Charting Felix

There are two questions on the lips of every resident of greater Seattle right now; what do you think of the new Pearl Jam album, and what’s wrong with King Felix? I can’t help you with the former, but obsessing over a 20-year-old pitcher on a painful-to-watch baseball team, well, that’s right up my alley.

Considering I spent the last few years telling everyone who would listen that Felix Hernandez is the best thing to come around since the thermos—sliced bread is old news—his performance to date in 2006 hasn’t exactly been what I expected. In his first 56.1 innings pitched, he’s given up an astounding 42 runs. For comparison sake, he gave up 26 runs in 84.1 innings last fall. He’s already given up twice as many home runs as he allowed last year, and he’s started two fewer games to date. The King has been more like the Court Jester this time around.

So, I did what any Mariners-obsessed blogger without a family to support would do; I decided to chart every pitch Felix has thrown this year, which means going back and re-watching all of his previous performances on MLB.tv. This is a bit of a time-consuming effort, as you may guess, and it’s still a work in progress. To date, I’ve charted four of his starts in their entirety, logging 405 pitches into a database.

I’m tracking essentially everything I can related to each pitch—velocity, pitch type, and result by batter, inning, and pitch count. Using this data, we can find out just how well Felix is commanding his fastball on 2-1 counts with two out or how many first-pitch curveballs he’s thrown after the fifth inning. You get the idea.

Due to some annoying production decisions by the guys controlling the television cameras (we really don’t need to see a kid eating ice cream when there is a game going on, fellas), I’ve only logged in the pitch type for 372 of those 405 pitches. That gives us a slightly smaller sample, but still one that we should be able to cull decent information from.

The goal, of course, is to answer the question of what is wrong with Felix? However, some would argue the actual answer to that question is nothing. After all, his peripheral statistics are still outstanding—54.4% ground-ball rate, 3.3 walks per game, 8.8 strikeouts per game, all adding up to a 3.65 xFIP, which is good for fourth in the American League among qualified starters.

Only Roy Halladay, Scott Kazmir, and Johan Santana have lower xFIPs in the AL. xFIP has proven to be a quality predictive tool, meaning that if Felix continued to pitch just as he is now, we’d expect him to pitch a lot more like one of the top five starters in the AL and a lot less like a member of the Twins starting rotation (sorry Aaron).

So why is there such a huge discrepancy between Felix’s xFIP and his ERA? Two things jump off the statsheet: Out of 101 qualified starters, he ranks 101st in home runs per fly ball and 95th in Defensive Efficiency Ratio. More of his fly balls are leaving the yard than any other starting pitcher in baseball, and only five guys have fewer of their balls in play being turned into outs.

These are two things that a pitcher has little control over, and so there’s arguably a good amount of luck involved. Every good analyst out there would heavily regress these two numbers to the mean when doing a projection for Felix’s future performances, and like xFIP, would probably come to the conclusion that he isn’t pitching all that badly.

However, while I don’t dispute that we’re likely to see a strong correction in Felix’s performance going forward, I don’t think we can chalk up all of his struggles to bad luck. By looking at the pitch charting data, I believe we can began to answer the real question—why is Felix struggling? We know that a lot of ground balls are missing his defenders and a lot of his fly balls are turning into souvenirs, but xFIP and similar metrics don’t help us understand what the cause may be. They just reassure us that it’s probably not a long-term problem.

That’s where the pitch charting comes in. If we want to attempt to isolate the cause of Felix’s struggles, we need to find out exactly how hitters have been successful against him. To get a more detailed look at his performance, I looked up his expanded splits on ESPN.com and scrolled down to the bottom, where the “By Inning/Pitches” breakdowns are available.

The first thing you’ll notice on that page is opposing batters are hitting .542/.633/.708 off Felix in the first 15 pitches of a ballgame. That’s a 1.342 OPS coming out of the gates. That’s Albert Pujols on a hot streak. You just can’t succeed if guys are starting the game by teeing off on you at that pace.

The performance tails off remarkably in pitches 16-30 (.206/.250/.412) and 31-45 (.200/.306/.300), so this is where the pitch charting can give us a hand. Is Felix doing something differently during the first 15 pitches of a game than he is the rest of the time?

Absolutely. I’ll offer the following table for your amazement:

Pitches	FB	CB	CH	FB%	CB%	CH%	Opp. OPS
1-15	48	5	5	82.76%	8.62%	8.62%	1.342
16-30	29	15	11	52.73%	27.27%	20.00%	0.662
31-45	31	14	11	55.36%	25.00%	19.64%	0.606
46-60	33	15	9	57.89%	26.32%	15.79%	1.008
61-75	30	18	6	55.56%	33.33%	11.11%	1.054
76-90	25	24	9	43.10%	41.38%	15.52%	0.543

In the four games I’ve charted, I’ve logged 58 of the 60 pitches that would constitute the first 15 of each ballgame. A remarkable 83% of those 58 pitches have been fastballs. The Mariners, as an organization, preach establishing the fastball early, but that is just ridiculous. Each start is basically the same. The first inning is all fastballs, all the time. And the opposing hitters are treating it like batting practice.

After the first 15 pitches, Felix starts mixing in his curveball and his change, and hitter effectiveness nosedives. That’s because his fastball is clearly his third-best pitch. As hard as it may be to believe, considering he has a 97 MPH fastball and can run his two-seam diving “sinker” up there at 95 MPH, Felix is at his best when he’s throwing junk. His curveball is as good as there is in the game, and the change-up that he doesn’t use enough is probably among the 10 best in baseball.

A Hardball Times Update
Goodbye for now.

The Mariners, in all their wisdom, however, have decided that every pitcher’s best pitch is their fastball, and they’re going to have Felix feature his fastball early in the ballgame to “set up the offspeed stuff.” Yet, for some reason, they have yet to realize the correlation between “first inning, establish fastball” and “first inning, Felix lit up like Christmas tree.”

If you run a correlation between the Opponents OPS column and the FB% column from the table above, you’ll find the correlation is a remarkable .858. Correlations run from -1 (meaning that they have a perfect inverse relationship) all the way to +1 (a perfect relationship), and +.858 essentially means that a huge amount of the opposing hitters success off of Felix can be directly tied to the amount of fastballs he throws.

More fastballs, more runs. Fewer fastballs, fewer runs. That’s the screaming message from the above table.

However, we need to look deeper. If Felix is throwing his fastball on 3-0 counts, and his breaking stuff on 0-2 counts, it’s only natural that hitters will be teeing off on the fastball. So let’s take a look at the data by ball/strike count:

Pitches	FB	CB	CH	FB%	CB%	CH%	Opp. OPS
0-0	60	24	17	59.41%	23.76%	16.83%	1.063
0-1	21	21	6	43.75%	43.75%	12.50%	0.889
0-2	17	13	3	51.52%	39.39%	9.09%	0.200
1-0	28	6	9	65.12%	13.95%	20.93%	0.880
1-1	10	9	6	40.00%	36.00%	24.00%	1.080
1-2	29	16	3	60.42%	33.33%	6.25%	0.353
2-0	13	0	3	81.25%	0.00%	18.75%	1.222
2-1	11	0	3	78.57%	0.00%	21.43%	1.000
2-2	10	17	3	33.33%	56.67%	10.00%	0.645
3-0	3	0	0	100.00%	0.00%	0.00%	2.000
3-1	7	0	0	100.00%	0.00%	0.00%	1.758
3-2	9	2	2	69.23%	15.38%	15.38%	1.155

Lots of information there—and yes, small sample size caveats heavily apply—but it all has the same general theme. On counts where you’d expect a hitter to be sitting fastball (1-0, 2-0, 2-1, 3-0, 3-1), Felix is complying, throwing significantly more fastballs when behind in the count than when he is ahead in the count. When he gets two strikes, you see an awful lot of curveballs as he goes for the punchout. However, if the hitter is able to work it into a fastball count, they almost always get one, and they’ve been punishing him for it.

What information can we gather from these two tables? To me, watching these games in detail, it is clear that there is a scouting report on Felix Hernandez circulating MLB that reads something like this:

First inning, all fastballs, so go up hacking. After first inning, only swing if it’s straight and hard. Struggles to throw fastball for strikes, so let him fall behind. In a good hitter’s count, swing for the fences—if it’s a breaking ball, you can’t hit it anyway.

That is what we’re seeing hitters do. They are attacking Felix early in the game, knowing that they are going to get a steady diet of fastballs in their first trip to the plate, and then hoping to get him to fall behind later in the game so they can sit dead red again.

When Felix is mixing his pitches effectively and keeping hitters off balance, he’s fine. However, he’s shown a tendency to become predictable at times, and major league hitters are really good at hitting what they know is coming.

There’s good news here for those rooting for the King. This is about as fixable a problem as you’ll find. It should be quite simple for the Mariners to recognize this pattern, make an adjustment to his pitch selection early in the game, and let the dominoes fall into place from there.

So, to answer the original question, what’s wrong with Felix? In my opinion, it’s as simple as pitch selection. As this 20-year-old kid gets more experience, he’ll learn how to pitch away from predictability, and hitters will resume flailing helplessly at his arsenal. When they don’t know what’s coming, they don’t have a chance.


Comments are closed.