Tuesday, November 11, 2008

A Look at Starting Pitching

Tampa Bay RaysThere has been a lot of talk this season about StatCorner and tRA, primarily by the guys at DRaysBay. I read articles explaining tRA such as this and this and compared StatCorner to advanced stats from Baseball Prospectus, the Hardball Times, and FirstInning. I've come away very impressed, so impressed that from now on player links will be to their StatCorner page. With this in mind, let's take a look at the starting pitching in the Rays organization using their StatCorner numbers.

First, here are the definitions of the stats used from the StatCorner Glossary:

xIP: Derived from xOuts, "xOuts stands for the expected number of outs a pitcher has generated based on his defense and park neutral outcomes." This is how much they pitched.

tRA: "tRA involves assigning run and out values to all events under a pitcher's control and coming up with an expected number of runs allowed and outs generated in a defense and park neutral environment. tRA is on a R/9 scale and does not involve any regression of the rates so while it should be more useful at determining a pitcher's true talent level, the best method for pitching projection is to use tRA*, the regressed version of tRA." This is how well they pitched.

pRAA: "Stands for pitching runs above average. Using tRA as the benchmark, the formula is (lgTRA * xOuts / 27) - xRuns." This is how much 'value' they added (or cost) their team based on the quality and quantity of their pitching.

Notes: StatCorner only covers down to full-season A ball, so no Hudson Valley, Princeton, DSL or VSL players. Data doesn't include innings pitched as a reliever, only as a starter. Age is as of today. Spot starters aren't included. Pitchers are listed by level, so for example Hellickson and Price appear on multiple lists.

Tampa Bay Rays (MLB):
Shields, James26.8MLB212.63.8424.4
Sonnanstine, Andy25.6MLB198.83.8223.2
Garza, Matt24.9MLB181.24.3211.1
Kazmir, Scott24.8MLB147.74.712.6
Jackson, Edwin25.2MLB174.85.68-15.7

Sonnanstine was nearly as valuable as Shields and yet there are those who want to trade Sonnanstine and keep Jackson. When Price is added to the rotation next year Jackson clearly needs to be the one to go, either to the pen or in a trade. Garza and a (hopefully) healthy Kazmir will round out the best rotation in the majors.

Durham Bulls (AAA):
Talbot, Mitch25.0AAA164.83.5423.5
Cummings, Jeremy32.0AAA81.73.858.9
Davis, Wade23.2AAA52.63.557.5
Niemann, Jeff25.7AAA127.94.406.1
Price, David23.2AAA18.74.221.3
Hendrickson, Ben27.8AAA141.55.44-9.6

Similar to Sonnanstine vs. Jackson, in the debate between Talbot and Niemann, the wrong one (Talbot) usually loses. I know Niemann was a higher pick and is a huge, projectable guy, but Talbot was much better in 2008. If we have to trade one, I'm hoping it's Niemann. Davis will get a full season of AAA next year, barring a callup. Cummings and Hendrickson are just filler.

Montgomery Biscuits (AA):
Davis, Wade23.2AA105.64.217.0
Price, David23.2AA54.03.776.2
McGee, Jacob22.3AA74.64.433.1
Rollins, Heath23.4AA23.04.341.2
Hellickson, Jeremy21.6AA73.64.710.8
Houser, James23.8AA88.94.800.0
De los Santos, Richard24.4AA70.64.91-0.8

Davis was better in AAA. Hellickson and Rollins should start the year in AA, Houser in AAA. McGee will miss most of the year, and De los Santos is a free agent.

Vero Beach Devil Rays (A+):
Hellickson, Jeremy21.6A+76.72.3818.9
Rollins, Heath23.4A+123.83.8510.1
Price, David23.2A+33.32.667.2
Morse, Ryan25.4A+102.94.69-1.1
Walker, Matthew22.2A+41.95.35-3.6
Mann, Brandon24.4A+122.25.46-11.8

Hellickson and Price just dominated, and Rollins was solid. Walker has a make or break year coming up following the suspension. Mann and Morse are filler.

Columbus Catfish (A):
Darcy, Jesse23.3A102.54.386.4
Cobb, Alexander21.1A129.45.13-2.7
Flores, Brian23.8A108.85.60-8.0
Ragan, Jason25.8A105.15.72-9.1
Hall, Jeremy25.1A94.65.91-10.2
Gibson, Glenn21.1A57.46.61-10.7

This is why I'm not as high on Cobb as most people. He is young, but he'll need to improve at Charlotte next year to convince me he's one of our top pitching prospects. Darcy did well after beginning the year in the pen, but because of age he needs to move quickly. Ragan was released. Hall and Flores are filler, actually Flores may have a future out of the pen as a LOOGY. Gibson was so bad he lost his spot in the rotation, was even worse out of the pen (8.07 tRA), and was demoted to Hudson Valley. As I wrote in early September, forget 2008 happened and start over in Bowling Green in 2009.

To summarize, with the addition of Price we will have the best starting rotation in baseball to begin 2009. But, as the need arises, we have plenty more talent on the way. The problem will be finding room on the big league roster, but what a great problem to have. As injuries occur and players are lost to free agency, spots will open up. And not all of our prospects will make it because of injuries, trades, or just hitting their ceiling in the minors. With any kind of offense, the Rays will be a formidable foe for years to come in the AL East.

Make sure to visit StatCorner. Excellent site, loads quickly, tons of information. Only one caveat, make sure you have lots of time, it's addictive.


  1. I still like Cobb. He was outstanding in every month except July. He got just smacked around in july. What was his tRA by month? July was the only month where his BAA was higher than .212 so im wondering why his tRA as a whole is so high. Maybe his BABIP against reveals he was lucky. But how could he be lucky in 4 months and unlucky (or normal) in July?

  2. Well tRA isn't available by month, so take a look at FirstInning on Cobb (I can't put charts in comments or I would do it for you). FIP was 4.75, so tRA of 5.13 isn't much different. Normal BABIP is ~.300, a little higher in A ball. Cobb's BABIP's were April=.256, May=.240, June=.218, July=.313, August=.222. So they only 'normal' BABIP was August, and he got torched. tRA is saying if he had normal BABIP for the season (his was .248, which isn't sustainable long term) he would be average. Now a few guys can maintain low BABIPs, but can Alex? In 2007 his was .312, almost identical to August's .313. So there is reason to believe he will return to normal next year (overall) and put up average numbers. Look at his last 10 starts in 'Game Log', six were bad, especially July 26th. On the plus side he is young and I think he may have just wore down, he went from 81.1 IP in 07 to 139.2 in 08. I'm not writing him off by any means, but 08 was average. Hopefully only growing pains.

  3. I think Cobb missed time with injury(in July, I belive), and he was very very impressive on April, May, and June, and pretty good in August. So I don't think it's out of the realm of possibility that Cobb was bothered by the injury for a lot of the month(he didn't make a start between July 9th and July 20th). Now, I'm not sure what the injury was, so this is somewhat just conjecture. But if you cheat and remove July, then his line for the season(in 118.2 innings) would be: 8-4 2.35 ERA 82 SO 30 BB, plus a lot of ground balls. So I think there's a lot to like with Cobb.

  4. And in re: tRA, this isn't aimed particularly at you, Doug, but this has bothered me ever since people started using it. Andy Sonnanstine's tRA was 3.84, James Shields' was 3.82. You said Sonnanstine was "nearly as valuable" as Shields, which is either improper word usage or poor logic. James Shield's ERA was about .75 less than Sonnanstine's. He allowed less runs in more innings, ergo he was more valuable. The point of pitching is to not allow runs, and Shields did a better job of that. If people want to use tRA as a predictor of future success, that's fine. I encourage that. There's absolutely nothing wrong with saying that Sonnanstine's process is more like to garner positive results than Edwin Jackson's process is. But to look back and say that Sonnanstine was as valuable as Shields based on tRA would be a mistake.

  5. Always feel free to question my logic. I figured you'd argue Cobb, but Sonnanstine? Correct me if I'm misunderstanding (crying baby is mindnumbing), but...

    Kevin: "If people want to use tRA as a predictor of future success, that's fine. I encourage that. There's absolutely nothing wrong with saying that Sonnanstine's process is more like to garner positive results than Edwin Jackson's process is. But to look back and say that Sonnanstine was as valuable as Shields based on tRA would be a mistake."

    Article: "tRA is on a R/9 scale and does not involve any regression of the rates so while it should be more useful at determining a pitcher's true talent level, the best method for pitching projection is to use tRA*, the regressed version of tRA."

    I used the "true talent level", not the "projection". Again, enjoy the debate, but not seeing my mistake, as far as I currently understand tRA and it's derivatives.

  6. I guess my point was that the word "valuable" was used wrong. A pitcher's value in the sense it was used is based on results. James Shields was more valuable last year because he allowed fewer runs in more innings than Andy Sonnanstine. Now, because of the similarity, you can argue that the only reason their ERAs were different was because of luck(I'd argue that, too, but that's a different story).

    In my opinion, the conclusion to be drawn from that is not that they were equally valuable last season, but that Sonnanstine's luck has a good chance at regressing next year, resulting in an improved ERA.

    tRA, as I understand it and take it to be, is about process. If the process is good(that is, you limit walks, throw strikes, get groundballs, etc), your tRA will be good. However, a good tRA doesn't guarantee good results, it just increases the likelihood of results.

    I measure value by result(which would be ERA), but I base future projections at least in part on process(which would be tRA).

    Example: A career .250/.300/.450 hitter randomly puts up a .400/.500/.600 line in 2008. However, his BABIP was ridiculously high, a higher-than-usual number of his fly balls left the park, and it's pretty obvious the year was a fluke.

    On the flip side, a career .330/.400/.550 hitter put up a line of .325/.390/.545 in 2008. His BABIP was normal, his HR/FB was in line with his career average, etc.

    In this particular case, Player A was more valuable in 2008. But if I had to pick a player for 2009, I'd pick Player B.

    It's not a perfect example, but I think it demonstrates what I'm talking about.

  7. I'm willing to concede "valuable" as it involves many variables (IP, tRA/ERA, contract status, health, age, etc.), but what I mean is rate times innings = value. Sonnanstine pitches a lot of innings at a good rate (better than Shields per tRA), so he's really valuable. I guess my comment was more towards the Jackson thing than Shields, clearly (big pic) Shields is more "valuable" to the org. Are you disagreeing with my prior post, or are we just arguing past one another re tRA?

  8. I definitely agree in regards to Sonnanstine/Jackson, but I still disagree about Cobb. I think he's a pretty good prospect.

  9. So, are we in agreement on tRA (= good thing) or not? Glad we are agreed on Jackson and Sonnanatine.

    On Cobb, I'll stand by my comments in the post. But as I said, he may have been tired, so the late June/July blowups might mean nothing. Feel free to show me why I'm wrong, but take a look at his HR/PA (up x 3 from 2007), his Home/Road splits (he won't be pitching in Columbus ever again, luckily), and tell me why he's better than Barnese or Moore, or Suarez or Cruz, or others. See FirstInning.com for the stats. Guess I just think he will get passed by them.

  10. ERA does not measure a pitcher's results. It measures team earned run prevention with a given pitcher on the mound, which is a completely different beast.

  11. Graham - Exactly, otherwise we wouldn't need tRA, FIP or DIPS to evaluate a pitcher's performance (not equating them in value). If there's anything you'd like to add about your stats, feel free. Hope I defined them properly, didn't want to get too technical.

  12. You did a good job explaining it, Doug. As far as I'm concerned, the less technical the explanation the better. The ideas are simple even if the mathematics sometimes becomes hard.