Numbers for the Numbered Starters

One of the more contrived areas of prospect analysis, in my opinion, is throwing a numbered starter grade on a pitching prospect. Scouts are the most consistent offenders of the trend: you’ll see a variation of “future no. 3 starter” in a lot of scouting reports. In a sense, it has become more philosophical — “a future ace” has non-contextual meaning, and so too does “middle of the rotation” or “back-end guy.”

I’ve advocated that we must think about prospects in terms of the WAR they will produce, and for pitchers, we must consider their potential strikeout, walk, home run and groundball rates (essentially, xFIP). Since the numbered starter grade has so much popularity behind it — I literally get “will X prospect be a #2 or #3” question every week in chats — it makes sense to put those designations in actual context, given the peripheral statistics we want to consider with prospects.

With that said, I will reinforce what Marc Hulet wrote earlier this year: essentially, the idea of a “#5 starter” doesn’t exist. Teams don’t use a #5 starter — they use a variety of replacement guys, generally — so if you’re projecting a prospect out as a #5 starter, you’re doing him a disservice. He’s either a reliever, a replacement-level player, or worse, a career minor leaguer. So, in this piece, we’ll be concerned with the 2010 statistical definitions of what #1-4 starters were. To get the appropriate number of players, I used 130 innings as a cut-off. This gave us 115 players, all of whom (except Brian Duensing) spent the vast majority of their season in the rotation.

Control (BB/9)

#1 Starter: 0.76 – 2.41 BB/9.
#2 Starter: 2.44 – 2.92 BB/9.
#3 Starter: 2.95 – 3.41 BB/9.
#4 Starter: 3.47 – 4.74 BB/9.

Cliff Lee was the king of command this year, with a walk rate almost 30% better than his next closest competitor, who just happened to be Roy Halladay. Guys like Shaun Marcum and Doug Fister took the next step this season by exhibiting ace-level command. It’s important to point out that BB/9 doesn’t always measure command, as a guy like Kevin Slowey can rank top five in the Majors in the category, but by missing on 21 pitches that ended up in the cheap seats, his xFIP was just 4.48. On the opposite side of the ledger was C.J. Wilson, who succeeds despite his proclivity for walks. He has success pitching consistently low in the zone, so he’s an example of a guy with command, but not control.

Swing-and-Miss Stuff (K/9)

#1 Starter: 10.95 – 7.86 K/9.
#2 Starter: 7.84 – 6.87 K/9.
#3 Starter: 6.86 – 5.44 K/9.
#4 Starter: 5.43 – 3.80 K/9.

What Cliff Lee was to the walk column, Brandon Morrow is to the strikeout column. While his control leaves something to be desired, Morrow’s raw stuff is off the charts. He was one of 13 starters this year that struck out one batter per inning. Since this happens so often in the minor leagues, I think we might forget just how rare it is for a pitcher to accomplish it at the highest level. Minor league strikeout kings like Wade Davis aren’t necessarily strikeout artists in the bigs. Without strikeouts, it’s a tough path to success: Carl Pavano had the most success with a terrible K/9, but his command had to reach career-best levels to achieve it.

Movement (GB%)

#1 Starter: 64.1 – 49.6 GB%.
#2 Starter: 49.4 – 44.9 GB%.
#3 Starter: 44.7 – 39.8 GB%.
#4 Starter: 39.5 – 28.3 GB%.

Movement isn’t quite described by GB%, but I think it does a fairly good job: good movement leads to weak hits, which keep the ball in the park. Tim Hudson was the only person this season to eclipse 60%, and he did it in a big way. The power of ground balls is maybe best exhibited with Ricky Romero, who was drowning in the minor leagues, looking like a bust, as a flyball pitcher in 2006 and 2007. Then, the former first rounder reinvented himself, and started focusing on ground balls. Success soon followed, and this season, he was one of 31 players to reach four wins above replacement. Staying east, Daisuke Matsuzaka‘s inability to record groundball outs has been one of the largest factors in his failure to live up to the hype in Boston.

Overall Rate Value (xFIP)

#1 Starter: 2.92 – 3.80 xFIP.
#2 Starter: 3.81 – 4.18 xFIP.
#3 Starter: 4.19 – 4.51 xFIP.
#4 Starter: 4.56 – 5.62 xFIP.

Now, these numbers are a little misleading — you wouldn’t call negative WAR guys like Javier Vazquez or Scott Kazmir “Number Four Starters.” Not in terms of how they’ve performed this season. While the scale falls off at the end, I think it does a good job at the beginning: I struggle to argue with calling anyone that had a 3.80 xFIP or better this season an “ace”. I laughed when I saw Trevor Cahill and Bud Norris back to back in the xFIP column — it’s a perfect description that there isn’t one way to succeed as a pitcher in the big leagues. Some pitchers, like Josh Johnson of the Marlins, can succeed in all three of the columns above. But whether it’s standing out in one category (Pavano), or consistency across all three (Randy Wells), there are a lot of paths to the same destination.

These numbers aren’t hard-and-fast cut-offs for discerning whether we’re going to call John Lamb a future number one starter, or a future number two. They are useful numbers for us to store mentally to put the distinctions that others insist on using into context. And, they serve as a nice tribute to the aces of 2010: Roy Halladay, Francisco Liriano, Adam Wainwright, Josh Johnson, Tim Lincecum, Cliff Lee, Felix Hernandez, Jon Lester, Mat Latos, Yovani Gallardo, Cole Hamels, Roy Oswalt, Jered Weaver, Ricky Nolasco, Hiroki Kuroda, Brandon Morrow, Derek Lowe, Dan Haren, Wandy Rodriguez, Justin Verlander, James Shields, Ubaldo Jimenez, Jaime Garcia, Jhoulys Chacin, Ricky Romero, Zack Greinke, C.C. Sabathia, Clayton Kershaw.




Print This Post





34 Responses to “Numbers for the Numbered Starters”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Paul says:

    I don’t really see the labeling of someone as a “#x starter” as something that needs upheaval.

    Vote -1 Vote +1

    • Bryan Smith says:

      And I realize that it will never be thrown away. It has its place in baseball, and in scouting reports, and always will. So, rather than continue to moan about it, I decided to add some literal context to it.

      Vote -1 Vote +1

  2. BS says:

    Damn, Joe Blanton could be considered a “#2 starter”. Blows my mind.

    Vote -1 Vote +1

    • don says:

      Hamels has the WORST walk rate in the Phillies rotation at 2.63 BB/9. I find that sort of hilarious.

      I think Blanton may have come back too soon from the oblique injury. He was terrible in the first half, but since the ASB he’s been striking a lot more guys out and he still doesn’t walk many, which is what he’s done most of the time since he came to the NL. He doesn’t keep the ball on the ground so he’s always going to give up some home runs, but he’s nowhere near as bad as he looked the first half of this year.

      Vote -1 Vote +1

  3. Jamie says:

    Halladay, Hamels, Oswalt.

    all 2010 staff aces.

    Vote -1 Vote +1

  4. Dan says:

    Any system that refers to Derek Lowe as an ace needs serious reconsideration. That cutoff needs to be below 3.5 or lower, IMO… I don’t believe that there are 28 “aces” in baseball.

    Vote -1 Vote +1

    • Bryan Smith says:

      But by literal definition, there are 30, no? If we have 150 rotation slots, there are 30 No. 1 starters. Like I said in the opening paragraph, surely the word “ace” has non-contextual meaning — and that is something better designed for Carson Cistulli to explore — but I wanted to show what it meant in 2010 at the most literal level.

      Vote -1 Vote +1

      • blackout says:

        This discussion has been had many times, Bryan. You can argue that there are 30 #1 starters by definition while designating only an elite sub-group of them as *aces*. Bronson Arroyo is the Reds #1 starter, but is not an ace by any traditional defintion or your newfangled one.

        Vote -1 Vote +1

      • Bryan Smith says:

        I realize it has, blackout. But I haven’t seen it applied to 2010 numbers anywhere yet (if it has, let me know, and I’ll link to it), so I thought it had value in bringing back up.

        Vote -1 Vote +1

      • blackout says:

        I’m not sure I see the value in measuring what I assume you’re referring to as 2010’s *aces* rather than the non-contextual version when the battle over #1 starter vs ace has presumably long been settled. I’d be wary of even seeming to conflate the two.

        Vote -1 Vote +1

      • blackout says:

        And I should add that I enjoy the exercise for the most part, and don’t want that quibble to overshadow what I think is a worthwhile discussion.

        Vote -1 Vote +1

  5. adohaj says:

    Ricky Nolasco an ace!??

    Vote -1 Vote +1

  6. rickie weeks says:

    So the boundaries in each category are determined by finding the top 25% of qualifiers, then the next 25%, and so on? This assumes there are the same number of “aces” as #4 starters in MLB, which I don’t agree with. I’d think the distribution is more normal, with a #3 representing the mean.

    Some people think #1 starter means one of the top-30 starters in MLB, whereas many others think there are significantly less true #1’s than teams in the league. I tend to agree with the latter. I mean, would you really say Wandy Rodriguez or Jaime Garcia could be the ace of a championship caliber team?

    Vote -1 Vote +1

    • rickie weeks says:

      Or Derek Lowe…

      Vote -1 Vote +1

    • Bryan Smith says:

      No, I wouldn’t. Ace is a term with non-contextual meaning, and if you talk to a scout, it means a lot of different things: a guy with 2-3 plus to plus-plus pitches that is capable of eating a lot of innings, commanding the zone, etc. But I do think a No. 1 starter, if we are being literal, implies there are about 30. And I think this piece, in a way, is designed to make you guys think how silly it is that Wandy Rodriguez or Derek Lowe were No. 1 starters this year. It’s working.

      Vote -1 Vote +1

      • rickie weeks says:

        Yea, I guess it comes down to figuring out exactly what the reader wants to know. But if the point was to illustrate the difference between a literal #1 (or top-30) starter and an “ace” or “#1” in the scouting context, then it’s definitely serving its purpose.

        Vote -1 Vote +1

      • mattpoin says:

        This makes sense to me. We should all be familiar with the idea that all #1 starters are not equal, but would still be the first guy to the mound in a season if all 30 of them were divided evenly between the teams. It’s like the NBA draft, for example–the guy who goes to the Lakers as the last pick of the first round (assuming they Lakers did not have an earlier pick as well) is still a #1 pick for the team. He just isn’t John Wall.

        Vote -1 Vote +1

      • Anon21 says:

        Look, that would be fine, except that Lowe is worse than two pitchers on his own team. So I’m back to thinking you need to recalibrate your system.

        Vote -1 Vote +1

      • blackout says:

        FWIW, having seen Wandy Rodriguez pitch a couple of times during his second half resurgence I have a lot less problem with him being listed as a #1, in any sense, than most. Excellent pitcher.

        Vote -1 Vote +1

      • J-Doug says:

        So has this question been answered? “So the boundaries in each category are determined by finding the top 25% of qualifiers, then the next 25%, and so on?”

        Did you select on your dependent variables, Bryan? Or did you make a determination based on who was #1, #2, etc… for their respective teams?

        I would be more interested in a study that did the latter, although I know that wouldn’t be easy. That said, great work, shared it on my website. I have my own take on it that I’ll be posting when the postseason is done.

        Vote -1 Vote +1

  7. Greg Maddux says:

    Quick glance at the player pages indicates only three of the starters are a “#1 Starter” in all four categories.

    Felix Hernandez
    Roy Halladay
    Adam Wainwright

    Woof,
    Mad Dog

    Vote -1 Vote +1

  8. Greg Maddux says:

    Oops, King Felix actually just misses the #1 BB rate bucket.

    Only Doc and Waino!? Handful of others are very close.

    Vote -1 Vote +1

  9. woodman says:

    Pretty cool that the Blue Jays have experts in the 3 categories: a strikeout artist in Morrow, a control artist in Marcum and a groundball artist in Romero. I’d have to say strikeout guys are the most fun to watch.

    Vote -1 Vote +1

  10. Sam A says:

    James Shields and ace?

    Vote -1 Vote +1

  11. Mr Punch says:

    I don’t think it’s true that there’s no such thing as a #5 starter – I can think of a number of teams (all pretty good) that have had a set 1-5 rotation or made only one change in a season (e.g., ’04 Red Sox replaced Kim with Arroyo). The Sox actually had an intended 1-5 this year, and a clear #6; they had 157 starts by these six despite DL time for (I think) four of them.

    I’m on the side of distinguishing ace from #1. Lowe’s the perfect example – no one has ever really thought of him as an ace. On the other hand, Curt Schilling was – although he was not the #1 for much of his career. Beckett was Boston’s opening day starter.

    Vote -1 Vote +1

  12. Omar Fired says:

    Out of your list of “aces”, which name isn’t bolded? Trying to get subliminal with the nonexistent Cy Young voters reading this piece?

    Vote -1 Vote +1

  13. Krog says:

    I like this article, but all of the metrics seem to be based on stuff while ignoring durability. There is a big difference between 130 IP and 230 IP. The ability to eat up innings is a huge responsibility for a Ace. Is there a way to include this in your next article?

    Vote -1 Vote +1

  14. RJodoin says:

    Krog, I agree that durability matters, but it’s often not in the hands of the pitcher (so to speak!), especially in the NL. A stat on “number of times being replaced mid-inning” might do the trick. I bet you someone has already invented it.

    Also, Bryan: does the league the pitcher is in matter here? Not that I want to create extra work if it’s statistically insignificant, but I would wager that the typical AL 9th batter hits better than the typical NL pitcher.

    (And yes, the Jays rotation will surprise many over the next few years.)

    Vote -1 Vote +1

  15. Phantom Stranger says:

    I would ascribe better movement for a pitcher in producing the fewest line drives, not ground balls. You just have to pitch low in the strike zone with sink to produce groundballs. It does not capture the important lateral movement which produces easy pop flies. Anyone know which pitchers gave up the fewest line drives as a percentage of hits given up? I bet it is Halladay.

    Vote -1 Vote +1

    • Nathaniel Dawson says:

      So that means line drives go for hits more often against Roy Halladay than they do for other pitchers?

      Vote -1 Vote +1

  16. Nathaniel Dawson says:

    What’s the point of including groundball rate in the analysis? GB% has a very low correlation to success at preventing runs. BB rate and K rate have much higher correlations, to the point that GB rate is trivial in comparison. It really tells you little to nothing about whether a pitcher is good at preventing runs. And since you’re using groundball rate, why not use flyball rate as well, since a higher flyball rate also shows a correlation (although very weak as well) to fewer runs scored?

    Vote -1 Vote +1

  17. Lorenzo says:

    I’m a little late to the discussion, but the article did a good job of measuring what’s measurable. One of the unmeasurables that is a key for an ace is performance in a high pressure must-win game. Both Halliday and Lincecum got their first postseason test, and did rather well. There’s no way to quantify that, you can only seek a consensus of opinion, and how accurate is that?

    Vote -1 Vote +1