Splitting Hairs on Felix Hernandez Season Splits

Felix Hernandez needs no introduction. If I introduced him to you, you would probably pee your pants. That's how no-introduction he is.

Introduction notwithstanding, the man they call The King had his most ineffective season last year since about 2008 when he wasn't just running roughshod over opponents like he was Samwell Tarly playing Red Rover with Kindergartners. But when 18 wins, a 8.52 K/9, and 3.53 ERA represents a disappointment, well you're doing alright. Or as Deion Sanders might tell Doug Baldwin, aight.

You may or may not have the same short-term memory issues that I have, but it doesn't take a deep dive in his stats to recall that Felix was having a particularly fine season through the "first half" (quotations denote arbitrariness of what they call the first half, which isn't a half at all. I digress) -- and on the whole, it was entirely King-like.

Over the first half, he threw 117.1 innings, giving up just 90 hits, 37 earned runs, while striking out 112 opposing batters, good for a 24.1% strikeout rate. He owned a 2.84 ERA (3.33 FIP), and his rate of allowing home runs sat at a stingy enough .77 HR/9. To that point, opponents slashed .211/.279/.323 with a .263 BABIP and just a .268 wOBA. Felix turned basically everyone into Chris Owings.

These results even include getting throttled by the Astros in Houston, giving up eight earned runs in ONE THIRD of an inning. Take that game away from the first half and, well, he's a lot like your college transcript without Physics 101.

His second half? Well, that's what I'm here to talk to you about.

Hernandez threw 84.1 innings, giving up the same number of hits he did over his first 117.1 (90). He gave up more earned runs (42), struck out just 79 batters (21.9% strikeout rate), earning a 4.48 ERA (4.26 FIP) and his HR/9 basically doubled to 1.39. In those 84.1 innings, opponents slashed .270/.329/.456, with a .320 BABIP and .339 wOBA. To be consistent, where everyone in the first half was Chris Owings, everyone in the second was more like Josh Reddick.

My immediate speculation of course is that Felix must have been pitching hurt, and the first place I typically run to is velocity. The results give us nothing, which, I suppose is very good news (data is from Brooks Baseball, and I've omitted their data on his cutter because he threw it about 20 times all year, and I'm not convinced it actually was a cutter at all, but I'll shut up now):

Pitch Type	Velo 1st	Velo 2nd
Fourseam	92.68	92.99
Sinker	92.09	92.51
Change	88.65	88.78
Slider	84.59	85.28
Curve	80.65	80.93

Now, I'm not physician, or an occupational therapist, or one of many well remunerated professionals -- and maybe it's entirely possible you could pitch hurt and still throw hard. However, Felix was, across the board, throwing harder in the second half in which he struggled to achieve the kind of results we're used to. But, in Yoda speak, dead his arm was not.

The second place I typically run to is release point. Because when it hurts to throw, you do funny things. Punch yourself in your throwing shoulder 100 times and try to throw a baseball. You'll look funny.

Using the data from TexasLeaguers.com, I looked at two of his most common pitches, his slider and his change. Each is an overlay of the first and second half release point, so just look for the date if you want your internal voice to say "before" and "after".

If anything, Hernandez does play it pretty fast and loose on his release points, and that might be wrapped up in his position on the pitching rubber -- but looking at these two pitches, he's consistently in the same blob from first half to next. When there's a clear drop in release point, you can usually tell there's at least something going on -- whether it is a new approach or your shoulder feels like it's getting stabbed every time you throw the ball. And for what it's worth, I looked at his fourseam fastball and curve, and there's no appreciable difference there either. But we're going to be .gif heavy today, so let's save some space, eh?

So velocity is good. Release point is good. Did anything change?

Certainly not a smoking gun, but there was a clear point where Hernandez and his brain trust decided to alter his repertoire. Whether that started on 7/19, I don't know -- but it certainly could have been a discussion over the All-Star break. Because the second half repertoire changed in a pretty notable way:

Pitch Type	1st Half	2nd Half
Fourseam	12.34%	15.20%
Sinker	32.30%	23.92%
Change	29.70%	26.23%
Slider	4.90%	7.41%
Curve	20.24%	25.23%

So almost a 10% drop in usage of his sinker, but still a significant pitch -- and a subsequent increase in his curveball, which by way of pitch values is easily his best. Perhaps an investment banker got a hold of him and told him to diversify, I don't know. But this seems intentional, not random.

In looking at his first half results, it's not very hard to understand why they might have made this change:

Pitch Type	Count	HR	BAA	SLG	ISO	BABIP		Observation
Fourseam	214	2	0.190	0.354	0.165	0.289		Solid
Sinker	560	6	0.288	0.470	0.182	0.276		Not very good
Change	515	1	0.210	0.254	0.044	0.301		Excellent
Slider	85	1	0.300	0.600	0.300	0.250		oof
Curve	351	0	0.082	0.082	0.000	0.135		Wow

His sinker was getting torched and his curve and change were both getting consistent outs, with the curve making batters look just silly.

So he makes the modification to fewer sinkers and more curves, a few more fourseam fastballs and sliders. What happened?

Pitch Type	Count	HR	BAA	SLG	ISO	BABIP		Observation
Fourseam	197	2	0.276	0.448	0.172	0.326		Pretty bad
Sinker	310	5	0.349	0.635	0.286	0.333		Very, very bad
Change	340	5	0.270	0.500	0.230	0.310		Wait, what happened here?
Slider	96	1	0.444	0.667	0.222	0.467		Usher!
Curve	327		0.175	0.225	0.050	0.280		Not as good, but still very good

The curve was still really quite good but his change became a disaster and nothing else was very good either.

Since we're here and we're geeking out on numbers, I find the data on right handed batters versus left handed batters kind of useful as well. Here are splits and results by offering versus left handed batters:

First Half
Pitch Type	Count	HR	BAA	SLG	ISO	BABIP	Observation
Fourseam	129		0.163	0.302	0.140	0.269	Very good
Sinker	291	5	0.328	0.627	0.299	0.293	Very horrible
Change	233	1	0.212	0.288	0.076	0.232	Good
Curve	228		0.081	0.081	0.000	0.125	Great
2nd Half
Pitch Type	Count	HR	BAA	SLG	ISO	BABIP
Fourseam	123		0.235	0.353	0.118	0.333	Ok
Sinker	160	2	0.250	0.438	0.188	0.222	Bad, but better
Change	155	4	0.280	0.600	0.320	0.250	Terrible
Curve	193		0.244	0.311	0.067	0.423	Ok, worse than 1st

(He threw his slider very little to LHB).

And then versus right handed batters by split and offering:

1st half
Pitch Type	Count	BAA	SLG	ISO	BABIP	Observations
Fourseam	85	0.222	0.417	0.194	0.316	OK
Sinker	269	0.246	0.308	0.062	0.259	Pretty good
Change	282	0.208	0.222	0.014	0.405	Very very good
Slider	85	0.300	0.600	0.300	0.250	Bad
Curve	123	0.083	0.083	0.000	0.154	Yes, we want this
2nd half
Pitch Type	Count	BAA	SLG	ISO	BABIP
Fourseam	74	0.333	0.583	0.250	0.316	Horrible horrible
Sinker	150	0.452	0.839	0.387	0.458	Horrible horrible
Change	185	0.260	0.400	0.140	0.387	Not very good at all
Slider	96	0.444	0.667	0.222	0.467	Stop it
Curve	134	0.086	0.114	0.029	0.125	Awesome

There's kind of a common finding here and that's the change-up. In general, his second half offerings weren't very good other than his otherworldly curveball, but the change went from good to bad overnight -- and that's not helpful when you throw it more than a quarter of the time.

Something of note about his change from the first half to the second half was general movement -- in that it was actually moving far more horizontally and far less vertically. And as you see above, the results vs. RHB weren't very good but the results vs. LHB went from pretty darn good to pretty awful. Below are the frequencies, velocity and movement of his change to LHB first half and second half:

Pitch Type	Count	Freq	Velo (mph)	pfx HMov (in.)	pfx VMov (in.)
1st half CH	233	26.42%	88.43	-6.21	1.32
2nd half CH	155	23.96%	88.42	-7.07	0.76

But movement is good, right? Certainly -- if you know where it's going. As I'm sure you're aware, they say the hardest thing to do in sports is hit a baseball, but just ask Randy Johnson about the nuances of a quarter inch release point -- it's the difference between a strike on the corner and a ball over John Kruk's head (yeah, I know that was for theater). So in the second half of the season, The King's change was moving almost a full inch more horizontally and a half inch less vertically. Which is super, unless you can't hit your spots with it anymore, which is ostensibly to blame here.

It's an imperfect example, but his change reminded me of the game against the Boston Red Sox in which he gave up three home runs -- two of them were clearly on a change-up, one to Pablo Sandoval and another to Jackie Bradley Jr. Against Bradley Jr., you don't see that horizontal movement as much, but the lack of vertical movement leaves it up as a meatball:

This isn't gotcha stuff. This is nuanced. Felix wasn't very good in the second half of 2015, and I'm perfectly willing to accept that it was dumb luck, bad juju, small sample size, what have you. The obvious indicators don't suggest any kind of major arm damage, so yay. But the data certainly suggests there was a demonstrable change in the way  Hernandez chose to approach hitters by way of his repertoire -- and for one reason or another, that led him from an excellent first half to a very mediocre second half. Heading into 2016, I'm certainly willing to bet we see that first half Felix again, and in order for the Mariners to truly contend, Stottlemyre and company will need to figure out how to get that production over 34 starts.

Go M's.