Welcome to the first "Doctoring the Numbers" of the year. Traditionally, DTN is a an in-season column, but I’m setting
that aside this week. I write these columns as a gut reaction to something I’ve recently encountered–an interesting statistic, an
compelling quote, or a thought-provoking assertion. Two weeks ago, Rob Neyer made a statement that, to me, definitely fell into the
last category.
In response to a reader’s e-mail questioning whether, as Conventional Wisdom states, left-handed pitchers tend to develop later than
right-handers, Neyer wrote:
"…Anyway, nothing magical happens for left-handers when they turn 26, or 27 or 28. Bill James studied the issue and ran a
chart including data for every left-handed pitcher born between 1920 and 1939 (The Bill James Baseball Abstract 1983, page
211). The lefties posted a .505 winning percentage from ages 23 through 26, and a .507 winning percentage from ages 27 through
30."
I was intrigued, and pulled out the 1983 Abstract, where I did in fact find a chart compiled by Bill James of the combined
win-loss records for all southpaws born between 1920 and 1939, separated by age. As James wrote:
"Did you ever hear the saying that lefthanders don’t find themselves until age 26? It’s obviously a ‘magic’ theory – there is a
magic about turning 26 – but people say it, so there must be people who believe it…on balance, though, there’s nothing to it.
Nothing too much happens to left-handed pitchers as a group at age 26. Their wins increase by 17%; their losses increase by
13%."
More precisely, the chart shows that all 25-year-old lefties went 617-627, a .496 winning percentage, while 26-year-old lefties went
721-711, a .503 winning percentage. The winning percentage of lefties from age 24 to age 30 is remarkably stable, never wavering
from .500 by more than 20 points and showing no obvious trend.
Before I give you some of the other details from the chart, I want to stop and talk about what might be wrong with this kind of
analysis.
The winning percentages suddenly jump dramatically at age 33 (from .501 at age 32 to .547), and stay above .540 consecutive years.
Why might that be? James theorized that this is because "pitchers who come through the ‘fastball crisis’ about age 30 often
come out on the other side as better pitchers than when they went in."
That’s one theory. It’s a theory that I find incredibly untenable, though, because of the study’s fatal flaw: by combining the
records of all left-handed pitchers together, irrespective of who they are, the study has an enormous selection bias: you’re only
counting the records of those pitchers who were good enough to pitch. Are pitchers aged 33 to 37 really better than pitchers in
their mid-20s, or is that most pitchers still throwing in their mid-30s are really good pitchers? How many rookie starters trying to
stick in the majors are 35 years old?
If we take James’s theory another step, the highest winning percentage on the entire chart was posted by 42-year-old pitchers, who
combined for a .690 winning percentage. Do pitchers really peak at 42, or is the study biased by the fact that the combined 29-13
record for 42-year-olds consist of Warren Spahn (23-7) and everyone else (6-6)?
To give you an idea of how this bias can work, I’m going to run a similar list on a different subgroup of players. Here is a chart
of the combined batting average of all switch-hitters born between 1920 and 1939:
Age BA Age BA Age BA 19 .261 26 .237 33 .253 20 .263 27 .252 34 .264 21 .269 28 .241 35 .241 22 .254 29 .250 36 .246 23 .257 30 .249 37 .242 24 .255 31 .250 38 .278 25 .248 32 .251 39 .247
Looking at these numbers, you would never guess that hitters peak at age 27, as they do, but that they hit their best in their teens
and early twenties, and have a smaller peak in their mid-to-late thirties. The highest combined averages were posted by 38-year-olds
and by 21-year-olds. Why? Because more than half the 38-year-old at-bats came from Maury Wills, who hit .281, and about 70%
of the 21-year-old at-bats came from Mickey Mantle, who hit .295.
I don’t want to knock James’s work; he was, after all, doing this in 1983, and probably compiled the entire list by hand. But the
truth is that his chart provides essentially no evidence on the subject of when left-handers peak. It certainly doesn’t provide
evidence on whether they peak later than right-handers do.
To tackle the subject, I decided to look at the data a little differently. I tried to determine at what age left-handed pitchers
were most likely to have a "breakout" season. To run that study, though, first we have to define a breakout season. A
simple definition would be a season in which a pitcher:
- made at least 20 starts
- had a winning record
- had an ERA under 4.00
- had a win-loss record at least two full games better than his previous career best
- had an ERA at least 50 points lower than his previous career best.
Between 1920 and 1999, there were 207 such breakout seasons. (Note that if a pitcher had more than one breakout season, only his
first one was counted.) The most recent examples are Russ Ortiz and Todd Ritchie (in 1999), and Bartolo Colon
and Jose Lima (in 1998).
Sixty-two of the 207 pitchers are left-handed. Here is how those 62 pitchers break down in terms of their age during their breakout
season:
Age Breakouts 21 3 22 7 23 8 24 13 25 5 26 16 27 4 28 2 29 0 30 1 31 2 32+ 1
The youngest pitchers in the group were Curt Simmons, barely 21 when he went 17-8, 3.40 for the Whiz Kid Phillies in 1950
(after going 4-10, 4.59 the year before), and Steve Avery, also 21 when he went 18-8 for the 1991 Braves (he went 3-11 as a
rookie in 1990).
While the data isn’t conclusive, I think it’s interesting to note that more pitchers had their breakout seasons at age 26 than at
any other age. However, exactly half of the 62 pitchers had their breakout season before age 25, so left-handers certainly don’t
seem to develop later than pitchers as a whole.
Or do they? We can’t claim that lefties develop at the same rate as pitchers as a whole without comparing them with a control group,
in this case, right-handed pitchers. Here the same list of how old the 145 right-handed pitchers were when they had their breakout
season:
Age Breakouts 19 1 20 2 21 6 22 10 23 15 24 20 25 28 26 24 27 11 28 12 29 4 30 2 31 3 32+ 7
Wally Bunker was just 19 when he went 19-5, 2.69 as a rookie in 1964 (he qualifies as a "breakout" because he
pitched four innings the year before). The four 20-year-olds to make the group are Milt Pappas, Dwight Gooden, Bob
Feller, and Jim Palmer. Not a bad bunch.
Whereas age 26 was the mode of the left-handed group, the most common age for right-handed breakouts was 25. However, only 37% of
right-handers peaked by age 24, compared to 50% of lefties. The average age of right-handed breakout pitchers was 25.93. The average
age of southpaws was 25.36, more than half a year younger than their counterparts.
In other words, while lefties frequently do peak at age 26, overall they do not appear to develop any later than right-handed
pitchers.
Let’s look at this topic from a slightly broader viewpoint. While left-handed pitchers may have their breakout season at the same,
or even a slightly younger, age as right-handers, could it be possible that another view of the data would lead us to a different
conclusion? What I wanted to find out was whether, given a left-hander and a right-hander of the same age, the left-hander has more
potential for growth than the right-handed pitcher.
The most direct way of determining this would be to take similar groups of young pitchers, split them into right- and left-handed
groups, and see how their collective careers turned out. I limited this study to rookie pitchers between 1920 and 1985 who:
- made at least 10 starts
- threw between 100 and 200 innings
- finished within four games of .500 (i.e. between 6-10 and 10-6)
- had an ERA within 0.60 of league average.
While that is a pretty wide range, the resulting groups had very similar numbers. For example, there were 51 such pitchers aged 21
or younger. The 14 lefties averaged an 8.6-7.8 record with a 3.85 ERA; the 37 righties averaged 8.4-9.1 with a 3.80 ERA.
The average career totals for these two groups?
LHP: 95 W, 88 L, 3.69 ERA
RHP: 100 W, 97 L, 3.70 ERA
There was essentially no real difference between the two groups. Here’s a chart for the career records of all pitchers in these two
groups, at every relevant age:
LHP RHP
Age W L Pct. IP ERA W L Pct. IP ERA
21- 95 88 .520 1640 3.69 100 97 .511 1727 3.70
22-23 67 74 .475 1254 3.86 74 71 .510 1282 3.74
24-25 66 57 .533 1105 3.72 46 50 .476 866 4.04
26-27 42 45 .484 772 4.08 41 42 .496 742 3.95
28+ 31 29 .521 517 3.73 44 37 .538 718 3.81
There’s no obvious trend to these numbers. Perhaps the easiest way to compare these two groups is to look at their innings pitched,
to get a sense of their career length. At age 28 and higher, the right-handers have a fairly substantial edge in career innings, but
I’m not sure how much of a real effect that is, given how small the sample size was in that age group.
What interests me is the large edge in IP (and overall performance) by lefties in the 24-25 age group. Perhaps, if lefties do take a
little more time to gain a foothold in the major leagues, those left-handers who are just establishing themselves at age 24 or 25
have a more solid future than right-handed pitchers of the same age?
I ran the same study again, but this time I changed the requirements in an effort to find rookies who had less-than-impressive
rookie seasons. Specifically, pitchers who:
- made at least 10 starts
- threw between 100 and 200 innings
- had a winning percentage of at least .333, but no higher than .500
- had an ERA no lower than league average, and no higher than 1.00 above average.
If left-handed pitchers really do develop a bit more slowly than right-handers, it’s possible that they might be more likely to
shake off a bad start to their career. Let’s roll the film:
LHP RHP
Age W L Pct. IP ERA W L Pct. IP ERA
21- 134 120 .534 2241 3.57 86 86 .500 1510 3.72
22-23 55 63 .469 1047 4.01 59 62 .486 1079 3.98
24-25 70 59 .543 1171 3.56 60 64 .482 1105 3.95
26-27 28 31 .472 569 4.27 33 38 .469 635 3.91
28+ 31 38 .447 570 3.86 18 20 .472 359 4.09
While the differences aren’t striking, these numbers lend a tiny bit of credence to the theory that left-handers develop a little
later than right-handers. In particular, lefties in that 24-25 age group win more games, lose fewer, throw more innings, and have a
significantly lower ERA than their right-handed counterparts. But right-handers do slightly better in both the 22-23 and 26-27 age
groups, so if lefties do have a legitimate development advantage, it’s obscured by a lot of noise.
This horse was dead a few paragraphs ago, so a few extra swipes at it aren’t going to hurt. I wanted to explore this topic from one
other angle. This final study is a retrospective one: what I want to do is isolate pitchers who developed into quality starters,
then take a look back to see how they were doing a few years before. The group for this study includes pitchers between 1946 and
1999 who:
- made at least 20 starts
- won between 12 and 18 games
- had a winning percentage of more than .500, but less than .700
- had an ERA between 50 and 100 points above league average.
For example, in the 24-year-old group we have 11 lefties and 27 right-handers. The most recent right-hander to qualify was Chan
Ho Park, in 1997. The most recent left-hander was Greg Swindell, in 1989. I looked at how each of those pitchers did
three years before the season in question, i.e., how Park did in 1994, or how Swindell did in 1986.
As a group, the lefties averaged 3.2 wins and 1.5 losses with a 3.38 ERA three years prior to their good season; righties averaged
3.8 wins and 3.3 losses with a 3.44 ERA.
Here’s that data in chart form, for every age between 24 and 33:
LHP RHP
Age W L Pct. IP ERA W L Pct. IP ERA
24 3.2 1.5 .686 44 3.38 3.8 3.3 .539 62 3.44
25 4.3 2.9 .600 67 4.05 3.2 3.6 .473 58 3.92
26 7.6 6.3 .547 129 3.46 5.5 5.1 .516 96 3.69
27 6.3 5.3 .540 102 3.59 6.3 5.7 .525 105 3.59
28 8.6 6.8 .560 134 3.15 8.6 8.3 .510 149 3.62
29 10.0 6.3 .613 159 3.17 10.1 9.3 .522 171 3.42
30 9.3 9.2 .502 169 3.42 12.4 11.0 .533 190 3.37
31 9.1 8.1 .529 146 3.05 10.2 8.1 .556 167 3.57
32 9.6 10.1 .487 175 3.87 10.5 10.5 .500 196 3.35
33 9.0 7.6 .543 145 3.02 14.6 7.6 .658 192 3.12
Well, I’ve had enough. There is simply no compelling evidence that southpaws develop any more slowly than right-handers do. This
chart, remember, shows us what the pitchers in each group did prior to having a good season. In other words, the worse the
performance in this chart, the worse those pitchers were three years before, and the more they’ve developed in the interim.
The problem is that the lefties, as a whole, were not any worse than the right-handers, and in some cases were better. In only one
group (age 32) did the left-handers compile an ERA more than 15 points worse than their right-handed counterparts three years prior.
In contrast, the lefties were more than 15 points better on four occasions, at ages 26, 28, 29, and 31.
I was hoping I’d have a nugget of wisdom to offer those of you brave enough–or masochistic enough–to slog your way through this
mess of a column. Unfortunately, the only thing I can offer you is this: Bill James was right. Maybe a good percentage of
left-handers have a breakout season around age 26, but that’s only because a good percentage of all pitchers break out at that age.
Yes, there are many examples of left-handers who didn’t reach their full potential until their late 20s, but there are just as many
examples of right-handers who took just as long. For every Sandy Koufax, there’s a Dazzy Vance. For every Al
Leiter, there’s a Red Ruffing. For every John Tudor, there’s a Bob Tewksbury.
So when you’re trying to decide whether a pitcher whose prospect luster has begun to fade is finally going to find himself, just
remember that the arm he throws with really doesn’t seem to matter.
Rany Jazayerli, M.D. is an author of Baseball Prospectus. You can contact him by
clicking here.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.
Subscribe now