IBA Ballot: Cy Young

Presented below is my ballot (and some justification) for one of the categories in the Internet Baseball Awards hosted at Baseball Prospectus. I’m just one person, and the whole point of having a vote like the IBA is to get a wide variety of (intelligent) perspectives, and so I will not feel in the list bit slighted if you don’t give a flip about this.

In the AL, let’s get it out of the way upfront: it comes down to Cliff Lee and Roy Halladay. We’ll get back to them in a minute.

For the rest of the ballot, there is a pack of starters within ten RAR of each other that I would consider: Jon Lester, John Danks, Daisuke Matsuzaka, and Ervin Santana. Lester leads at +65 RAR, and while his peripherals aren’t as strong as his actual RA, that’s true for the whole group except Santana, who is last at +55 RA, and who is still only about even with Lester in eRA and dRA. Thus, I give Lester the third spot.

Matsuzaka’s odd season has been well-documented; the stats I list don’t capture it, really, although the fact that just 48% of his starts were quality hints at the issues. I give Danks the edge over both Daisuke and Santana.

That leaves the question of relievers; it will come as no surprise to sabermetrically-inclined readers that I am less than impressed with Francisco Rodriguez as a Cy Young candidate. If any reliever deserves that type of recognition, it is Mariano Rivera. He pitched two more innings, with a RA over one run lower, a RRA almost two runs lower, an ERA about .8 runs lower, an eRA almost two runs lower, and a dRA over one run lower. It’s a clearly superior season by any context-neutral measure you’d like to look at. WPA? Rivera leads him +4.47 to +3.33. I would also put Soria, Nathan, and Papelbon ahead of Rodriguez among AL closers. None of that is said to belittle K-Rod; he may not have had a great season, and he may be grossly overpaid in short order, but he’s still quite good, he’s only 26 even though it seems as if he’s been around forever, and he has a fine track record. He’s wasn’t a worthy Cy Young contender in 2008, though.

Lee v. Halladay. Upfront, yes, I am an Indians fan, although I consider myself well below average on the partisan scale. Also upfront, whatever conclusion I draw says nothing to very little about who I feel is a better pitcher--it's solely about who was a more valuable pitcher in 2008. Going backwards or forwards in time, I would take Halladay in a heartbeat.

Halladay’s big advantage up front is 23 more innings; Lee counters with a .42 run edge in RA, and .19 runs in ERA. In terms of eRA, Halladay bests Lee 3.23-3.08, and they are essentially even in dRA (3.33-3.36, advantage Lee). Win-loss record, evaluated superficially against team W%, favors Lee, .915 to .643. I don’t want to go any deeper than that, since it really has no bearing on my choice, but it does tell the story of why Lee will win the actual award.

In terms of value against baseline (based on RA), I have Lee as +80 (replacement)/+51 (average), and Halladay at +77/+44. So my initial inclination is to give the edge to Lee.

One point that has been offered in Halladay’s favor is that his average opposing batter was better. According to Baseball Prospectus’ figures, the average Halladay opponent hit .266/.342/.425 while Lee’s hit .262/.330/.405. Plugging those lines into ERP, the differences are significant--Halladay's opponents had a 5.04 RG versus 4.60 for Lee (for reference, the AL average was 4.78). Of course, as others have noted, this is based simply on the performance level of those batters this year, not their true talent.

However, I think that going too deep into quality of opponent leads to some tricky issues about what constitutes value. You may consider what follows to be a case of paralysis by analysis, but so be it.

If we have two pitchers, one in a tough division (like the AL East) and one in a weak division (like the AL Central), we would expect a random pitcher from the first team to face a tougher average opponent than a random pitcher from the second team. If we assume that the two pitchers’ true talent is the same, and that they each have the same degree of “luck” for lack of a better term, we would expect the second pitcher to allow less runs, win more games, etc. despite the fact that he has pitched exactly as well as the first pitcher.

You can choose to adjust for this--but you can also argue that from a strict perspective of value, those extra wins for the second team are every bit as real. Had the Blue Jays and the Indians been competing against each other for the wildcard, Toronto would not have gotten bonus points in the standings for facing tougher opponents. One can thus argue that Halladay shouldn’t either. For lack of a better term let’s call this the “actual team wins” argument.

Of course, that raises the issue of baseline. I assume that a replacement level starter allows runs at 125% of the league average--but if he faces opponents that are 5% tougher, we would expect such a pitcher to allow something more like 130%. And thus one can also argue that Halladay should be compared to a higher (in terms of RA) replacement level, since what we are trying to measure is the marginal difference between Halladay and a scrub in his circumstances. Inserting a theoretical replacement level into the discussion is a point against the “actual team wins” argument made in the last paragraph. But I don’t consider it a deathblow to that argument.

Anyway, the Indians’ opponents, weighted by games, had a park-adjusted R/G of 4.70 (I did a general park correction, not a specific one based on where the game was played, which would be preferable). The Blue Jays’ have a R/G of 4.79. Weighting by innings, Lee’s opposing teams had a R/G of 4.62, Halladay 4.81. Lee faced opponents at 98.3% the R/G of his team average, Halladay 100.4% of his. If you compare Lee to a baseline .983 times what I was initially using (the league average times 1.25) and Halladay to the same times 1.004, you get each at +78 RAR. (This approach accepts the premise that strength of opposition should only be adjusted for by comparing a pitcher to his team, not to the league).

I don’t know why the results of the opposing slash lines as given by BP and the weighted team R/G given here vary so much. Obviously, one takes into consideration the actual identities of the batters and one does not, but I don’t think that is necessarily a selling point for BP’s approach. What if you studied the opponents and see that more good left-handed hitters got a day off against Lee? If that were the case (I do not know it to be), that would not be something that I would want to hold against his value. And along the same lines, it should be noted that the opposing hitters’ stats don’t account for platoon differences--maybe more mediocre right-handed hitters get an opportunity to play against a tough lefty and drive the pitcher’s composite opponent numbers down.

Again, I’m not saying that either of those scenarios is the case--it would take a lot of digging to figure it out, and I don’t want to get that involved here. But I do not think it is a given that the slash opponents’ figures from BP are the 100% correct choice if one looks to adjust for quality of opponent.

I have already droned on and on about this and I have not even mentioned a key factor like the quality of team fielding. It appears from team DER that this is a slight impairment for Lee vis-à-vis Halladay, but you know how fielding statistics are. The real takeaway from all of this is that there is a tiny margin separating these two. I think either would be an acceptable choice. I am going to take the coward’s way out and choose Lee because that is what the consensus opinion of the masses will be. But if you want to make a case for Halladay, I won’t put up a fight.

1) Cliff Lee, CLE
2) Roy Halladay, TOR
3) Jon Lester, BOS
4) RP Mariano Rivera, NYA
5) John Danks, CHA

The National League race is closer in RAR but in the end, my choice was clearer. Off the bat, no reliever is on my radar; Hong-Chih Kuo led at +27 RAR, followed by Carlos Marmol (+24) and Brad Lidge (+21). Lidge would get the biggest boost from leverage credit, but it’s not enough to get him into contention.

Again, there are two starters that stand out: Tim Lincecum and Johan Santana. For the other spots, Ryan Dempster, Cole Hamels, Dan Haren, and Brandon Webb are my pool of possible choices.

While I would least want Dempster going forward, he did have a fine season, +58/+32, with solid peripherals (3.14 RA coupled with a 3.38 eRA and 3.60 dRA). Cole Hamels is at +27/+56, but he did no better in peripherals (3.45 RA paired with a 3.62 eRA and 4.10 dRA).

Webb and Haren are an interesting pair since they are teammates. Webb pitched ten more innings, but his RA was .18 runs higher, so Haren beats him in RAR +54-+52. Webb was better in eRA (3.20 to 3.51), but dRA favors Haren (3.25 to 3.42). They faced essentially the same quality of opposing batter; .255/.327/.398 for Haren, .254/.325/.393 for Webb. You can basically flip a coin, and mine comes up Webb.

Finally, Lincecum and Santana. Santana pitched seven more innings with a RA .10 higher, an ERA .02 higher, an eRA .64 higher, and a dRA 1.11 higher. Since Lincecum rates even with Santana in the value measures (+72/+43 versus +71/+42) and thrashes him in peripherals, I think he’s the clear choice in the end:

1) Tim Lincecum, SF
2) Johan Santana, NYN
3) Ryan Dempster, CHN
4) Brandon Webb, ARI
5) Cole Hamels, PHI