2007 American League SP Analysis
January 11, 2008 Leave a comment
A couple of weeks ago, I presented the Seidman SP-Effectiveness Model, which took into account a large majority of statistics that deem a pitcher to be effective and weighted them with points based on how important/rare they were. The system is designed to take into account various factors that need to be taken into account in order to level the field of play between those on good or bad teams, those with or without run support, and those either called up/injured or those just plain bad.
Not surprisingly at all, Jake Peavy ended up being first, five points ahead of his competition, but the order of those that followed him turned out to be a bit more surprising than I thought. Everything made proper sense, though, because the pitcher cannot be blamed for his team not scoring for him or not getting decisions in brilliantly-pitched games.
Essentially, my SP-Effectiveness Model answers the question – What would happen if a pitcher was rewarded every time he pitched well and negated every time he pitched poorly?
I also introduced my statistic, the AQS, or Adjusted Quality Start, which extends the general rule of 6+ IP and 3 or less ER to also include games of 7.2+ IP and 4 or less ER. Based on my analysis of innings pitched by starters and the frequency of when they were lifted for relievers, coming one out short of the eighth inning truly merits being allowed to give up that fourth run.
If you have not yet read the NL Article on this same subject, I highly suggest you click the below link – that way you will understand the rubric and reasoning.
To read the NL 2007 SP-Effectiveness article, and see the results, click here.
In this article, I am applying my model to 2007 American League pitchers. Just like the NL, there were some expected results, as well as some initially peculiar results that make sense upon further thought. Additionally, just like with my NL post, I did not apply this to every American League pitcher. Instead, I selected 1-3 pitchers from each AL team. Before the 2008 season begins I will plug every pitcher from both leagues into my system to see who was worst – which is always fun.
I will not explain all of the statistics or points values, since I did that in the previous post on the NL, but I will say that I did consider the fact that AL managers did not have to worry about pinch-hitters. Due to this, I considered making the IP requirements more stringent with the AL, but the fact is that even though they do not need to be removed for pinch-hitters, they are facing an extra offensive player (not a pitcher in the 9th spot). They should, in theory, give up more runs and have just as good of a reason to come out of a game.
Overall, though, only a few more AL pitchers had over 225 IP than NL pitchers and so it was not worth changing. The biggest difference in both leagues was the average IP/game of the selected pitchers. AL starting pitchers accounted for 66.2% of the total IP in 2007, whereas NL starting pitchers accounted for 63.5%. Though the numbers are pretty close, when we are dealing with over 23,000 IP in a league that extra 2.7% equates to approximately 600 IP.
- To view the raw statistics of all the pitchers used, click here.
- To view the list of AL SP used in the order of effectiveness points, click here
Again, if you wonder why certain statistics are used and/or why they were assigned certain points, please read the previous NL article linked at the top of the page.
I do not want to post a table of 28-30 pitchers, so you will have to click the link to view the results spreadsheet, but I will list the top ones below.
- CC Sabathia, +84
- Dan Haren, +76
- Fausto Carmona, +74
- John Lackey, +72
- Roy Halladay, +68
- Johan Santana, +60
- Mark Buehrle, +59
- Josh Beckett, +58
- Justin Verlander, +58
- James Shields, +57
- Javier Vazquez, +57
- Kelvim Escobar, +57
- Joe Blanton, +57
In the National League, the odd ranking was Chris Young, whose barometrical statistics suggested he should have been ranked higher. In the AL, Beckett falls into the same category. The issue here has nothing to do with Beckett’s numbers, but rather the fact that there were other pitchers who were not as lucky as he was in getting run support or solid bullpen help.
Of the players listed above Beckett, both Santana and Haren had 7 tough losses, Buehrle and Lackey had 5 tough losses, Halladay led MLB in IP/gm and CG, and Carmona had more legit wins and less legit losses.
Essentially, there is nothing wrong with Beckett’s 2007 numbers, however there were other pitchers who happened to perform better in certain areas than he did.
The Red Sox had a dynamite bullpen, so going to Okajima or Papelbon was something that just about any manager would feel comfortable and justified in doing, whereas some of these other teams needed their starters to last longer.
No, this system does not take into account any sort of clutch factor, where I am sure Beckett would excel, but it does level the playing field to show which pitchers were the most effective, based on the numbers they individually put up.
Just like the conclusion that was made in the Snell/Zambrano comparison, this is all about consistency. The quality of Josh Beckett’s AQS’s may have been far greater than those of the other pitchers, however they occurred less frequently compared to the same other pitchers. Even though his good-great games may have been astounding, when he was having average or bad games, the other pitchers were still having good-great games.
Beckett had an AQS 67% of the time (20 of 30 starts) while those listed were 73% and higher. This is not necessarily a measure of how good a pitcher was in his good games, but rather how often he was good.
ADJUSTED W-L RECORDS
One of the major reasons we considered Beckett to have been so good this past season was his record. If he was only 15-9, like Dan Haren, there would not have been a Cy Young debate.
That tends to be a problem because, as I will get into in the next category, W-L records do not differentiate between these Cheap Wins and Tough Losses. If we gave every pitcher a Win for each Tough Loss, and a Loss for each Cheap Win, Beckett’s record would not have been 20-7. It would have been 19-8.
There is not a huge difference between his 20-7 and 19-8, but when we do the same for the AL pitchers above him in points, we get the following records: Sabathia (21-5), Carmona (23-4), Santana (19-9), Haren (21-3), Buehrle (15-4), Lackey (22-6).
If we are going to use W-L record as a barometer, and include these Tough Losses and Cheap Wins, all of those above records are either better than or equivalent to Beckett’s 19-8.
Based off of just looking at the Adjusted W-L records, if we were to use that as the barometer for the Cy Young Award or the best pitcher, the debate would not be between Sabathia and Beckett – it would be between Haren and Carmona. I am not saying it should have been between Haren and Carmona, but rather that if we are going to use W-L as an “end-all” statistical solution, we should at least use the Adjusted W-L, or the True W-L.
I described the different types of wins in the NL article but I did not mention the statistic “True W-L Record.” In order to properly evaluate pitchers, W-L records have to be broken down and examined. Some pitchers will get tremendous run support and win games even if they only last 5.1 innings and give up 4-5 runs.
Then there are some who will go 6.2-7.1 innings, give up 2-3 runs, and lose. After separating these Cheap Wins and Tough Losses from a W-L record, we are left with a record of legitimate wins and losses – games that a pitcher deserved to win or lose based on performance.
A legit win occurs when you record an AQS and win, and a legit loss occurs when you do not record an AQS and lose.
The difference between True W-L and the Adjusted W-L I used in the Beckett comparison is that the True W-L does not include Cheap Wins or Tough Losses. True W-L only includes games in which the pitcher recorded a win or loss when either decision was merited.
You can see these True W-L Records in the raw statistics spreadsheet, but I have listed the best ones below. In parenthesis next to the True W-L Records are the Actual W-L Records.
- Dan Haren, 14-2, (15-9)
- Kelvim Escobar, 14-3, (18-7)
- Fausto Carmona, 18-3, (19-8)
- Josh Beckett, 17-5, (20-7)
- Chien Ming-Wang, 17-5, (19-7)
- CC Sabathia, 17-5, (19-7)
Again, we see that if win-loss was to be the “end-all” tool to evaluate a Cy Young Award or the best pitchers, Haren and Carmona would be atop the list.
BOB GIBSON, GREG MADDUX, PEDRO MARTINEZ
For fun, I decided to plug some legendary seasons into my system to see what the end results were. Yes, it is impossible to perfectly compare a season from 1966 to one from 1996, but still it is interesting to see how they would rank. To do this, I took the 1968 season of Gibson, the 1995 season of Maddux, and the 2000 season of Martinez. The points results for the three were:
- Bob Gibson, 1968, +178 pts
- Pedro Martinez, 2000, +104 pts
- Greg Maddux, 1995, +97 pts
And there you have it. By the middle of February I should have a spreadsheet/PDF made up of all NL and AL pitchers plugged into this effectiveness model. That way we can see who were the absolute worst as I am sure we will find some surprises and unexpected names there.
The biggest surprises to me in both leagues, in a positive turn, were Bronson Arroyo and James Shields.
The most unexpected finishes were Beckett and Chris Young, as I predicted they would be higher.
An interesting thing to look at is how players on the same team ranked next to each other. In the NL, Zambrano is widely thought of as the #1 of the Cubs, yet Ted Lilly finished much higher. In the AL, Kazmir is definitely thought of as the Rays ace, yet Shields ranked 9th out of the pitchers used here, and Kazmir finished 20th.
And, since the Yankees have to be stubborn, both Pettitte and Wang tied in effectiveness points.
This model is not the end-all solution to determining who the best pitchers are in a given year, but it is a darn good predictor and estimator since it equalizes the field of play and makes sure it is known that you do not have to be on a great team to be a great pitcher or have a very effective year.
This measures a specific season, where some players may be better than others, even if they are nowhere near better in a retrospective look at their careers.