The Quality of Cole Hamels’ Opposition

We’re used to making little adjustments all the time. Most commonly, it’s because of ballpark environment. A .350 wOBA in San Francisco is a hell of a lot more valuable than a .350 wOBA in Arizona. Sometimes you’ll also see adjustments for era, which is relevant now given increasing strikeouts and decreasing runs. There are raw stats, and there are adjusted stats, like, say, wRC+, but there’s one adjustment we seldom talk about even though it’s right there in front of our faces. What about the opposition a player actually faces?

It’s like strength-of-schedule, on the player level. No one debates the utility of strength-of-schedule measurements. Now, in baseball, what’s convenient is that the samples get pretty big so we can generally get away with assuming that things even out. Over broad windows, no one’s going to face exclusively awful opponents or awesome opponents. But in certain cases, it’s worth digging in when we have a suspicion. As such, I want to go into more detail on something I noted about Cole Hamels earlier.

Just for the sake of easy comparison, let’s line Hamels up with Jon Lester, Max Scherzer, and James Shields. Three of those are free agents, but Hamels is also highly likely to move, perhaps to someone who misses out on the FA market. This is a table of WAR values, where I’ve taken regular WAR and RA9-WAR and averaged them 50/50. Except for the Steamer line. That’s just regular WAR, because that’s what’s projected.

Year Hamels Lester Scherzer Shields
2012-2014 13.2 12.3 15.8 12.6
2014 4.9 6.0 5.5 3.8
2015, Steamer/200 2.7 3.5 3.9 3.0
2015 Age 31 31 30 33

Hamels, Lester, and Scherzer are of similar ages, with Shields a little older. Over the last three years, Scherzer’s been the best, with the others more or less equal. In just this past season, Lester was awesome, and then Scherzer was a little less awesome, and Hamels was a little less awesome, and Shields was still less awesome. The projected order goes Scherzer — Lester — Shields — Hamels, but then those are just projections, right? Hamels, for a long time, has been really good. No one questions that he’s a good pitcher.

This is around where we usually stop. But I wanted to investigate Hamels’ quality of opposition, compared to the other guys. Hamels has lived in the National League East. The others have all lived in the American League. I went over the last three years, and matched each hitter seen to the same hitter’s single-season wRC+. Then I weighted everything to come up with an average hitter wRC+ faced. This is the best quality-of-opposition measure I can come up with in a day. And the results, I’d say, are revealing:

Year Hamels Lester Scherzer Shields
2012 90 97 100 102
2013 88 100 96 98
2014 89 98 100 103
Average 89 98 99 101
Hitter Comp Everth Cabrera Alex Avila Brian McCann Eric Hosmer

You’ve got single-season average wRC+, then you have the average of the last three seasons. In the last line, you see a hitter who has posted the same wRC+ over the past three years. On average, since 2012, Cole Hamels has faced a bunch of Everth Cabreras. James Shields has faced a bunch of Eric Hosmers. There’s little difference between the AL guys, but Hamels is far removed. You have to assume this has worked to Hamels’ benefit.

How might one be able to adjust for this? It’s easy, after all, to say that Hamels has faced weaker competition, but, what does that mean? It’s time to try something. Over the past three years, Hamels has allowed a .293 wOBA. His batting opponents have averaged a .302 combined wOBA. I decided to work through the odds ratio method backwards. I wound up with a value of .304 — that is, based on the math, Hamels’ “true-talent wOBA” over the past three years would be .304. Or, 11 points higher than what we observe, based on his results.

Of course, that’s still a good mark. Still a better mark than average. But it does make a real difference. There are a couple ways to figure out the impact. For example, over the past three years, Hamels has averaged a 4.4 RA9-WAR per 200 innings. Pitchers with wOBAs right around .304, however, have averaged a 3.2 RA9-WAR per 200 innings. Alternatively, you can just calculate the difference between a .293 wOBA and a .304 wOBA over Hamels’ 2,601 plate appearances. That comes out to about 21 runs. Or, an average of almost a win a year.

That would all be a quality-of-opposition adjustment for Cole Hamels. He’s faced relatively weak opponents. If you retroactively have him face roughly average opponents, you’d expect him to be almost a win worse a year. He’s still better than average, by a decent amount, but the gap shrinks. To whatever extent the usual numbers favor Hamels over James Shields, Shields has faced far tougher opponents. So interested teams don’t need to worry about adjusting him for that. For Hamels, I think it’s necessary.

This obviously all gets complicated, and well removed from the field of play. On the field of play, Hamels has the same fastball as ever, and a really fantastic changeup. The quality of his stuff is independent of the hitters he faces. It’s not because of the hitters that he’s been durable. It’s not because of the hitters that he’s got playoff experience. Hamels is good enough to pitch anywhere, and he’s recognized as a front-of-the-line kind of starter. There’s a reason teams are interested in him.

But at the end of the day, expensive players get evaluated mostly by how they’ve done. The numbers drive the contracts and the trades, and the numbers get adjusted by any half-smart organization. Everybody understands ballpark effects. There are also, sometimes, opponent effects. Adjusting for the latter makes Cole Hamels look less valuable than his actual numbers, and that’s just one of the reasons why it’s proving so difficult for Ruben Amaro to find a deal to his liking. The Red Sox don’t care so much about how Hamels might do in Philadelphia. They care about how Hamels might do in Boston.



Print This Post



Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.


Sort by:   newest | oldest | most voted
emdash
Guest
emdash
1 year 6 months ago

Did you remove the Phillies themselves from his quality of opposition? Their offense hasn’t been very good over these three seasons and I’d imagine they’d bring down the average pretty well.

Also, I don’t believe the hitters’ numbers have been adjusted for the quality of pitchers they faced in their division, which seems pretty important in determining the opposing hitters’ level of skill.

unabashed phillies fan
Guest
unabashed phillies fan
1 year 6 months ago

but have you thought abt how your measures say he is a better player than all thsee other free agent SPs and how many world serieses he has been the MVP of? that doesn’t just grow on trees, Bryant+ coming from Cubs with philly eating no money easy.

Spencer00
Guest
Spencer00
1 year 6 months ago

+1 because I still have faith in humanity and assume this is satire.

Baltimorechop11
Guest
Baltimorechop11
1 year 6 months ago

Can you run for Cueto and Kershaw 2014? Both had same number of starts against sub 500 teams, while Cueto had extra 7 against winning teams. Curious to see if that would alter the numbers.

Bob
Guest
Bob
1 year 6 months ago

Its November 19th. Christ wait till the Lester signs his 8 year deal and Ervin Santana gets 90 million and then see how the pieces fall.

Christ
Guest
Christ
1 year 6 months ago

With the Marlins paying a guy $325 million, I might not wait until Lester or Santana sign their deals. Hell might have already frozen over.

Phillyfan425
Guest
Phillyfan425
1 year 6 months ago

What if you stated it as Shields “averaged” to face Asdrubal Cabrera, Scherzer “averaged” to face Logan Morrison, Lester “averaged” to face Delmon Young, and Hamels “averaged” to face J.J. Hardy? Would it have changed your opinion on the “quality of opponents” that each faced? Because over the last three years, each one of those players fall into the correct category that Jeff listed.

Nik
Guest
Nik
1 year 6 months ago

Amaro is such an idiot for not taking a crappy prospect for his franchise pitcher in order to rebuild.

Wilbur
Guest
Wilbur
1 year 6 months ago

Pot, meet kettle.

*BG*
Guest
*BG*
1 year 6 months ago

Always thought Philly fans were thicker skinned than this.

Ebenezer
Member
Member
Ebenezer
1 year 6 months ago

They may be, on the whole – this is a small sample size.

Jim
Guest
Jim
1 year 6 months ago

The Philly fans you are thinking of do not visit FanGraphs. Most Phillies fans think Cole is a pretty boy and harbor resentment towards him.

Deelron
Member
Deelron
1 year 6 months ago

Don’t worry, he won’t take any prospect and will just half-ass rebuild. Again.

Nik
Guest
Nik
1 year 6 months ago

Isnt WAR already park and league adjusted? Why jump through hoops to get these numbers? If these had any validity whatsoever, why aren’t they part of the WAR calculation to begin with? Also fWar for Hamels is quite different from his rWAR.

Psy Jung
Guest
Psy Jung
1 year 6 months ago

These aren’t park or league adjustments. Hamels has just faced lesser oppenents for no particular structural reason.

regfairfield
Guest
regfairfield
1 year 6 months ago

So you proved Hamels faces pitchers, which WAR already adjusts for.

ElJimador
Guest
ElJimador
1 year 6 months ago

“I went over the last three years, and matched each hitter seen to the same hitter’s single-season wRC+. Then I weighted everything to come up with an average hitter wRC+ faced.”

Does this include pitchers? If so, how exactly did you weigh to come up with those single season wRC+ faced? Because a pitcher in the NL is going to face the opposing pitcher not more than twice a game on average but if you were to take every one of those pitcher/hitters and average their single season wRC+ the same as every real hitter who they face more often then that would still give you an average wRC+ faced but one that seriously underestimates the overall level of competition that Hamels actually faced.

Please elaborate on your methodology on this and if you could please also let us know the average wRC+ against if you removed pitchers from the opponents’ sample. Thanks.

Paul Clarke
Guest
Paul Clarke
1 year 6 months ago

Baseball Reference uses something like this in its WAR calculations: RA9opp, or what the pitcher’s opposition would be expected to hit in an average park against an average pitcher and defence. Hamels’ figures with NL average in brackets:

2014: 4.01 (4.02)
2013: 3.93 (4.08)
2012: 4.24 (4.31)

That works out to a .08 runs/9 IP advantage for Hamels over 640 IP. That’s 5.57 runs total, or about 0.6 wins, which is quite a bit less than Jeff’s results. The difference seems to be due to B-Ref blaming the Phillies’ defence for about 17 runs lost over the same innings.

LTG
Guest
1 year 6 months ago

I’m pretty sure the defense is not factored into the RA9opp. It has its own column and serves to calculate the RA9avg to which the pitcher’s RA9 gets compared. B-Ref just thinks the difference in Hamels’s opponents is not as great as Jeff’s evidence here seems to indicate.

Paul Clarke
Guest
Paul Clarke
1 year 6 months ago

That was a lot clearer in my head than on the page. What I’m saying is that when Jeff takes the opposition’s .302 wOBA and Hamels’ .293 wOBA allowed and calculates the “true-talent” wOBA, he’s not taking into account that when facing Hamels the opposition was also facing the Phillies’ defence, which hasn’t been that great the last couple of years. Against the Phillies’ defence and an average pitcher, those hitters should be expected to hit better than .302 – I make it about .307 if you use UZR and about .314 if you use DRS. If you use one of those numbers as an input to the odds ratio calculation then Hamels’ “true-talent” wOBA will come out better.

LTG
Guest
1 year 6 months ago

Right. So, I agree that in order to complete the analysis Jeff does in the article the Phillies’ terrible team defense must be included. I merely disagree that B-Ref’s RAopp includes that factor already. From that it follows that B-Ref offers evidence that somehow Jeff is overestimating the run-differential between Hamels and the others based on the wOBA-against data. As I say below I suspect the odds ratio method has something to do with the difference.

Paul Clarke
Guest
Paul Clarke
1 year 6 months ago

I’m not saying that B-Ref’s RAopp includes defence, in fact I said that it assumes an average defence (and park, and pitcher). What I did was to compare Hamels’ RA9opp with the NL average RA9opp; as both RA9opp figures make the same assumptions, that works to isolate just the advantage from strength of opposition. The, or at least a, reason that Jeff gets a larger number is that his method doesn’t isolate strength of opposition: it includes defence too (and park effects, but CBP is pretty neutral).

However, now that I think about it there’s a defect in my method too: I compared Hamels’ RA9opp to the average of all NL pitchers. I should probably have compared it to the average of all NL starting pitchers, because NL starters pitch to pitchers more often than NL relievers do. Unfortunately B-Ref doesn’t provide league average RA9opp just for starters, and Fangraphs won’t tell me wOBA against just for starters either.

Paul Clarke
Guest
Paul Clarke
1 year 6 months ago

Hey, Baseball Prospectus has a “Pitcher’s quality of opposition” report. If I limit that to NL pitchers who’ve thrown more than 150 innings (to weed out the relievers and part-time starters) I get the following figures for opponent’s TruAv for Hamels (estimated NL starter average in brackets):

2012: .254 (.256)
2013: .239 (.254)
2014: .255 (.257)

That averages to a .003 advantage for Hamels. I can’t find the precise method of converting TruAv to runs, but it’s a linear weights measure scaled so that average is .260, as opposed to wOBA which is scaled so that average matches OBP (.317). So as an estimate multiply by .317/.260 to convert to wOBA and divide by wOBAScale (1.275) to convert to runs per PA, then multiply by Hamels’ total batters faced (2601):

.003 X .317/.260 / 1.275 X 2601 = 7.5 runs

If that calculation can be trusted, the advantage Hamels gained due to weaker than average hitters was worth a bit less than one win over three seasons.

Paul Clarke
Guest
Paul Clarke
1 year 6 months ago

That .239 for 2013 should be .249 by the way.

LTG
Guest
1 year 6 months ago

Wow. Thanks for the doing the leg work on this. I understand now what you meant. And completely agree. So that’s a win for everyone, especially rational defense.

LTG
Guest
1 year 6 months ago

‘defense’ should have been ‘discourse.’

Nathaniel Dawson
Guest
Nathaniel Dawson
1 year 6 months ago

Not only that, LTG, but you spelled it wrong as well. It should be “defence”.

Nathaniel Dawson
Guest
Nathaniel Dawson
1 year 6 months ago

5.57 runs would better be rounded to .5 wins, using (what I think is current runs to win conversion of) 9.3. Minor and perhaps trivial, but it seems we might be delving into the minor and trivial here.

Paul Clarke
Guest
Paul Clarke
1 year 6 months ago

9.3 is about right for 2012-2014, but 5.57/9.3 = 0.5989.

Nathaniel Dawson
Guest
Nathaniel Dawson
1 year 6 months ago

Ha! Very true. Maybe I should be less concerned with spell-checking my posts and more with the math-checking. Although both at times need work.

LTG
Guest
1 year 6 months ago

I appreciate your response to the clamor. But I’m wondering about the odds ration method used as you did. Tango came to this conclusion: “the Odds Ratio method only works on single confrontations, and not “group level” results. So, a .600 team facing a .400 team will NOT necessarily follow the Odds Ratio (log5) process.”

I’m not sure to what extent this bears on your findings. But I take it that you applied the method to a group level result and not a single confrontation. It might be that the odds ratio method overestimates the difference between the competition Hamels has faced and that faced by the other pitchers.

Josh
Guest
Josh
1 year 6 months ago

the stupids sure came out of the woodwork to comment on this article! it’s like a dozen mini RAJs running around fangraphs.

Jeff, very good analysis that brought my attention to something I overlook on a daily basis.

Ebenezer
Member
Member
Ebenezer
1 year 6 months ago

Good analysis, Jeff. I realize it would take considerably more analysis, but I wonder how much of a factor this plays with some players who switch teams, whether due to free agency or trade, afterwhch their performance seems to collapse. I’m also wondering what have been the most extreme cases of player stats being helped or hurt by quality of opposition.

ML
Guest
ML
1 year 6 months ago

Really wish this type of data was more readily available. Interesting to analyze but so difficult to aggregate and compute, especially manually. Nice read.

KK-Swizzle
Guest
KK-Swizzle
1 year 6 months ago

Props to Jeff for writing a good article AND replying to numerous (somewhat hostile) comments. I’ll be honest, I was surprised that this had such a big impact…a win per year is over 20% of his perceived value! I love articles that think outside of the box like this

Josh I
Guest
Josh I
1 year 6 months ago

Well written article, and I have to say I’ve become a big fan of yours Jeff. If I could suggest one thing, it would be to include a similar NL pitcher as well for comparison’s sake. Outside of that, I love the analysis and am impressed with how well you’re handling some of these jerks in the comments.
On a different, impractical note, let’s say Hamels isn’t traded until next deadline. I wonder if they could sneakily line up his starts (“family issues”, “hangnails”) to avoid Washington, St. Louis, etc. to try and inflate his numbers and his value. Would this work, or be more effort than it’s worth?

Cole Hamels
Guest
Cole Hamels
1 year 6 months ago

As I faced relatively inferior hitters, perhaps I faced easier questions when I took the SAT and got a 1510/1600 (although difficulty should be accounted into the score).

DNA+
Guest
DNA+
1 year 6 months ago

Jeff,

Since you’ve gone to the trouble of calculating this figure for all pitchers, you should plot the distribution so that we can see how unusual (or usual) each of these pitchers are in their strength of batters faced. It would also be interesting to plot leagues and divisions on their own so that we can see if the differences can be explained by the team a pitcher plays for.

ASK
Guest
ASK
1 year 6 months ago

Jeff,

Since you went to the trouble of breaking it down to each plate appearance, might it have been more telling to use each hitter’s wRC+ vs. the handedness of the pitcher in question? Is it possible that Hamels, as the only LHP in the group, faced more platoons than the others thereby facing more lower wRC+ hitters who happen to have higher platoon split wRC+’s? Of course, I may be completely off-base here.

DNA+
Guest
DNA+
1 year 6 months ago

Jon Lester is generally considered to be left handed.

vivalajeter
Guest
vivalajeter
1 year 6 months ago

I’ve only heard him referred to as being a Southpaw. Maybe Fenway is facing the opposite direction of every other ballpark, and a pitcher’s left arm faces North?

hk
Guest
hk
1 year 6 months ago

Of course he is. What I meant to say was “being the only NL pitcher in the group” as I believe – albeit without any evidence – that NL teams seem to platoon more than AL teams.

LTG
Guest
1 year 6 months ago

hk = ASK? My world is rocked.

ASK
Guest
ASK
1 year 6 months ago

Long story…multiple computers…”hk” populating the “Name” box for some reason.

Sam Fuld
Guest
Sam Fuld
1 year 6 months ago

Excellent article, even better comment section.

Harry
Guest
Harry
1 year 6 months ago

Just throwing something up there… So, the assumption is Hamels numbers wouldn’t be as good if he was going against better opposition. But is that true? I think there is something to be said about a pitcher rising to the competition. I think a better determinant would be to look at how Hamels has performed against better opposition during his career (e.g. interleague play.

Bonus Wagner
Guest
Bonus Wagner
1 year 6 months ago

Harry,

Tell us about the values of “clutch” and “grit” next…..scrappiness too if you have the time. Murray Chass is giving you a standing ovation somewhere in Michigan.

LTG
Guest
1 year 6 months ago

While you might not like the intangibles he references, it is in fact possible that a pitcher fair more or less the same against a wide range of quality of hitter. There’s no physical law that says a pitcher’s performance against various hitting talents is a smooth curve.

Harry
Guest
Harry
1 year 6 months ago

Bonus Wagner – Thank you for the usual, lazy response. To be clear, I am a big proponent of advanced stats, and what I suggested has nothing to do with grit or clutch and should be rather easy to measure.

awalnoha
Member
awalnoha
1 year 6 months ago

I get this but I think that Hamels would have gotten out better hitters, his performance is likely not related to his quality of opponents at least on the micro level. You see that good and great pitchers tend to get out about everybody except for the occasional hitter that just does well against that particular pitcher. If the difference was extreme then you might see it affect the outcome, say it was the equivalent of facing Trout or Cabrerra compared to a more league average hitter.

Jeff, could you compare the pitchers above by collecting the results when they face those same classes of hitters. If I were the Phillies I would build a table that shows him compared to others pitchers in wRC categories by increments of 10-20, assuming it would show that he was still good against the better competition.

Alex
Guest
Alex
1 year 6 months ago

How much of this has to do with simply playing in the NL and getting face the oopsing pitcher? I have to think NL pitchers as a whole face opponents with a lower wRC+ simply because they face a pitcher ~3 times a night.

novaether
Member
novaether
1 year 6 months ago

If you’re going to use RA/9 then you have to look at defense. The Phillies defense posted -149 DRS over the past 3 years, so give Hamels 15% of the innings and you get about -22 runs over the 3 year span. That nullifies the 21 run benefit from the wOBA calculation.

David
Member
Member
David
1 year 6 months ago

for a more fully expressed counterpoint on the value of Cole Hamels, please see the goodphight.com

David
Member
Member
David
1 year 6 months ago

Look, you can take these pitchers and slice them a lot of ways. Depending how you slice, one or the other of them looks better. FanGraphs seems to be focusing on ways that make Hamels look not as good as Lester, Shields, or Scherzer, like a step below. People who have watched Cole Hamels pitch a lot might disagree because of there natural biases. But if you slice another way Hamels might fit right in with this group.
How about I cherry pick K%-BB%.
Over the last three years, Scherzer leads the way among these pitchers around or exceeding 20%. Lester put in an awesome 19.2 in 2014, but otherwise significantly lower in 2013 (12.2) and 2011 (11.2)
Shields 14.5, 13.5, 17.5
Hamels 16.8, 16.8, 18.9

LTG
Guest
1 year 6 months ago

You frame the disagreement as if its stats vs. watching Cole. Given rWAR’s evaluation of the relative value produced by these pitchers, I’d say the disagreement is more like stats vs. stats, specifically, FG’s premises vs. BR’s. Or at least that’s the disagreement worth attending to.

dfa
Guest
dfa
1 year 6 months ago

Can you please add quality of opposition stats to fangraphs

Kenny Powers
Guest
Kenny Powers
1 year 6 months ago

Yea, start with pitcher vs hitter career numbers.

Kenny Powers
Guest
Kenny Powers
1 year 6 months ago

For a site that likes to measure pitchers on what they can control, basing an argument on something they can’t control like their opposition, seems kind of cynical.

KDL
Guest
KDL
1 year 6 months ago

So…what you’re arguing is that it’s weird to consider facing the Tigers and Padres as different?
I’m not sure what you’re driving at with this comment.

Pitchers do not set the schedule, true.
But facing weaker talent is a tangible thing.
And understanding this helps us understand the true talent of the pitcher.

Pitchers also are at the whims of the defensive quality behind them.
And those whims are removed because it is beyond the pitcher’s control.
And applying these controls help us understand the true talent of the pitcher.

Put another way…
What’s more impressive, and likely speaks to more talent:
Striking out the Padres 7-8-9 hitters?
Or striking out Kinsler, Miggy, VMart?
Both are graded the same by FIP…but are you really wanting to consider those events as 100% the same?

Kenny Powers
Guest
Kenny Powers
1 year 6 months ago

Still cynical.

KDL
Guest
KDL
1 year 6 months ago

Still confused

John
Guest
John
1 year 6 months ago

People need to understand that jeff isn’t trying to hate on cole hamels. Everyone knows why you would want to trade for cole hamels; he is on of the best pitchers in baseball on a contract that isn’t much of a discount in aav but is only 4 or 5 years depending on what team gets him. This shorter contract minimizes risk. What isn’t common knowledge is why you may not want to trade for cole hamels and I think that is why jeff is writing the articles. He is reeling us the cons because we all know the pros.

wpDiscuz