The Dodgers’ Fifth Starter
Amusing quote from Buster Olney’s column yesterday:
“We’d love to have a bona fide No. 5 starter,” said Colletti.
Olney frames this quote in response to Colletti listing a handful of names that qualify as unknowns, has-beens, or never-will-be types. Why is this amusing? Because, a bona fide number five starter generally fits into one of those categories. Matthew did some work on rotation slots in reality by tRA in late 2008. He found that number five starters were something like 24% below league average. Dave Cameron chimed in below in the comments section and it’s worth a read, if only for some constructive thinking. Matthew surmises the number five starter in his piece as such:
TO BE PERFECTLY CLEAR: By this definition, a #5 starter is probably not what most people term a #5 starter. I assume that when most people talk about a #5 starter, they talk about some mythical rotation that almost never misses a start and this person being the worst pitcher on it. For the most part, those rotations do not happen. A #5 as defined below represents the combined worst starters to have actually pitched. In my opinion, this is the more useful definition, because this way, having health is properly weighted and you get a notion for the level of scarcity that exists.
By this definition, the Dodgers do have a bona fide number five starter already. Even last year, the Dodger pitcher who made the fifth most starts was Eric Stults. He started 10 times and had a 5.54 xFIP. The list of alley cats Olney provides includes Charlie Haeger (profiled on this very site multiple times), James McDonald, Ramon and Russ Ortiz, and even Josh Towers amongst others. Frankly, I had no idea a few of those guys were still pitching baseballs. McDonald, though, is an interesting option.
McDonald has made nearly 90 starts in the minors, including 42 between Triple- and Double-A. At both destinations McDonald struck out at least nine per nine and walked between three and four batters per nine. He turned 25 in October and started last season in the Dodgers’ rotation. He would only make four starts, as he walked 14 in 13.1 innings and struck out only six. Upon a move to the bullpen, McDonald looked like his minor league self, posting a SO/BB of 2.4 and striking out roughly one batter pr inning.
His stuff doesn’t seem to stink, either: a low-90s fastball, curve, and change. Each pitch was whiffed on at least 8% of the time. His fastball shows great “rise” which makes up for some lackadaisical run. Those whiff rates will likely decrease upon a move back to the rotation, but McDonald’s body of work makes him more appealing than the Ortizes of the world. Plus, who knows, maybe he turns into more than a bona fide number five.
45
Oh good…one more thing that Colletti doesn’t get
Well, he learned everything he knows here in S.F.
Your definition of the #5 starter is self serving. Nobody would dispute that the worst 1/5 of starts degrade pitching performance. The problem is that manager’s do not know who their worst starters will be. The #5 spot is often the place where you get the most bang for the buck, a veteran coming off an injury, or a young rookie in his first year. Saying that the bottom 25% of starts are below average is not insightful. Check out the preseason depth charts, there will be many great options in the #5 spot before the season is over.
Saying that you wish you had another innings eater, or a promising rookie, or more start pitching depth may have been a better statement than wanting a bona fide #5 starter, but none of these things are bad to wish for.
I suppose taking this analysis further into other areas would yield “pick only stocks that go up”, or “date people with good futures”, or maybe “work only at places that will grow”.
“The problem is that manager’s do not know who their worst starters will be.”
But they do know who their worst starters are, and probably work them into the #5 spot of the rotation since you can skip them occasionally. As far as I know, there’s no data point attached to a rotation spot like there is for batting order.
How would you suggest he identify #5 starters more appropriately? Easy to say “you’re doing it wrong”…little tougher to be constructive. ;-)
I have a proposal couple proposals. Mathmatically, you could consider the 5th starter to be a position in a rotation- which is cyclical. Since the opening day starter is usually the ace, and they usually work everything else around him, the 5th starter is the guy who is pitches the 4rth start after him (5th start total) unless he has already pitched a game in that period. Teams do not tend to shuffle the rotation midseason, so this should be relatively straightforward.
This will work so long as the ace remains healthy. Since he may not, you can probably extend your coverage by doing a similar check about who does the 3rd start after the 2nd starter of the season, and the 2rth starter after the 3rd guy. (Again, excluding anyone you have already designated as being the 1rst, 2nd, or 3rd). If any of the initial 3 stop playing or get out of sync with the rest of the rotation (such as by injury) you stop using them and rely on the remaining guys. The guy who has the highest count of such starts is your 5th starter. So basically, you get to immediately exclude 3 guys from being the 5th starter and then use their start positions to infer who tended to be the last guy they would use (and by association, also the most likely guy to be skipped, if that is a useful stat for this).
Alternatively, you could consider the “5th starter” to be a purely psychological construct rather than a physical position. Determining who the 5th starter is then is easy in theory, but harder from the available data and tools people tend to use here. Basically, anybody with a news webscraper could parse articles and find quotes by the manager that include the name of a pitcher and terms that mean “5th starter.” The guy with the highest score on that metric is your 5th starter.
Or if we want to be really lazy, we can just take a mode of the rotation depth charts over time from the major baseball sites. Whoever occurs most often in the 5th spot on those depth charts is your 5th starter.
So, there are 3 approaches. 1. Mathematically infer who is the 5th guy in a cycle of at most 6, relative to positions that are assigned initially to other members of the cycle. 2. Use text analysis to determine the manager’s opinion on the matter (since he is filling out the lineup cards). 3. Take the most common opinion from reporting sites.
On a side note, I can see the usefulness of the methodology used in the piece presented. It’s a nice way to globally say what a 5th tier starter would mean, in league terms. With that said, when trying to assess the performance of anyone who actually was slotted to that portion of the rotation it clearly throws out any of that info.
This makes this statement: “He found that number five starters were something like 24% below league average.” completely worthless. So we took all starters, binned them into 5 tiers, and the worst tier is 24% worse than the average? That seems pretty reasonable. If I took all starters and binned them into 5 categories by uniformly splitting them by FIP, I am relatively sure that according to that metric the bottom 20% in FIP will be 24% worse than the average. In fact, for many types of distributions- I could state that with certainty as a property of how I’ve made that split.
Additionally, though meant to compensate for injuries, the metric he is using doesn’t. In fact, it basically can make your ace a 5th starter if he happens to not pitch as many innings as expected and performs poorly when he does (which can tend to happen with injuries). If anything, I would say injuries would have a pretty major effect on slotting guys into these categories due to the use of counting stats like IP and RA.
So… interesting stuff, but I’m not sure it gives a real view into what should really be classified as a 5th starer.
One suggestion would be to select, for each team, the pitcher that was the 5th starting pitcher to start a game for that team (it wouldn’t necessarly be the pitcher that started the 5th game of the season, due to off days some teams have a 4 man rotation for the first few weeks of the season).
That way, the pitcher selected would be the one that the manager selected as his 5th starter at the start of the season (barring injury), and we could then see how those pitchers do when compared to the larger population of pitchers.
I did something similar to what you suggested in response to a 2008 article from BP. It is 2 years old, but the age of the response reinforces my point even more because you can see how guys lower in the depth chart at one point can have an even more pronounced effect going into the future. I contend that your point of “manager know who their worst starter are” is not true. There are many surprises that emerge in the rotation throughout the season.
Jim Baker is his January 4 Prospectus Matchups articles states:
Fifth Starters: Who Cares?
Something I’ve harped on the last couple of offseasons is this: don’t waste your eyeball time on articles about pitchers battling for fifth-starter slots in spring training. It just doesn’t matter. David Pinto does an excellent job of hammering this point home in this piece. His thesis, well-reinforced with the appropriate statistical evidence, is that fifth starters are a solid lead albatross around the necks of the large majority of teams. The real curse of the five-man rotation is not only that the top two guys don’t get more starts, but that fifth starters get so many. On the other hand, they do help hitters generate more outrageous stats, so they are not entirely without purpose.
Using David Pinto’s definition of a 5thstarter,
How do No. 5 starters contribute to a club? By definition, the fifth rotation slot is for the team’s worst starter. Given the current cost of pitching, the ability to put five quality starters in the rotation is difficult. Can these pitchers make the difference between a team winning and losing?
One can believe that the team’s worst starters do not contribute much to a team’s success. However during spring training, General Managers do not know who will perform the worst. Many starting rotations are set incorrectly and those starters who during spring training are battling for fifth starter slots turn out to be productive starters later in the season. After all Carl Pavano opened the season for the Yankees, while Fausto Carmona was battling for that fifth spot with the Indians.
My definition of a fifth starter is anyone who started for an MLB team in 2007 but did not start in one of the team’s first 4 games. I may have missed an injury that caused a starter to rehab but for the most part, I believe that I have captured those starters that were battling for that fifth spot.
There were some notable #4 starters, Brad Penny, Adam Wainwright, Tim Wakefield, Chad Gaudin, Rich Hill, and Greg Maddux but here are the top 21 guys who were not among their teams top 4 in 2007.
VorpStarter
64.0Fausto Carmona
40.7Chad Billingsley
38.2Jeremy Guthrie
34.9Brian Bannister
34.0Dustin McGowan
31.1Shawn Marcum
29.1Jered Weaver
27.1Kyle Kendrick
25.8Scott Baker
25.1Yovani Gallardo
21.1Micah Owings
20.3Steve Tracshel
19.4Tim Redding
18.7Jesse Litsch
16.7Sean Marshall
16.4Chris Sampson
13.5Chad Durbin
13.4Josh Fogg
13.1Justin Germano
12.9Lenny Dinardo
12.7Joel Pineiro
Some other players of note are Matt Garza, Kason Gabbard, Ubaldo Jimenez, Clay Buchholz, Brad Thompson, Jon Lester, Phillip Hughes, and Claudio Vargas.
The truth is, there are several starting pitchers who turn out decent seasons after battling for that fifth spot during spring training. Roles change throughout the season. If a pitcher is getting a long look during spring training he will likely get a shot later.
Ned is afraid to give any young player a chance. He is doing all he can to push Blake DeWitt out of the way and voicing his desire to get a #5 starter. The Dodgers have cheaper, younger options. He could be saying that he wants starter depth, a pitcher that can fill in for any Dodger starter that goes to the DL. Why sign a guy and name him the #5 starter? How about pretending there will be competition or allow for the new pitcher to be a long-reliever in addition to a starter. Teams use more pitchers than 5 in a season and the Dodgers have some good depth. Ned, shop for more depth, not a nebulous designation that usually features replacement-level (or worse) pitching and a lack of reliability.
As a side note, after examining the criteria given for 5th starters I’m convinced there has to be a better way than setting up arbitrary bounds on IP and etc. A team only has so many outs annually, approximately. There’s about 162 games and approximately 27 outs per game (maybe a bit extra due to extra innings). In any event, you can calculate how many outs a team needs to complete its season. And it wants its RA to be above average over the course of the season.
So I would recommend measuring the performance level of a pitcher as their ability to adjust the mean of that distribution as compared to the average performance. So basically, we know the following is globally true:
LeagueERA = (All Runs Against)/(All IP)
Teams want to beat this with their team ERA:
TeamERA = (Team Runs Against)/(Team IP)
We know what the Team IP will be, with good confidence. I would posit that a good metric as to the quality of a pitcher is to measure how good/bad the remaining innings would have to be in order for the team as a whole to have a replacement level ERA (6ish?). We’ll call this the replacement ERA. So then:
replERA = (Pitcher RA + Rest of Team RA)/(Team IP)
replERA *IP = (Pitcher RA + Rest of Team RA)
Rest of Team RA = replERA *IP – Pitcher RA
Rest of Team ERA = (TargetERA*IP – PitcherRA)/(Team IP – Pitcher IP)
So now we know how bad the rest of the team would have to be in order for that team to meet that goal. It takes into account both the IP contribution and the performance contribution, smoothly. The only place where this doesn’t work is if you account for all innings (the denominator will be 0). It should be noted that higher numbers on this scale are a good thing, so we could always shift this to be (replERA – Rest of Team ERA) to make positive numbers bad and negative numbers good. Or whatever. And one might want to increase the magnitude, since some numbers could be small but significant. Those are presentation details. One might also want to use an adjusted run estimator as opposed to actual RA maybe (i.e. like the tRA used in the article). I would assume one could alter the presentation to make this similar to ERA+ but taking IP into account instead of just raw innings.
It’s in many ways similar to what is presented, but without the need to artificially bin guys IP requirements and the like. People with low IP will naturally regress towards replacement ERA in this system, since their impact is negligeable (and it is assumed they’re replaced by whatever is ‘the worst’ in this system). . Good pitchers will drive the metric upwards and really really bad pitchers will drive it downwards slightly, especially if they pitched a lot of innings. But in general, it would give a decent metric as to how far a pitcher pushed their entire team performance away from replacement level. And isn’t that what is actually important for a starter by and large?
So that’s my thought. Take it or leave it.
Note: Replace TargetERA with replERA anywhere noted. I see I neglected to adjust one of those.
I believe that a5th starters should be differently analyzed pre-season and in- and post-season. Pre-season, starters should be analyzed like they are set in team’s plans, like everybody would make 32-33 starts, and additionally “position” 6 – 8th starter should be presented and analyzed. Before season, nobody can know if team will be lucky like CWS in 2005, will they have one starter go down, but other 4 pretty much there for every start, so only one position is empty (NYY in 2009), or if they would need 10+ starters, like NYY most of time 2005 – 2008. So, we can only analyze predicted 5 starters and prepared back-up starters. When we see how many starters were injured, we can analyze each spot (for example, 5th spot was started 28 times during the season, and starters were: 5th 10 times, 6th 4 times, 7th 6 times, 8th 1 time, 9th 1 time, 10th 6 times). Of course, when one starter goes down, everybody gets up by one spot.
McDonald was given the chance to stick in the rotation last year and he blew it. But he is still the best possible choice. He should be given another chance.
Right now it is Stults’ job to lose. So far in exhibition play he’s pitched the best out of the candidates with 2 solid performances. McDonald got destroyed in 2 innings of work against the White Sox, and Haegler couldn’t find his spot yesterday (though he can be really on and off and I don’t see him as a 5th starter). Stults has been extremely consistent in stints in the majors over the last two years with an ERA in the low to mid 4′s. He can also go deep into games. I don’t know why he hasn’t been given the time of the day by the organization because he has done everything asked of him. He’ll have one bad outing and get dropped from the 25 man roster. So is Colletti’s way with young talent. Brad Penny can butcher 8 games in a row and still have a spot on the rotation???
Stults is also a lefty which gives him the advantage over Haegler, McDonald and the Ortizs since the Dodgers already have three RHPs in the rotation. All in all though, you know you are going to have to deal with injuries and need a solid long relief guy since Chad Billingsley and Clayton Kershaw can still be really inconsistent. This looks like a job well fit for McDonald and/or Jeff Weaver, if he makes the team. They can make spot starts to give the old guys (Padilla and Kuroda) extra rest, and start in cases of injury. You then keep the other guys around in AAA in case needed. Stults has earned his spot on the team and if he gets denied again the management is a bunch of dumbasses.
Advantageously, the article is actually the freshest on this worthy topic. I agree with your conclusions and will thirstily look forward to your forthcoming updates. Saying thanks will not just be enough, for the wonderful clarity in your writing. I will immediately grab your rss feed to stay abreast of any updates. Good work and much success in your business efforts!
“Hapy B’day geny.wmv From: narubhai8 Views: 0 0 ratings Time: 01:20 More in People & Blogs”