Not Quite Yet, Cliff

This afternoon, Eric presented the case for Cliff Lee, asking the somewhat rhetorical question of whether he’s already wrapped up the Cy Young award. Well, Roy Halladay called, and he told me to tell Eric “not so fast, buddy”.

First, the comparison of traditional stats.

Wins: Lee (16) over Halladay (13)
ERA: Lee (2.45) over Halladay (2.72)
Innings: Halladay (182) over Lee (161 2/3)
Strkeouts: Halladay (155) over Lee (128)
Complete Games: Halladay (7) over Lee (2)

If you don’t like to go beyond the 1950s numbers to evaluate pitchers, then you have to decide whether you like quality (Lee) or quantity (Halladay). Lee’s surface numbers are a bit better, but Halladay’s pitched more. I’d imagine most voters who base their decisions on Wins and ERA will go with Lee, but for those of us who think pitchers should be evaluated by metrics a little more current, let’s take a look at how they’ve performed in things they can control.

BB/9: Lee (1.22) over Halladay (1.53)
K/9: Halladay (7.66) over Lee (7.13)
HR/9: Lee (0.39) over Hallday (0.64)
WPA/LI: Halladay (3.97) over Lee (3.90)

Lee has a lead in the two command factors of FIP, walks and home runs, by slight but significant margins. Halladay has a similar lead in dominance, getting more strikeouts and ground balls. Hallday’s FIP of 2.99 is half a run higher than Lee’s 2.48, almost all of which is tied to Lee’s home run prevention. On a per inning basis, Lee’s got the advantage, no doubt.

However, as WPA/LI shows, the extra 20 innings that Halladay has thrown matter. If Lee had thrown the same 182 innings that Halladay has completed, his advantage based on FIP would be FIP (or ERA, as both have ERAs close to their actual performance) would be about 8 or 9 runs. Halladay, however, needs to get credit for pitching more often, so we need to adjust the difference in innings to account for the other pitchers Cleveland has had to use in lieu of Lee going deeper into the game.

Lee has allowed 46 runs in his 162 2/3 IP, but the Indians bullpen has been terrible this year, allowing 5.38 runs per nine innings as a group. If we multiply that RA times the 20 inning difference between Lee and Halladay, we get an additional 12 runs given up. That would push Lee + bullpen’s total to 58 runs allowed in 182 innings, which is just six runs better than Halladay’s runs allowed total.

The extra innings Halladay has pitched close the gap, but not entirely – Lee still has the lead right now, even after accounting for quantity. However, with a month and a half to go in the season, there’s still certainly time for Halladay to make up the difference and then some.

Lee’s season has been great, but Doc Halladay isn’t that far behind. Let’s not give him the Cy Young just yet.



Print This Post



Dave is the Managing Editor of FanGraphs.


Sort by:   newest | oldest | most voted
NickP
Guest
7 years 11 months ago

This feels like the Bad Back-up Effect.

Some pitchers have bullpens that are better than they are. This sort of analysis would actually help them, no?

Josh
Guest
Josh
7 years 11 months ago

what about if halladay had pitched 20 fewer innings and his bullpen had pitched 20 more innings? what would the comparison look like then?

Josh
Guest
Josh
7 years 11 months ago

perhaps i should have clarified: how many runs is halladay saving his team by pitching those 20 extra innings?

the better way to look at this (my opinion) would be to use the average runs allowed by all bullpens per inning – no reason to punish lee for being on a team with such a lousy bullpen – and see how much more valuable it is to the average team to have their starter pitch 20 (or 30) more innings per year.

Trev
Guest
Trev
7 years 11 months ago

I agree with Josh.

Penalizing Lee for the quality of his bullpen is just like penalizing Halladay’s teammates for not giving him the run support for 4 more wins.

For what it’s worth, the Toronto bullpen has been excellent this year — 3.01 RA/9. Using your equation, Cliff Lee + Toronto’s pen (7 RA in 20 IP) = 53 RA. This pushes Lee’s advantage to 11 runs over Halladay.

Using AL average bullpens (4.30 RA/9) for Lee we get 56 RA, eight runs better than Halladay.

Trev
Guest
Trev
7 years 11 months ago

Because I really don’t want to get back to work, here’s another useless exercise: How about we compare the total runs allowed in the games Halladay and Lee pitched in.

Halladay made 24 starts (and one scoreless 2.1 inning relief appearance). Setting that aside, he “asked” his bullpen to pitch 36.1 innings.
w/ TOR bullpen: 76 RA
w/ avg bullpen: 81 RA
w/ CLE bullpen: 86 RA

Lee made 23 starts and averaged 7.03 IP/GS. If we give him one more start at his current IP and RA rates, he “asked” his bullpen to pitch 47.1 IP.
w/ TOR bullpen: 64 RA
w/ avg bullpen: 71 RA
w/ CLE bullpen: 76 RA

What does this all mean (besides I really like to procrastinate)? That in 24 starts, Halladay and Lee’s teams both give up 76 runs despite the difference in bullpen quality.

Trev
Guest
Trev
7 years 11 months ago

RE: Dave’s comment making my last post useless

But then we’re making a judgment we can’t make on how many innings Lee would pitch in an ideal world, pitch to “his heart’s content”. I don’t think we can judge whether Lee goes back out for the 8th because Wedge needs him to or because that’s Lee’s optimal usage pattern, irrespective of the bullpen behind him. [How do we know what Lee’s usage would be in a world where Lee pitches as many innings as he wants versus a world where he pitches as many innings as Wedge says so?]

That said, you could use Lee’s pitches/GS to help your point. From 2002-2007 Lee averaged 97 pitches/start while he’s averaging 106 this year. But this could just be a product of increased effectiveness.

Scott Batura
Guest
Scott Batura
7 years 11 months ago

Yeah, but this IS the Cy Young– 1950s metrics are apparently all that is relevant. That given, then yes, it looks like Cliff Lee will be a shoo in this year.

Finn
Guest
7 years 11 months ago

For what it’s worth, the Neyer/James Cy Young Predictor has the following AL Cy Young Top 5:

1. Francisco Rodriguez (154.4)
2. Cliff Lee (150.1)
3. Joe Saunders (124.8)
4. Jonathan Papelbon (124.2)
5. Roy Halladay (122.4)

The formula is ((5*IP/9)-ER) + (SO/12) + (SV*2.5) + Shutouts + ((W*6)-(L*2)) + A 12-point bonus awarded for leading your team to the division championship (pro-rated based on the current standings).

Tom Lyons
Guest
7 years 11 months ago

Dave,

I don’t find your argument at all compelling against using an average bullpen’s replacement of Lee’s innings v. the Indians bullpen’s replacement of Lee’s innings. If you’re not going to isolate Lee from his teammates in evaluating Lee, then Halladay deserves the same failure to isolate, which you haven’t done.

The manager’s decision to continue riding a starter — particularly an ace — largely comes down to pitch count, not bullpen. No matter how awful your bullpen, when your ace is spent, your ace is spent. Halladay has thrown 14.46 pitches per inning vs. 15.02 per inning for Lee. Why? In large part because Toronto’s defense (8th in DER) ends innings faster than Cleveland’s defense (21st).

If you’re going to give Halladay a bump because he’s been of slightly less quality in 21.1 more IP of quantity, that’s not unfair. But you have to weed out some portion of those 21.1 IP to Toronto’s defense — NOT all to Halladay — as they ended innings more effectively, and thus slowed Halladay’s approach to his pitch count, and thus extended his IP.

Peter Bendix
Guest
Peter Bendix
7 years 11 months ago

I don’t think we should penalize Lee for his usage. It’s one thing if there is something about Lee that suggests that he is posting quality numbers in part because he is being used very carefully, but I do not believe this is so (if, for example, he’s removed after 75-80 pitches per start). But this is not the case, and, as such, we have no reason to believe that Lee would be particularly harmed if he pitched 20 more innings.

If anything, the Indians bullpen HURTS Lee, as they are less likely to strand runners that Lee leaves on base, and Wedge is MORE likely to leave Lee in even if Lee is tired, thereby hurting his numbers.

It might be another case if we were comparing the exact same run prevention spread across different amounts of innings (such as CC Sabathia vs Josh Beckett last year). But Lee has been better at preventing runs, and as such I think he deserves the nod over Halladay.

Plus, he has so many more wins :).

Big Steve
Guest
Big Steve
7 years 11 months ago

Not a very dtrong effort here Dave. Using the Cleveland bullpen in part to evaluate Cliff Lee’s performance defies logic. I could go through all of the reasons, but the other commenters have done a pretty good job sinking your argument.

Scappy
Guest
Scappy
7 years 11 months ago

If anyone cares, I just crunched the WPA+/- numbers and Halladay has earned a 14-7 record while Lee should be 13-3. Halladay has effectively pitched more games than Lee, but he has not pitched as well. That said I don’t think I would complain either way.

Now 2005 was not an exceptionally good year for the voters. Based on WPA Santana was 16-9 while Colon was 14-9, through in the SO and ERA and there is no reason for Santana to not win the award.

The win statistic as it is used today is pretty awful. If I played fantasy baseball I’d lobby for quality starts to be used instead of wins.

wpDiscuz