Notice: Trying to get property 'display_name' of non-object in /var/www/html/wp-content/plugins/wordpress-seo/src/generators/schema/article.php on line 52
keyboard_arrow_uptop

This week, position players join pitchers and catchers at spring training. For those who travel to Florida or Arizona to cover the teams, reporting on the same story lines can grow tiresome. For others, enjoying watching the same story lines pop up again and again is half the fun.

One traditional source of March copy has been position battles. Teams, wisely or not, sometimes view spring training as a sort of tryout, and so beat writers can craft a narrative around Player A vs. Player B, while maintaining that both players simply want what’s best for the team. Later in the spring, players will battle for the last roster spots-which is a battle with serious implications for the players and their careers. But this early in spring training, the dominant story lines are about rotation spots, particularly the fifth starter’s job. Aside from their dramatic value, do such battles have any significance?


Camptown Racetracks Five Weeks Long, Oh, Doo Dah Day

Let’s start with one of 2010’s battles for a fifth starter job. The most notable is being waged in Tampa between Yankees rotation candidates Joba Chamberlain and Phil Hughes. Both were highly touted starting pitching prospects in the minor leagues, but each has experienced setbacks at the major-league level. Chamberlain, saddled with the unfortunate (and inaccurate) perception that he is more valuable as a reliever, must battle the “B-Jobbers” in addition to Hughes. Hughes, on the other hand, is likely subject to an innings limit and was also effective as a reliever last year (his 3.84 WXRL ranked 14th in the majors last year). This sounds like a perfect battle between worthy adversaries, doesn’t it?

The world champs aren’t the only ones having trouble identifying their fifth starter. The NL champion Phillies, too, do not yet have a clear fifth starter. The primary battle is between the 47-year-old [sic] Jamie Moyer (4.65 SIERA last year) and the newly hirsute Kyle Kendrick (3.34 ERA at Triple-A in ’09). This one’s got built-in archetypes, as the aged veteran attempts to return from multiple off-season surgeries to battle the uneven youngster one last time.

Perhaps slightly less compelling is the battle in Bradenton, Florida. Pirates pitchers Daniel McCutchen (4.94 SIERA in 36 1/3 major-league innings) and Kevin Hart (5.19 SIERA in 81 innings) match up in a battle of young but not particularly promising arms. Though a recent AP report gives Hart the edge in the battle, Raise The Jolly Roger thinks McCutchen is the clear choice. Will the Pirates, who traded for Hart from the Cubs last year, forego a chance to look smart?


Is This Going to be on the Exam?

A more basic question we could ask about all of these battles is whether the ostensible winner will end up pitching more than the loser. Put slightly differently, what difference is there between the pitcher who pitches the fifth-most starts on a team and the pitcher who pitches the sixth-most starts? After all, most teams don’t even need a fifth starter until a few weeks into the season, and use every opportunity they get to skip the weakest link. On the other hand, pitcher injuries are all too frequent. Combined with trades, injuries mean that there is a good bit of rotation in the, er, staff. Of course, you want numbers, and who am I to get off on being withholding?

I ranked each team’s pitchers by games started in 2009. I then took the fifth- and sixth-ranked pitchers on this score and subtracted the number of starts for the sixth man from the number of starts for the fifth man. No team’s sixth starter started fewer than six games (Brad Penny and Ryan Sadowski for the Giants both did). No team’s fifth starter started more than 26 games (Rich Harden of the Cubs). On average, fifth starters took the ball just over 16 times, while sixth starters averaged just less than 12 starts. That’s a very small margin given the importance attached to spring battles. Here’s the full list:

Team      5th Man     5 GS  6th Man       6 GS  Delta
Braves    T. Hanson    21   T. Hudson      7      14
Dodgers   E. Stults    10   V. Padilla     7       3
                             J. Weaver     7
Giants    R. Johnson   17   B. Penny       6      11
                             R. Sadowski   6
Cardinals Wellemeyer   21   M. Boggs       9      12
Cubs      R. Harden    26   S. Marshall    9      17
Mariners  J. Vargas    14   I. Snell      12       2
White Sox C. Richard   14   B. Colon      12       2
Phillies  C. Lee       12   B. Myers      10       2
Tigers    J. Washburn   8   D. Willis      7       1
D'backs   Y. Petit     17   B. Buckner    13       4
Angels    M. Palmer    13   S. O'Sullivan 10       3
Yankees   S. Mitre      9   C. Wang        9       0
Reds      H. Bailey    11   J. Lehr       11       0
Rays      Sonnanstine  22   S. Kazmir     20       2
Marlins   A. Sanchez   16   A. Miller     14       2
Pirates   I. Snell     15   J. Karstens   13       2
Rangers   V. Padilla   18   B. McCarthy   17       1
Red Sox   C. Buchholz  16   D. Matsuzaka  12       4
Indians   C. Pavano    21   A. Laffey     19       2
Blue Jays B. Cecil     17   Rzepczynski   11       6
Royals    K. Davies    22   B. Chen        9      13
                             S. Ponson     9
Athletics G. Gonzalez  17   J. Outman     12       5
           V. Mazzaro  17
Mets      J. Maine     15   O. Perez      14       1
Padres    C. Young     14   J. Peavy      13       1
           T. Stauffer 14
Astros    F. Paulino   17   R. Ortiz      13       4
Twins     K. Slowey    16   C. Pavano     12       4
Nationals G. Mock      15   ---           --       0
           S. Martis   15
           J. Martin   15
Orioles   R. Hill      13   K. Uehara     12       1
                            C. Tillman    12
Brewers   D. Bush      21   M. Burns       8      13

One thought you might have is that better teams tend to have less turnover in their rotation. But a quick regression of the delta term against team wins revealed a very low correlation (r-squared = .005). More significant, however, was the correlation between team-wide starters’ ERA (r-squared = .14). This suggests that perhaps some of the explanation for why a fifth starter might get more starts than the sixth starter is due to the overall quality of the team’s starters. However, the low correlation also points out the essential randomness of the baseball season. The excellent Braves (3.52 starters’ ERA in ’09) got just seven starts from their No. 6 (Hudson), but the nearly as good Dodgers (3.52) got just three more starts from their No. 5 than from their No. 6. At the other end of the spectrum, the poor Brewers squeezed a 5.37 ERA out of their starters and yet their fifth starter (Dave Bush) took the ball 21 times.

Question of the Day

Certainly, there are better ways of measuring which is the fifth and which is the sixth starter than mere games started. But does this simple method show that, over the course of the season, even the guys who lose out on the fifth starter’s job can still expect to start at least 7-10 games per year? Is there a psychic significance to making the rotation out of spring training I am overlooking? How about an economic difference?

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
You need to be logged in to comment. Login or Subscribe
brooklyn55
2/23
Clarity? SIERA? Mixing of SIERA and ERA? This is getting like the Olympics in INCREASING the level of difficulty (of understanding).
ScottBehson
2/23
Omar Minaya once wisely said (and then did not take his own advice) that you don't build a rotation of 5 starters, you build a rotation of 8 or 9 for the long season.
jpaternostro
2/23
Which is why, of course, the Mets have rolled out such luminaries as Jose Lima, Jeremi Gonzalez, Dave Williams, Alay Soler, Brian Lawrence, Jorge Sosa, Brandon Knight and Claudio Vargas the last few years.
yankeehater32
2/24
Omar Minaya's theory works better with the Theo Epstein's ability to identify and acquire talent.
BillJohnson
2/23
"Psychic significance," probably not, but there are abundant effects in roster construction, particularly if the nominal sixth starter is out of options and therefore can't be stored at AAA with a "break glass in case of emergency" sign around his neck. There's probably an interesting roster-construction study in your data somewhere, although I can't immediately see how to do it.

Another point, btw, is that getting the fifth starter "right" is more important for some teams than others. Wrangling over the fifth starter doesn't make any sense at all if you're Pittsburgh or Kansas City or Washington. If you're Colorado or Minnesota (or several other teams), by contrast, the one or two games that the "right" fifth starter might gain you compared to the "wrong" one who's relegated to the sixth spot could loom very large at the end of the season. I really don't blame those teams, the ones who are serious contenders but not locks to win their divisions, for hosting these battles, nor commentators for homing in on them.
smitty99
2/23
Here's another way to look at it. Last season the Phillies top four starters (in games started) were Blanton; Hamels; Happ and Moyer. Eight other guys had starts for a total of 51 starts!

While Cliff Lee and Pedro Martinez were two of those guys, Brett Myers; Chan Ho Park; Rodrigo Lopez; Andrew Carpenter; Kyle Kendrick and Antonio Bastardo were the others.

Almost all teams get a pretty high number of starts from guys who aren't in their original first four or even five. The battle for the 5th starter really pales in comparison to the number of starts given to mid season trade guys and injury replacement guys and guys who get called up to replace someone who just isn't doing the job.

I think a lot is made of who is going to be the 5th starter in the spring but it doesn't end up being as important as who are going to make all those other starts.

TheRedsMan
2/23
I think you set up the analysis in such a way as to make the conclusion both self-evident and misleading. The battle in Spring Training is not for what player will get the 5th most starts. You've introduced a selection bias wherein if a guy who breaks camp as the "5th starter" performs well and, perhaps most of all, stays healthy, he won't be looked upon as the 5th starter using your definition.

But that's not what a 5th starter is. The 5th starter is the guy whom the team expects to be the 5th most effective starting pitcher. Thus, the point of comparison which is more appropriate is the performance of the the guy who opened the season as the "5th starter" compared to the rest of the guys who made starts but who were not part of the rotation when the season started. And the reason the battle in ST matters is because it's possible that a better pitcher will never get an opportunity to be the "5th starter" if he doesn't win the job at the start of the season. Your analysis obscures this.

For example, let's say a team gives a guy, let's call him Joe Veteran (or Kip Wells if you prefer), the 5th starter job and he puts up a 5.50 ERA while making 28 starts. He gets skipped a few times but pretty much keeps his rotation spot all year. Meanwhile, Stud McRookie loses out in the ST competition and goes to AAA.

Midway through the season, the team's #2 starter goes down and Stud McRookie gets called up. From that point forward, he makes 14 starts and puts up a 4.50 ERA. Let's also assume that a collection of other guys get a handful of starts due to various injuries.

Looking at the guys who opened the season in the rotation:
#1 starter = 34 GS, 3.25 ERA
#2 starter = 18 GS, 3.75 ERA
#3 starter = 26 GS, 4.25 ERA
#4 starter = 30 GS, 4.75 ERA
Joe Veteran = 28 GS, 5.50 ERA
-------------------
Stud McRookie = 14 GS, 4.50 ERA
other guys = 12 GS, 6.00 ERA

Using your methodology, we wouldn't consider Joe Veteran the #5 starter. You'd be looking at the guy who was expected to be the teams' 2nd most effective starter as the "5th starter" because he spend half the year on the DL. And then you'd be saying that the 5th starter decision wasn't that important because there was only a 4 start difference.

But that's not the right point of comparison. We should be looking at the fact that Joe Veteran put up 28 GS of a 5.50 ERA while Stud McRookie only got 14 GS while putting up a 4.50 ERA. That's going to mean 1.4 wins worth of difference because of your choice. That's not a huge deal, but it's significant in my book.
tbsmkdn
2/23
This is a legitimate concern, and it's one that I considered. However, I must dispute your assertion that I set up the analysis so as to make the conclusion "self-evident" and "misleading." I think that's an unfair characterization that was uncalled for. To the degree you are suggesting I am being disingenuous, I must deny the charge.

Allow me to present the argument slightly differently. Look from the standpoint of the ostensible sixth starter. By design, he is the sixth best starter on the team. That means, assuming he can stay healthy, he should get at least the sixth most starts on the team. Even though my data only looked at who received the 5th and 6th most starts, it still demonstrates by implication that guys who enter the season as the 5th starter and stay healthy will often end up with more starts than that. The exact same logic applies to the guy who "loses" the spring training battle, and implies that sixth starters are nearly guaranteed to join the rotation for a considerable amount of time at some later point in the season.

The win consequences are a separate matter not considered here. The best practice (somewhat trivially) is always to give playing time to players in decreasing order of value. Plenty of things are going to affect that, but what you call a guy ought not to. I don't think focusing on the nomenclature is particularly helpful, nor would focusing on who is the starter who pitched most often on Tuesdays be.

The simple point I intended to demonstrate is that the sixth starter can anticipate significant playing time despite the fact that he lost the battle.
tballgame
2/23
I have to agree with cdmyers here. Without implying that the selection bias was premeditated, it has the effect of making the conclusion self-evident. If the point you were trying to make is that teams need more than 5 starters to get through a season so losing the battle for 5th is not a huge imposition, it is a fair point and it deserves to be framed differently.

Looking for the exact point in the column, this - "A more basic question we could ask about all of these battles is whether the ostensible winner will end up pitching more than the loser. Put slightly differently, what difference is there between the pitcher who pitches the fifth-most starts on a team and the pitcher who pitches the sixth-most starts?" isn't as slight a difference as you set out (again, just my opinion).


If you want to look at 5th and 6th starter battles from last spring, you may wind up with a very different conclusion. The veterans that lose the competition (e.g., Colon) may end up getting waived, while young players with options can be sent down and saved for in season, and others get to while away in relief. Who were the winners and losers of these camp battles last year? I think you'd find the losers do not consistently wind up ranking in that team's top 6 GS (guessing 50-60% do). In particular, I'm guessing players without options that lose that battle become roster footnotes.
TheRedsMan
2/24
Tommy, sorry if my comments came across as mean-spirited as it seems they may have. That certainly was not my intent. Rather, I think that while your ultimate conclusion is certainly accurate, there is a lot of turnover so the 6th guy is likely to get a fair number of starts, that your analysis did not really address the question you posed.

You set up the analysis with "A more basic question we could ask about all of these battles is whether the ostensible winner will end up pitching more than the loser." This very clearly means specific Player A (the guy who wins the #5 job) vs. specific Player B (the guy who loses the #5 job).

However, you then immediately re-frame the question fundamentally by making it not about the players in competition in the spring, but by the players who eventually wind up with the 5th and 6th most starts at the end of the season. For all we know, those guys could have been the #1 and #2 starter out of camp --- telling us nothing about the eventual opportunities of the guy who made the team and the guy who didn't. By doing it this way, all we're learning about is the general distribution of starts among starters, not about the starts for the two players on whom the entire article is premised.

So I've posited that despite your conclusion, which might lead one to the understanding that the answer to this question is 'no', spring training battles don't really matter, that the answer is actually yes, such battles do have significance. Because while it's highly likely that the #6 starter will see some action, he will not likely see as much action as he would have had he begun the year in the rotation -- or as much as the guy to whom he lost the ST competition. Unfortunately, you didn't really analyze this comparison.

Yes, rotations are turbulent and the guy who loses the competition for the #5 job will get some opportunity to pitch. We can agree on that. But the amount of difference between some opportunity of unknown proportion due to roster turbulence and the opportunity that the loser would have had if he had earned the job out of spring training is still an open question.
mpirani
2/24
This is a statistical analysis of the GIGO type. I would posit that, given the number of pitchers who break down, the "fifth starter" going into the season, if he's at all effective, would frequently end up being the 3rd or even second starter by the end of the season in terms of games started.

To summarize the results of this "study": there's not much difference between the number of games pitched by the guy who pitches the fifth highest and sixth highest number of starts per team. Really? Wow! I'd be quite interested if there's much difference between the number of doubles hit by the players that rank fifth and sixth on each team too, that's about as relevant and interesting. To quote Diego Montoyo, "I don't think that means what you think it does."
irablum
2/25
I know I'm coming into this dicussion late, but to be frank, I was hugely dissapointed by the article. As soon as I looked at the data used I knew there was a problem.

Rich Harden? Vincente Padilla? Cliff Lee? Tommy Hanson?

This is an example of not doing your homework. Not all teams treat their 5th starter the same way. Some teams (most notably the mid-90's Braves) worked a 4+ rotation where the 4 top starters always pitched on their normal rest and the 5th starter pitched whenever needed. Other teams had a more true 5 starter rotation and allowed their pitchers more (or less) rest so that they would continue to go in order.

Whats more important is that rotations evolve over the season, and the 5 guys who start the season as the rotation usually don't all end up taking their turn each year. A perfect example is last years Rangers. In 2009, the Rangers broke camp with a Rotation as follows: Kevin Millwood, Vincente Padilla, Brandon McCarthy, Matt Harrison, and Kris Benson. Their number 6 starter turned out to be Scott Feldman, as he was the first person put into the rotation when it was obvious that Kris Benson sucked (something that most of us fans saw during spring training).

An hour or so with retrosheet (or some other boxscore like site) would allow you to have reconstructed each teams starting 5 in that way, with the number 6 starter as the first pitcher outside the starting 5 to start a game.

In the case of the Rangers, its quite obvious that the difference between their number 5 and 6 starters was immense, but in the opposite direction, as Feldman ended up with an SNLVAR of 5.6 in 31 starts, and Benson had a -0.4 SNLVAR in 2 starts. This example would have been an extreme addition to your main thesis.
tbsmkdn
2/25
I'm sorry to hear you are disappointed. I do my best to provide an interesting perspective, and it sounds like I've failed you this time. I'll work hard to win your readership in the future.
irablum
2/25
Incidentally, I found it amusing that you started off your discussion with the talk about the fight between Joba and Phillip Hughes for the 5th spot in the Yankees Rotation and then under the Yankees you listed the 5th starter as Sergio Mitre and the 6th as Chien-Ming Wang, when my research showed that the Yankees 5th starter in 2009 was Joba Chamberlain and their 6th starter was... Phillip Hughes.
tbsmkdn
2/25
Just to be clear, I defined the fifth and sixth starters differently than the traditional definitions. If that makes you distrust my conclusions, fair enough. But I only looked at number of games started. That's why I listed Mitre and Wang.
irablum
2/26
Its not a matter of trust. The problem is that you measured something different than what your purpose was. It would be like me trying to measure who was the best fisherman by seeing who brought home the largest store bought catfish.

Let me quote you back to you: "A more basic question we could ask about all of these battles is whether the ostensible winner will end up pitching more than the loser. Put slightly differently, what difference is there between the pitcher who pitches the fifth-most starts on a team and the pitcher who pitches the sixth-most starts?"

The two sentences here seem to follow from one another, but in reality they don't. With the volatility of todays teams, the times where the pitcher who starts the fifth most games and the pitcher who starts the sixth most games are almost never the two pitchers who were vying for the fifth spot in the rotation. I gave you two examples above. In one, (the Rangers), the "winner" of the number 5 spot (Kris Benson) quickly pitched his way out of the rotation and even off the team, while the "loser" ended the season as the teams number 1 starter. In the other, (the Yankees), the "winner" ended up as the number 4 starter all year, and the "loser" was the number 5 starter for a month or so before settling into a role as the primary setup man.

Now, if your analysis was to go through each team, identify who their number 5 and 6 starters were coming out of camp in 2009 and telling what happened to them, that would have been a great article. (and in fact, what I was expecting when you started talking about this years camp battles). Instead you took a much easier way out by "identifying" each teams 5th and 6th starters based on number of starts, ignoring trades and injuries, and then tried to perform regression analysis on it.

Once again, as I said above, seeing you talk about Rich Harden as a number 5 starter for the Cubs bothered me. (though apparantly the Cubs had him as their number 4 starter behind Zambrano, Dempster, and Lilly) A little checking reveals that the Cubs number 5 starter at the beginning of 2009 was Sean Marshall with Randy Wells as their number 6 starter. Marshall ended up the year as a 6th or 7th inning pitcher, behind Chris Marmol and Kevin Gregg.

If Rich Harden upset me, I'm sure Phillies and Mariner fans objected to the idea of Cliff Lee as a 5th starter, as well.

cdmyers
2/23
At least some of the so-called 5th starters logged so few starts because they were acquired mid-year in deadline deals. I'm looking at you Jerrod Washburn. These aren't 5th starters in the traditional sense; their low number of starts is due pretty clearly to their only being around for half a year.

It would take a bit of digging, but the comparison you want to be making is between pitchers who are the 5th pitcher to start a game for each team and players who are the 6th pitcher to start a game for their team during the year. That's a pretty clear expression of team's preference orderings coming out of camp, and would give a much clearer picture of whether winning the positional battle matters.
cdmyers
2/23
Also see Kaszmir (traded away by the Rays) Lee (Traded for by the Phillies). Neither of these are 5th or 6th starters by any stretch of the imagination.
tbsmkdn
2/23
Right. The point of the exercise isn't to claim that Cliff Lee is a fifth starter. It's to show much much turbulence there is in a major league staff, which suggests that even sixth starters get plenty of starts.
cdmyers
2/23
Fair enough, but you say the goal is to see "whether the ostensible winner will end up pitching more than the loser." That's not what you're looking at in a sizable number of these comparisons.
mpirani
2/24
The point of the exercise is that you crunched some numbers that don't support answering the questions you were trying to address. No sweat; that's why this kind of stuff should be peer-reviewed before it's posted. But you continue to talk around the criticisms: there's "much turbulence" in a staff, which suggests that sixth starters get plenty of work? Maybe so, but your study does nothing toward answering that proposition. Nada. Zip. Zero. It's really kind of embarrassing, as far as I'm concerned. As you can tell, I just get rather worked up over people applying statistical techniques in such a sloppy fashion. As you ask above, "is this going to be on the exam?" If so, you flunked.
tbsmkdn
2/24
Thank you for taking the time to comment.
misterdelaware
2/23
Where's Colorado? Their 5th starter (Cook) had 27 GS, their 6th starters (Morales & Contreras) had 2 each.
benharris
2/24
He's right. The Rockies had 5 pitchers start at least 27 games, and four of them started 30.
StarkFist
2/24
Cook isn't the Rockies' 5th starter. He lost starts due to injury issues.
misterdelaware
2/24
For the purposes of this article, Cook would qualify as the Rockies 5th starter as he made the 5th most starts for the team in 2009.

Jimenez: 33
Marquis: 33
de la Rosa: 32
Hammel: 30
Cook: 27

(I'm a bit curious why this isn't being addressed. I know its only a one team omission but the fact that Colorado is the biggest possible outlier on the high side makes it significant.)
tbsmkdn
2/24
Yep, good catch! Colorado must have slipped off my spreadsheet at some point in the transition. I regret the error.
EnderCN
2/23
To be fair to the poor Brewers 4 of their 5 initial starters got hurt last year(Looper and Parra pitched through it with poor results and had surgery after the season) and one(Gallardo) wore down because he was hurt the year before. Bush really wasn't the 4th starter his total was like that because he got hit by a line drive.
canada
2/24
I might be looking at this a different way, but as some have noted injuries and player moves distort these figures. I think the more relevant comparison would be between the guy who broke camp as fifth starter (eg. made his first start the game prior to the #1 starter's second game) and the first pitcher to be starter #6 when any of #1 through #5 stops starting due to injury or poor performance.

PWHjort
2/24
Reminds me of one of my favorite Rob Neyer quotes (http://sports.espn.go.com/espn/blog/index?entryID=3999762&name=Neyer_Rob):

"We probably spend just a little too much psychic energy obsessing over fifth starters. In the 1940s, a baseball writer might refer to a team's "Big Four," even though no team was going to get through a season using just four starters. Granted, there is unpredictability and instability almost everywhere. But for many teams, that No. 5 slot in the rotation is particularly unstable, and the real key isn't picking the right guy in March; it's having viable options over the course of the season."
omarwhite
2/24
Following up on some other posts, the fifth and sixth most starts on a team looking back at the season is a function of injury, player movement and ineffectiveness. The more interesting question is how decisions on 5th and 6th starters effected the number of starts given to those players (and how that impacted wins). The Rays were chosing among Price, Sonanstine and Niemann. There were other considerations, but choosing Price over Sonanstine cost the Rays 10 or more starts by Price and forced them to struggle through 10 or more starts with Sonanstine. Hanson on the Braves, the decision to give Reyes and then Meldin the nod over Hanson cost the Braves ten starts by Hanson (would the Braves season have been different if he started in the rotation). These decisions (both the spring training battles and the decisions regarding signing veterens to fill the fifth slot (e.g. Brad Penny) can be significant because they can result in a pitcher starting 10-15 games that they shouldn't and/or delaying an impact pitcher (particularly, when service time issues come into play) and having them pitch only a half a season. While many more of these decisions are irrelevant, the Price/Sonnastine decision or the Penny/Bucholz/Masterson/Smoltz decisions could have been far more important if the Rays didn't have a number of other things run against them last year.