09 September 2008

Roy Halladay is Better Than Cliff Lee


Baseball Prospectus' Joe Sheehan has a column this morning arguing that some of Cliff Lee's success this season is due to the fact that he has faced much softer competition than his chief competitor in the AL Cy Young race, Roy Halladay. This sounded awfully familiar to me, since I had spent the better part of last week arguing that exact point in two different forums after some Cleveland fans got on my case for supposedly belittling Lee's accomplishments after he won his 20th game.

It turns out, however, that Sheehan's argument is based on the hitters that Lee and Halladay have faced, whereas my argument had to do with the starting pitchers who had opposed them. Here's Joe:


Cliff Lee has made 28 starts this season, Roy Halladay 29. Of those, 13 are in-common starts: the A’s, Rays and Rangers twice, and the Angels, White Sox, Reds, Royals, Twins, Yankees and Mariners once. Those starts cancel out. Of the remaining starts, there seems to be a very wide gap in the caliber of competition, enough to at least mention. Of the 15 starts Cliff Lee does not have in common with Halladay, nine have come against teams in the bottom third in offense, as ranked by team EqA, and none have come against a team ranked in the top six.

[...]

Let me run the data this way, because I think it illustrates the point. The following numbers are the team EqA ranks for each not-in-common opponent, highest to lowest.

Halladay: 3, 4, 4, 4, 9, 9, 9, 11, 11, 14, 14, 14, 14, 17, 18, 18

Lee: 7, 7, 7, 12, 13, 13, 21, 22, 22, 25, 26, 27, 28, 28, 28

It helps if you read those numbers right to left. It’s clear from this data that Cliff Lee has seen a significantly inferior set of opponents than Halladay has.
If it's not as clear to you as you might like it to be, let me help you out this way: The average EqA rank for Lee's opponents is about 19th, while Roy's opposition has averaged a rank just under 11th in the majors. Right now, for example, Baltimore ranks 11th in MLB in EqA, and they've averaged 5.1 runs per game. Houston ranks 19th in EqA, with just 4.47 R/G. The Runs Scored difference is exaggerated by their respective park effects, but you get the picture.

This average, however, is a disservice to you, as it does not sufficiently express the disparity between these pitchers' competition. Over the course of 15 starts, even if we took those Runs/Game numbers at face value, we'd have a difference of only 9 or 10 runs total, far less than one per game. If you look at Equivalent Runs (BP's attempt to normalize and neutralize for everything under the sun) the difference between 11th and 19th is more like 3 runs over 15 starts, i.e. not much.

This of course does not tell the whole picture, which is probably why Joe didn't present it that way. Halladay and Lee have not faced these "averaged" teams 15 or 16 times. They've faced actual teams, and the teams they've faced have been very different in terms of their offensive prowess. This helps to explain why Lee's ERA is lower than Halladay's. He's capitalized on his comparatively soft schedule.

My arguments last week
, in the comments section of my blog and of the post on Bleacher Report, centered around Lee and Halladay's respective opponents on the pitching mound, i.e. the starters their teammates had to face. One of the main reasons that lee is now 21-2 is that his teammates score almost six runs per game when he pitches, while Halladay gets only 4.75 R/G.

Looking at starting pitchers, these two have only five opponents in common: Zach Greinke, James Shields, Chien-Ming Wang, Sidney Ponson, and Matt Garza. If you take them out of the mix, the aggregate records of their respective foes are as follows:

Lee: 156-174, 4.60 ERA, 1.41 WHIP, 2.1 K/W
Roy: 195-176, 4.35 ERA, 1.37 WHIP, 2.3 K/W

The biggest difference here is the W/L records, though with enough data recovery there are notable (if not enormous) differences in all the pitching numbers. (These may be bigger if we were to adjust for park factors and such, but this is too time consuming as it is.) Lee has faced twice as many 10+ game losers as Halladay has (10 to 5), while Halladay has faced several more 10+ game winners (11 to 7), and this naturally leads us back to Sheehan's analysis, of how much tougher the offense has been against Halladay.

It may look like Lee has pitched slightly better overall, but the weak opposition on offense has helped his ERA tremendously, and the huge amount of run support he's gotten, thanks to facing some inferior pitchers, has helped a lot. Give Halladay 6 Runs per game and he's 22-5 instead of 18-9, and suddenly we've got a real race for the CYA.

Unfortunately, the BBWAA voters like shiny objects such as a 21-2 record, and rarely pay attention to things like how many times a pitcher gets to face the horrendous Kansas City Royals (hint: four). Neither do they fret much over whether Dontrelle Willis or Livan Hernandez or Clayton Richard or Chris Lambert or Carlos Silva were as tough to beat as Jose Contreras and and Josh Beckett and Andy Pettitte and Rich Harden and Jon Lackey. (Hint: No.) They just look at the pretty numbers in the newspaper and then vote whomever the heck they feel like voting. Which, this year, will undoubtedly and unfortunately be Cliff Lee.

I'm not saying that they shouldn't vote for him, just that they ought to think about the process a little more than they probably will.

Stumble Upon Toolbar

15 comments:

Anonymous said...

This is called....OVERANALYSIS....Cliff Lee has no control over whom Cliff Lee faces.

Of course Halladay will face the AL East and those great lineups more often than Lee.

You are really reaching here...any topic can and should be deconstructed to a certain extent. But this is a little much, sorry.

Anonymous said...

I actually just ran the same thing just now for fun...

In the games Lee and Halladay don't have in common Halladay faced teams with an average of 4.99 runs per game, while Lee's teams scored 4.53 runs per game

Clearly Halladay faced tougher hitters, I'm not sure how to correct for that...I woke up this morning 100% sure Lee deserved the Cy Young. Now I'm not so sure, but put a gun to my head and I'd still choose Lee. What he's done this year is flat out amazing.

Anonymous said...

I suppose I should add that I'm a huge Jays and Halladay fan, and think your headline "Halladay better than Lee" is a no-brainer.

I just think sometimes the Cy Young goes, not to the best pitcher, but the pitcher who had the best year.

dbvader said...

Light blue against a glaringly white background is not good for the eyes.

Livingston said...

Isn't the point of objective statistical analysis to eliminate factors that a player cannot completely control (which is the argument against judging pitchers solely based on wins)?

So, why are we supposed to judge Lee based on another factor that he cannot control at all (the schedule)? Yes, some pitchers do ask to be moved to match up against an ace or managers do it for them. But both Cleveland and Toronto have been effectively out of contention much of the year. That argument would be baseless.

So, we're left with Lee having a lower ERA and higher WHIP than Halladay. The WHIP is largely a factor of IP - Halladay has 8 CGs to Lee's 4 (although they have the same number of CG SHO), giving him 17 more IP than Lee. Both are goign to be well over 200, so neither has been protected from going as far as they could. Lee's walked slightly fewer per plater appearance; Halladay's struck out slight more; Lee's base runner's per 9 is slightly better and he's allowed significantly few HRs. Lee's FIP is about .6 better than Halladay's.

So, using the conventional or unconventional statistical measurements, it is close and Lee wins. Just as judging's someone's season on total wins (like Lincecum), it isn't fair to deduct points for mowing down a slightly weaker competition. Isolating the 15 starts not in common is an exaggeration of the overall quality of opposition since the season has to be judged as a whole.

One factor out of their control that might be fair to consider (since it clearly affected Roy's win total) is run support. But if wins are an exaggerated measurement, then I guess that doesn't matter.

Travis M. Nelson said...

I don't think we should blame or penalize Lee for having faced softer competition, just that we shouldn't ignore the fact. Consider these two pitchers:

Pitcher K: 14-6 3.45, 117 K's in 156 IP

Pitcher O: 14-9 3.72, 144 K's in 182 IP

Pretty close, right? Pitcher O is Roy Oswalt, while pitcher K is Kei Igwa, and those are his numbers in AAA this year. Nobody thinks that Igawa is almost as good or better than Oswalt, because we know he hasn't faced the same level of competition. It's an extreme case, but the logic holds. Competition matters.


Oh, and as for dbvader's comment: I'm sorry about that. It's a dark blue background in IE, but in Firefox you get the white, and I can't figure out how to change that.

Wally said...

Livingston, how in the world is having a lower WHIP due to pitching more innings and not due to giving up fewer hits and walks per inning? WHIP is a pretty lame stat anyway. How many hits vs. walks? What kind of hits? All make that stat pretty terrible.

"Isolating the 15 starts not in common is an exaggeration of the overall quality of opposition since the season has to be judged as a whole."

While this is true, I don't believe the case being made is that Halladay is better because of those 15 starts. The argument is that the traditional stats don't tell the whole story because of those 15 starts (and the RS).

Travis, congrates on getting the Wednesday Wangerdoodle link too. You have good posts, I'll admit you're site isn't a regular of mine, but I do read it time to time, and I'm always impressed. Keep it up.

Anonymous said...

I'd like to add that even tho(theoretically) CLee doesn't have a say in who he faces, his manager and pitching coach most certainly do! As does Halladay's. There is long history of managers arranging to have their horse face their best competition and Halladay's usage leaves no doubt to this. Lee's looks intentional as well. Nothing wrong with that, but the pattern IS human and should be acknowledged.

steve said...

Last anonymous:

At the start of the year, Lee was the 5th starter, so obviously his usage pattern was different. Since the ASB, he has been our clear #1.

Did Wedge hold Lee back against the Mariners so that he could instead face the White Sox because he wanted Lee to face "inferior competition"? Seems like he would have put him out there against the M's if his intent was to inflate Lee's numbers.

Lee has a big lead over Halladay in PRAA, WARP3, WPA, and IIRC Pitcher VORP. He's the better pitcher this year.

Anonymous said...

Cliff Lee has 5 no decisions. one of them he gave up zero runs in a 9 inning outing. Two of them he gave up 1 run, one he gave up 2 runs.

He could realistically have been 25-2 right now!

Travis M. Nelson said...

Cliff Lee has zero "tough losses" and one "Cheap Win" according to ESPN.com's stats page. That's a loose definition, because he's also got a win for allowing 4 runs to KC in 6 IP. Halladay had a similar outing, 4 ER in 6.2 IP to Minnesota, but he got the win, so those two are basically a wash.

Halladay has four tough losses and zero cheap wins, plus a no-decision in which he threw 7 shutout innings against the Pirates.

Edge to Halladay, again, if only slightly.

steve said...

Again, dude, you're really reaching. Yes, the difference between Halladay and Lee isn't as much as Lee's inflated W-L numbers would suggest, but even if you neutralize their records, Lee STILL wins.

You want to argue things outside of their control? How about defense? BTW, Lee's BABIP is actually higher than Halladay's, and yet he still beats him in opponent's OPS. You think Lee likes pitching with Ryan Garko and Casey Blake and Jhonny Peralta and Andy Marte behind him? You don't think he'd rather have John McDonald and Scott Rolen?

No matter how much people try and stretch it, Lee is having the better season. He leads by 17 in PRAA (which is a huge advantage). He leads by 11 in Runs Prevented. He leads in WARP3, he leads by 13 in pitching VORP, he leads in everything, really. And the whole CG argument? Halladay has one more GS and he pitched 2.1 innings of relief. Lee's and Halladay's difference in IP/GS is miniscule, less than a batter per start, and Lee is 2nd in the majors in IP/GS btw; Halladay is 1st, of course.

Anonymous said...

The Jays and Tribe recently played against each other. Does anyone know why when Haladay and Lee were matched up against each other Wedges pushed Lee back a day? Halladay ended up losing to Byrd anyway, and Lee pitched against a weak rookie Richmond and won easily.

Both teams may have realistically been out of the race for a long time, but Toronto thinks they have a shot. Mathematically they do I suppose. The Indians know they have no shot, and came to that realization a long time ago.

Anonymous said...

I'm a loyal Halladay fan, but I must say with all honesty that this was an off year for Halladay. I have watched Halladay at times where he has gotten run support and then totally exploded the next inning walking a guy, then giving up a hit, and another hit, and another. I thought, "this is not like Halladay". In my eye he doesn't deserve the Cy Young as well as Cliff Lee does because voters expect greater out of Halladay then they do Lee. Nobody expected Lee to win 22 entering 2008. NOBODY.

Anonymous said...

"This is called....OVERANALYSIS....Cliff Lee has no control over whom Cliff Lee faces."

The Indians do, to a certain extent. At least with respect to match-ups to some extent.

I'm not alone in thinking that Lee's random extra day's rest kept him from facing Halladay. The Tribe decided to rest him for no reason and that he would go against AAA filler Scott Richmond the next day.

I'm not saying that happened all year... Lee has had a phenomenal year. I personally believe that Halladay's was better.

And why are people referencing the overall average runs scored by the offenses per game? The number of runs scored for AJ Burnett don't help Roy Halladay. Cliff Lee got a fair amount more support than Halladay did.

Cliff Lee's RS: 6.13
Roy Halladay's RS: 4.72

It's very, very close overall. And it will go to Lee based on the W-L record.