are his only seasons of 9 or more WAR. In fact, per Baseball Reference his WAR totals in 1963-65 are 10.6, 11.0 & 11.2 and he had a total of 9 seasons in his career with a WAR of 9 or more (1954, 55, 58, 60 & 62-66). I think you were looking at his oWAR totals for 63-65 which indeed are 9.1, 9.0 & 9.5.

Here's an interesting thing - WAR & WS agree on at least 1 thing - Mays' best 5-year run is 1962-66 (age 31-35) which scores at 52.3 WAR & 197 WS.


YOU ARE CORRECT. THANK YOU FOR THE CORRECTION.
6:49 PM Aug 17th
 
tigerlily
Thanks Bill. However, I think you've got an error in there on Mays' WAR totals. You state that his best three years according to WAR are 1963-65 when he scores at 9.1, 9.0 & 9.5 and that these are his only seasons of 9 or more WAR. In fact, per Baseball Reference his WAR totals in 1963-65 are 10.6, 11.0 & 11.2 and he had a total of 9 seasons in his career with a WAR of 9 or more (1954, 55, 58, 60 & 62-66). I think you were looking at his oWAR totals for 63-65 which indeed are 9.1, 9.0 & 9.5.

Here's an interesting thing - WAR & WS agree on at least 1 thing - Mays' best 5-year run is 1962-66 (age 31-35) which scores at 52.3 WAR & 197 WS.
11:24 AM Aug 15th
 
Zeth
Being fascinated of late with Orel ('The Shiser') Hershiser and Bob Welch, the two most statistically similar long-career players in history by the old method, I looked them up; these are their respective top 5 seasons by age:

1. Hershiser 29; Welch 30
2. Hershiser 26; Welch 33
3. Hershiser 28; Welch 26
4. Hershiser 30; Welch 28
5. Hershiser 25; Welch 23

A pretty damn good match except for Welch's outlier age-33 season (1990, of course).

By Bill's second method outlined here, if I did my data entry right, they score at 842. I have no idea how high or low that might be, or how far up one another's comp lists by that method it is. Their careers are exactly the same length and pretty close to the same shape, but Hershiser started and finished four years older than Welch did, which drags down their value pattern similarity.
10:15 PM Aug 13th
 
SteveN
Thanks for the larger font. Much easier to read.

I had a really great comment early in you article, but, it seems to have been lost to the bins of wherever my mind is.
12:07 PM Aug 13th
 
Rallymonkey5
I do like the system, another interesting tool to analyze players.

I’m not buying that Hornsby is a special case here. I’m familiar with your writings on him going back to the historical abstracts, but saying that teams trading him and didn’t get worse, trading for him and not getting better boils down to a single team: the 1928 Braves.

The Cardinals won the world series his last year. The Giants won as many games as should have been expected when he played his one year there.

The Braves were bad, got Rajah, he had a great year yet the team was even worse. Legit data point there.

The Cubs traded a poopoo platter for him, improved as expected, and went to the world series.
12:03 PM Aug 13th
 
evanecurb
Interesting system. I'm interested in how predictive the system might be. I'm also interested in how well the system works for good-not-great players, i.e. players who might be expensive to re-sign but are not all-time greats.
11:27 AM Aug 13th
 
DavidHNix
Lil Stoner? Hip hop artist, right?
11:24 AM Aug 13th
 
 
©2021 Be Jolly, Inc. All Rights Reserved.|Powered by Sports Info Solutions|Terms & Conditions|Privacy Policy