Jered Weaver versus King Felix is not even close

Rich Lederer decided to compare the careers of Felix Hernandez and Jered Weaver, two pitchers whom he claims have much closer numbers than analysts like Dave Cameron are giving credit for. Maybe they are, but I think this line is unforgivable:

Based on age, velocity, and batted ball info, maybe Hernandez projects as a better pitcher than Weaver. But the reality is that Felix has not outpitched Jered to this point. Or, if he has, the difference between the two has been miniscule [sic].

I mean, huh? Hernandez does not only project to be a better pitcher, he far and away has been the better pitcher, and it’s not very close!

Felix Hernandez xFIP 2005-2010:

2005: 2.76
2006: 3.34
2007: 3.34
2008: 3.87
2009: 3.42
2010: 3.87

Jered Weaver xFIP 2005-2010:

2005: N/A
2006: 4.30
2007: 4.76
2008: 4.28
2009: 4.48
2010: 3.22

If somebody wants to compare their SIERA’s, I’m pretty sure Felix will win in a landslide as well. I mean, Weaver is absolutely owned by Felix despite being older in every year pitched (small sample size for 2010, of course). I don’t think it’s possible to rule out Weaver becoming a #1 starter, but with his track record, age, and skill set, it’s very, very unlikely. I find it completely disingenuous that Rich is using Win-Loss, WHIP, H/9 (!?!?), and ERA when there are so many better stats.

In fact, the graphic he puts at the top of the page (see below here) is so, so deceptive. This is exactly the type of maneuver that is made to make two players look much more similar than they are. Let’s see why:

Okay, so Felix has a lot more innings pitched. Better for him. He has more wins, but who cares. He has a lower ERA, but that also is murky. He has a better ERA+, but that is just derivative of ERA. His WHIP is much worse, but that’s probably due partly to the next stat, H/9, which is a silly stat completely dependent on defense. Felix gives up less homers, but this also deals greatly with luck/Safeco. He walks more people, but strikes them outs much more as well. Oh, and then the SO/BB to cap it off, which is completely redundant.

Look, Felix has been hands down the better pitcher so far, at a younger age, and will continue to be. I like picking on #6org as much as the next guy, but give me a break.

Update:

Taken from a comment from the article:

FELIX HERNANDEZ                  JERED WEAVER             DIFFERENCE (JW-FH)
YR AGE IP    FIP  xFIP WAR  AGE IP    FIP  xFIP WAR  AGE IP    FIP   xFIP  WAR
05 19   84.1 2.85 2.76 2.6  23                       +4  -84.1             -2.6
06 20  191.0 3.91 3.34 3.8  24  123.0 3.90 4.30 2.6  +4  -68.0 -0.01 +0.96 -1.2
07 21  190.1 3.75 3.34 4.1  25  161.0 4.06 4.76 3.1  +4  -29.1 +0.31 +1.42 -1.0
08 22  200.2 3.80 3.87 3.9  26  176.2 3.90 4.28 3.4  +4  -24.0 +0.10 +0.41 -0.5
09 23  238.2 3.09 3.42 6.9  27  211.0 4.04 4.48 3.9  +4  -27.2 +0.95 +1.06 -3.0
10 24   44.0 4.52 3.87 0.3  28   44.0 3.11 3.22 1.1  +4    0.0 -1.41 -0.65 +0.8

And taken from the comments section here:

SIERA
2006: Felix 3.28 Weaver 3.78
2007: Felix 3.10 Weaver 4.34
2008: Felix 3.84 Weaver 3.89
2009: Felix 3.39 Weaver 4.10
2010: Felix 3.83 Weaver 2.78

Using Recurrent Neural Networks to Predict Player Performance
Technology is rapidly advancing possibilities in decision-making.

Print This Post
Pat Andriola is an Analyst at Bloomberg Sports who formerly worked in Major League Baseball's Labor Relations Department. You can contact him at Patrick.Andriola@tufts.edu or follow him on Twitter @tuftspat

Leave a Reply

30 Comments on "Jered Weaver versus King Felix is not even close"

Notify of
avatar
Sort by:   newest | oldest | most voted
Colin Wyers
Guest
Colin Wyers

Why xFIP? Why ignore Weaver’s lower HR/FB rate?

Pat Andriola
Guest
Pat Andriola

Colin, a lot has to do with this:

http://www.hardballtimes.com/main/article/how-well-can-we-predict-era/

I like to regress homerun rates. Here are the FIP numbers even if you believe Weaver has the ability to limit his HR/FB to a lower-than-average rate:

Jered Weaver:

2005: N/A
2006: 3.90
2007: 4.06
2008: 3.90
2009: 4.04
2010: 3.11

Felix Hernandez:

2005: 2.85
2006: 3.91
2007: 3.75
2008: 3.80
2009: 3.09
2010: 4.52

It’s closer, as I said, but you simply cannot argue a minuscule difference. FIP-based WAR, which Fangraphs uses, has Felix worth 7.5 WAR more than Weaver over his career so far.

Some guy
Guest
Some guy

This article is why people hate stats nerds.

Try to be less of a indignant douche next time.

TheTeeto
Guest
TheTeeto

What a terrible argument.  You only use xFIP as your retort?  You argue that the numbers aren’t really that close, when they really are?  The simple argument here is age—Felix put up his numbers in the bigs when most players are in the minors.  There’s a big difference between a pitcher at 23-27 than there is from 20-24.  That’s all you had to say…

Colin Wyers
Guest
Colin Wyers

If we prorate Fangraphs WAR out over 180 IP, we get:

Felix: 4.1
Weaver: 3.5

That’s a difference of about .6 wins a season (ish – 180 IP is just a “back-of-the-envelope” number).

Is that significant? How do you define “significant?” I think given what we know about the accuracy of our ability to estimate runs from components and wins from runs, a half-win difference over a season is within measurement error.

Pat Andriola
Guest
Pat Andriola

TheTeeto,

That’s just not true. In terms of xFIP they are not close at all, and considering we have FIP-based WAR which takes into account innings pitched we can see that Felix has been worth 7.5 more WAR over his career.

For even more context, Baseball Projection’s numbers:

Jered Weaver career WAR: 14.6
Felix Hernandez career WAR: 18.2

That’s ~3.6 WAR, almost a full good season’s worth, between them.

Pat Andriola
Guest
Pat Andriola

Colin,

Should we prorate it to ~180? Why not give Felix credit for pitching well over a longer period of time?

Again, I still think xFIP is a better indicator, and would like to know SIERA numbers.

I think .6 wins per season (which is about what I got from Baseball Projection’s numbers), while conservative, is still enough to say that Felix has been the better pitcher (age taken into account to), or at least that the difference is more than miniscule.

elwin
Guest
elwin

SIERA
2006: Felix 3.28 Weaver 3.78
2007: Felix 3.10 Weaver 4.34
2008: Felix 3.84 Weaver 3.89
2009: Felix 3.39 Weaver 4.10
2010: Felix 3.83 Weaver 2.78

Sean Smith
Guest
Sean Smith

Felix may project better, but for actual performance so far Weaver has been very close – 0.7 WAR over 4 years since Weaver’s debut.

As for projection, it’s too much of an unknown anyway.  Pitcher projections are nowhere near as reliable as hitter projections.  A big reason is that nobody knows how to project health.  Two of Felix’s advantages – doing this at a younger age and throwing more innings per year – may actually make him a bigger risk for future injury.

Mitch Brannon
Guest
Mitch Brannon
I agree with the article in general, however isn’t saying H/9 is “a silly stat completely dependent on defense” a little simplistic? Surely the pitcher has at least something to do with the frequency and velocity at which his pitches come off of bats. You might have done better using “largely” instead of “completely.” I think what Some Guy may have inelegantly tried to say is while your argument has merits, you probably could have made it without coming across as attacking Rich Lederer. Lederer’s point seems to have been that while Felix is the better pitcher, it isn’t all… Read more »
Pat Andriola
Guest
Pat Andriola

Mitch,

Yes, “largely” is a better word.

In terms of re: being kind of a dick, I certainly didn’t mean to come off that way (may be stressed a little by final exams) and admire Rich’s work for sure; however, this was just a pretty unscientific and I think slightly sleezy way to compare the two pitchers.

Mitch Brannon
Guest
Mitch Brannon

You’re right, he used some silly stats and you did well to note the double-counting. As a stats guy who deals with government types, I’m sensitive to how easy it is to come off as sounding smug or defensive, even if you’re not.

MangoLiger
Guest
MangoLiger

I guess what I’m saying is that these results shouldn’t come as a shock to those of us who think that Felix is clearly the better pitcher. It’s just a reminder that, while he had a bright future, he wasn’t clearly the better pitcher until very very recently.

Note that you can get the same results with Zack Greinke, who has an almost identical career ERA to Weaver. People talk about small sample size a lot, but large sample size can be a problem too—if you are trying to measure a non-stationary signal.

Nick Steiner
Guest
Nick Steiner
I agree with Colin.  The only reason xFIP was more predictive than FIP in the study Colin wrote was that he used single season sample sizes.  Weaver has over 700 career innings pitched in a home ballpark with a very close to neutral HR park factor http://www.hardballtimes.com/main/blog_article/hr-fb-park-factors/ So I see no reason to regress his HR/FB all the way back to league average.  xFIP is seriously only useful in a small sample size, once you get to 5-6 seasons it doesn’t have much value.  Weaver’s career FIP is .35 points higher, and the two’s career ERA+, which adjusts for Safeco… Read more »
Colin Wyers
Guest
Colin Wyers
Using CHONE’s ERA projections: Felix: 3.38 Weaver: 4.10 That’s a difference of 0.72 runs. Sounds pretty cut and dried, dunnit? But it isn’t. For ‘07, the RMSE between CHONE and observed was 1.084. (I should be using more recent figures, but I can’t find those readily and for the purposes of illustration it doesn’t matter. Actually to be pedantic each projection has several factors that influence it’s error and that should be accounted for, but for simple illustration that REALLY doesn’t matter.) When the difference between two players’ forecasts is smaller than the estimated error for either forecast – that… Read more »
Nick Steiner
Guest
Nick Steiner
I think the ERA+ argument is the strongest for Weaver vs. Felix.  The ONLY reason we use FIP as an evaluative tool is that ERA has stuff outside the pitchers control which are distributed both randomly and with bias.  But ERA also includes MANY inside the pitchers control that FIP does not account for.  Over 700+ innings, the randomness of ERA should start to fade away and the only things remaining are the biases.  Park factors are one bias, but ERA+ strips that away (or at least it should).  The other bias is defense. According to FanGraphs, the Mariner’s have… Read more »
Colin Wyers
Guest
Colin Wyers

There are other biases in ERA that persist over time, the greatest being the bias in unearned runs. (Ground ball pitchers will have more unearned – and thus fewer earned – runs than a similarly talented FB pitcher, over an arbitrarily large sample.)

But what xFIP does is it regresses several pitching skills 100% to the mean. Over a small sample, this is probably indistinguishable from proper regression – this makes it a better projection then observed results.

Over the samples involved, though, there’s no reason for us to prefer xFIP over more expressive measures of past pitching performance.

Nick Steiner
Guest
Nick Steiner
Yes forgot about that one.  But the ground ball bias in ERA is one that actually works against Felix right?  He’s a ground ball pitcher so his RA will be higher than his ERA relative to the league. Weaver has a career 3.84 RA, while Felix has a 3.89 RA!  So Weaver has allowed fewer runs than Felix despite pitching in a decidedly more hitter friendly ball park with defenses of basically equal quality.  Rich’s comparison looks spot on to me right now.  Right about xFIP.  I can live with the stat in a small sample size because it’s simple… Read more »
MangoLiger
Guest
MangoLiger

“Clearly better” was my subjective opinion. Let’s try this: Felix is better than Weaver with a 74.672% probability (assuming a normal distribution).

Are you claiming that a pitcher needs a CHONE-projected ERA of 3.016 to be considered better than Jered Weaver? Because there are no starting pitchers that meet that criteria.

Colin Wyers
Guest
Colin Wyers

And for this:

For even more context, Baseball Projection’s numbers:

Jered Weaver career WAR: 14.6
Felix Hernandez career WAR: 18.2

That’s ~3.6 WAR, almost a full good season’s worth, between them.

Yes, it’s a full good season’s worth – 2005, in fact, where Hernandez had 2.9 WAR. Weaver made his debut in 2006. From that point on, there’s .7 rWAR between them.

I think that’s a damn sight different than “not even close.”

Pat Andriola
Guest
Pat Andriola

Yes, but as I said, I prefer Fangraphs’ WAR, which is at ~7.5.

In terms of Baseball Projections’ numbers, that would be closer than “not even close,” but also not at Lederer’s “minuscule difference.” Somewhere in between.

Nick Steiner
Guest
Nick Steiner

Pat, can you explain why you think FIP (FanGraphs WAR) is better than Run Average (Baseball Projection WAR) over the sample sizes involved? 

I’ve already shown that there is not a defensive bias involved, and because Baseball Projection WAR is park adjusted, there is no park bias.  That leaves the difference between FanGraphs WAR and BP WAR a combination of pitcher skill (controlling timing, controlling quality of balls in play, holding runners, etc.) and random variation. I would think that over 700 – 900 innings the amount of skill captured is more prevalent than the amount of random variation.

Sean Smith
Guest
Sean Smith
Those ERA projections include the efforts of the defense behind them.  Most people seem to think Seattle’s defense is better.  The ERAn – neutral context for each are 3.44 for Hernandez, 3.93 for Weaver.  Still an edge for Hernandez, but a bit closer.  The main reasons for his advantege is that Felix’s best year is also his most recent, and his groundball percentage. Hernandez projects better than Weaver.  I don’t think anyone has suggested otherwise, but the difference in their actual performance to date has been much smaller than the 0.50 gap in projected ERA. As for last season, anyone… Read more »
Pat Andriola
Guest
Pat Andriola

Nick, I could very well be wrong, but what about th serious differences in SIERA posted above, which should take into account a lot of the things we’re talking about.

Nick Steiner
Guest
Nick Steiner

SIERA regresses HR/FB 100% giving it the same problem as xFIP.  Given that’s where a lot of Weaver’s value comes from, it’s going to systematically underrate him.

MangoLiger
Guest
MangoLiger
I think what we’re glossing over here is that Felix has gotten much better over the last five years and Weaver has been more level (with some improvement), so of course when you average everything out it looks about the same. The implicit assumption being that “true talent” was static that whole time (from age 19 to 24 !), so we can just average away the 7 WAR season as noise. That assumption is absurd, and if you abandon it, you have to weight recent performance more heavily, which effectively also reduces your sample size, but also blows up any… Read more »
jrdo410
Guest
jrdo410

Felix was just traded for Weaver – which fans are pumped and which are pooping their pants?  Thats about as much analysis as you need.

TheTeeto
Guest
TheTeeto

Pat Andriola said…
TheTeeto,
That’s just not true…

What’s not true?  Did you even read my post?  My point was the entire argument is ridiculous because Weaver is almost 4 years older.  Comparing the two is silly and pointless.

Pat Andriola
Guest
Pat Andriola

TheTeeto,

I took your “what a terrible argument” to be in reference to mine based on what came after. That’s all.

Voxx
Guest
Voxx
It’s hardly that cut and dried, jrodo. That factors in age and contract, which shouldn’t be heavily taken into account in this discussion.  The two pitchers are honestly a lot closer than most give them credit for.  Is Felix better? Yes. But while xFip is a good stat for smaller sample sizes, I do believe ERA and, especially, ERA+ are superior stats when looking at sample sizes as large as 4-5 seasons.  However, there’s one large caveat to be had here.  Felix has pitched better in the last year, despite them being fairly similar when their entire careers are taken… Read more »
wpDiscuz