Archive for pitch analysis

With the advent of the Statcast era, a lot of research has been done in attempts to measure the effectiveness of a particular pitch based on its flight characteristics. As has been noted in the past, quantifying a pitcher’s stuff and command is no easy task. However, over the past few months I have worked to build my own models in an attempt to evaluate the “filth” of any given pitch, taking more of a probability-based approach. I introduce to you my Probabilistic Pitch Scores and xWhiff metrics.

When evaluating the quality of a particular pitch, I focused my interest on three different binary outcome variables: whether or not the batter swung at a pitch, whether or not the batter whiffed on a pitch, and whether or not a pitch was thrown for a strike. Thus, my goal was to train three different types of classification models corresponding to each of these variables: a swing, a miss, and a called strike. For the actual outcomes of these models, I was less interested in the model’s decision and more interested in the predicted probability. For example, if a batter swings on a pitch with given flight characteristics, what is the probability that he will whiff? These probabilities were utilized as the basis of my metrics.

Read the rest of this entry »

Last July when the Oakland Athletics traded for Blake Treinen, Jesus Luzardo, and Sheldon Neuse, providing the Washington Nationals the services of relievers Sean Doolittle and Ryan Madson, the Athletics trade was more notably about the prospects of Luzardo’s pitching and Neuse at third base. The Athletics are in the middle of a definitive rebuild; outside of the Athletics organization, the Treinen piece of the trade was simply for his services as a simple bullpen body to hit average (hopefully) and pander on through the remnant of the season as Doolittle and Madson’s replacement.

Contextually, when the trade occurred last July, Treinen was 28 and hitting his ceiling in the Nationals development methodology. He lost hit job in April of 2017 then slowly settled into an awful 5.73 ERA with 48 allowed hits in the first half. There is no magical analytic which explains why he was bad, no pedantically bad situational deterioration – Treinen was simply bad.

Specifically, Treinen was bland on his first pitch (which is a telling sign of a pitcher struggling), lofting sweet, contact worthy pitches to the upper zone. Overall, batters swung and made contact throughout his first-pitch zones. Hence, the same area that created his bizarre downward trend was the first area where Treinen began his 2018 correction. He has cut high zone pitches on the first-pitch count, aptly cutting contact to the high zone. Batters are attacking his first pitch less, taking (or trying to take) a more principled approach to seeing him out.

The batting delay has created Treinen’s most formidable analytical point: the second highest and lowest qualified-reliever swinging strike rate and ERA at 18.8 percent and .93, respectively.

Any improvement this stark demands a resurgence across all pitching categories. First, Treinen has begun to shift his fastball placement. In 2017, on a micro fastball scale, Treinen offered either the fastball outside (44 percent balls) or distinctly inside (76.1 percent contact), failing to hit the lateral sides of the inside zone. That crisp distinctness allowed batters to perceive where his fastball was going to land, allowing a 156 wRC+. In 2018, he has avoided letting the fastball fall distinctly outside of the zone (30 percent balls, 72.7 percent contact), thus offering more variety within more controlled movement. His fastball has kept batters to a negative 20 wRC+.

When batters have been contacting his fastball, they have a 47.1 percent fly-ball rate. Perceptually, this is an alarming rate, less that placement becomes intrinsically important. In a string of reactions, Treinen is enforcing a greater chase rate (38.7 percent, increase from 30.8 percent) while decreasing his chase contact rate from 52.2 to 38.6 percent. This has cut the ability of batters to find barrel contact (2.1 percent), thus cutting his opposite field hit-rate to 19.8 percent. In short, Treinen is deriving contact that is easy to field.

Hypothetically, this might be a philosophical adjunct to the Athletics analytic mantra. The team is shifting on 30.4 percent of left-handed batters, an increase from 17.1 percent last season with the Washington Nationals. Neither team shift at a dramatic rate, but there is a slight difference between the Nationals fielding chart and the Athletics fielding chart behind Treinen. The slight difference may not be the main reason, but the Athletics awareness of how to help Treinen with more movement is, at minimum, an interesting note.

However, for all the jovial notation Treinen’s fastball is receiving, his main-pitch, the sinker, deserves even more praise. He has kept batters to a .200 average, already hitting 17 strikeouts on the sinker for a 56 wRC+. Much like the fastball, the sinker is moving less, but in a crisper fashion (9.7 rating down from 11.9). On a meta point, Treinen is simply more confident and educated in his pitching approach – on a simple eye-test, he is perceptually prepared where to throw. Scrupulous timing and more variety in placement of his sinker has lowered contact to 71.1 percent (was an egregious 86.2 percent last season). The one data-point which fundamentally incorporates how good his sinker has been is the increase in outside swing percentage (43 percent) while decreasing outside-contact (52.1 percent).

In short, pitches that were not meant to be hit, were being hit in 2018, and Treinen has prevented that from occurring. Fundamentally, this is on the breadth of a slight mechanical edit observable with a release point grouping to the right. In the example provided, between 2017 and 2018, Treinen is releasing his pitches with more elevation, signifying a change to his release philosophy.

The Athletics somehow have tapped into Treinen to bring out the best of him; the only question they have to answer is whether he is trade-bait or a long-term staple to the bullpen.

On the eve of July, the month of definitive do or die competition, the Colorado Rockies optioned their opening day starter, Jon Gray, to Triple A Baseball, putting a temporary halt to a season which should have been superlative. Gray was positioned to be the Rockies Ace pitcher, the de facto strike out machine. He did so, posting an MLB fifth best 11.64 K/9 with a WAR of 2.5, breaking most of his projections.

Yet, Gray’s demise and optioning is a reminder that a pitcher’s job, in the end, is to play the averages and get out of situational disaster to end innings with the formidable zero still on the board. Gray was pitiful at cleaning up the base path with a 63.1 percent left-on-base percentage. His 5.77 ERA was slowly flowing up since the beginning of the season. His MLB best 14.33 K/9 for June was met with only 27 innings of pitching, 62.2 percent left-on-base, and an ERA of six. Troubled outings and difficulty finishing starts were trending, not the outlier.

There is an odd note, however, on Gray’s optioning to Triple A. German Marquez, who finished eight innings of one-run pitching in a 3-1 win over the Los Angeles Dodgers last night, has even more developmental problems. Marquez had an even more troublesome June analytically, with an equal 62.2 percent left-on-base, an era of 6.75, a FIP of 5.26, and nine home runs allowed. Hypothetically, there are two reasons the Rockies have decided to option their ‘best’ pitcher instead of the more developmental Marquez. First, the Rockies may be admitting they are going to be sellers at the deadline, and this is the beginning to positioning certain pitchers for sale. However, this would be a very un-Rockies tact to take for a team who has been stubbornly boisterous about ‘competing’. Second, Gray may be more fixable than Marquez, with a quick stint in AAA allowing him to resolve fundamental mechanics away from the stench of scrutiny. (This hypothetical is what the remnant of the article will focus on). Or, it may be a mix of both hypotheticals, with time telling which carries more weight in organizational decisions.

Optioning Gray becomes a matter of establishing finishing touches, helping him to make his strikeouts effective. In a matter of plate discipline, batters are attacking zone pitches 5.8 percent more than last season, back to a career average of 65.8 percent. Yet, he is throwing less to the zone (43.8 percent) while batters are making drastically less contact (80 percent in 2017, 70.2 percent in 2018). All those numbers lead up to a compelling 13.2 swinging-strike percentage and the conclusion Gray ought to be even better than last season when he finished with a 3.67 ERA and a 3.18 FIP.

The pitch arsenal has seen some slight edits, with a cut to fastballs and a rise in slider percentage of five both ways. Velocity has remained mechanically the same, thus, batters should not be exploiting his pitches at this rate. The problem, however, becomes that batters are exploiting this edit by forcing perceptual chaos on Gray, in which he doubles down on throwing distinct pitches with little movement variation.

Gray’s slider placement, on a meta level, has not changed, nor has the contact basis. However, what has dramatically shifted between 2017 and 2018 is how batters are making contact. In 2017, there were three zones which batters had near .100 averages against Gray; in 2018, that rating has gone up to seven, with an egregious .250 to double down on the pain. Strategically, Gray attacks the shadow of the zone with his slider when ahead and moves up to inside the zone when behind. It is not so much a matter of controlling placement but controlling the count and situation.

A false sense of security in the slider has created situational derisiveness on Gray’s fastball. Gray has developed a distinction with his slider as his ‘shadow’ pitch (3.3 PITCHf/x movement rating, down from 5.4) while his fastball is his ‘heart’ pitch (8.7 PITCHf/x movement, down from 11.3). Thus, when in trouble, Gray’s intents become clear, and his fastballs have been straying more inside. The brevity in fastball movement has lead to batters grouping his fastball and hitting at a .172 average from the middle to right, lower portion of the zone.

In short, Gray’s problems result not from mechanical duplicity, but from strategic duplicity – a loss of confidence. Since Gray’s goal is to hit strikeouts, when bases are empty, he has a 13.84 K/9 rating; when runners are on, his K/9 falls to 8.61 while BB/9 raise to 4.19. With runners in scoring position, his FIP takes a jump to 4.94.

Situationally, the flop begins when situational leverage ebbs from low to medium with 96.6 and 61.8 percent left-on-base, respectively. Unfortunately for Gray, his troubles begin regardless of time through the order. He has allowed 21,19, and 21 runs through the first, second, and third time through the order, respectively. The underlying tact of how batters destroy Gray can be seen in slugging at a .390, .485, and .527 percentage through the order.

What can Gray fix in Triple A baseball? In two words, strategical variety. Despite being able to land more strikeouts, Gray has become less effective by staying stuck in a rut, unable (or unwilling) to hide his slider and fastball with movement. Situational aptitude and learning how to pattern his pitches will be essential to turning Gray into an effective strikeout machine.

After signing a massive seven-year, 217-million-dollar contract with the Red Sox this past offseason, David Price got off to a slow start. After his May 7th start against the Yankees in which he gave up six earned runs in just 4.2 innings, Price’s ERA stood at a whopping 6.75 yet his peripherals remained strong. He had a 2.98 FIP and 11.5 K/9. However, he was giving up hard contact over 41 percent of the time. The immediate fix was a mechanical issue noticed by Dustin Pedroia that was limiting Price’s leg lift and diminishing his velocity. Frustrated with his failures, Price vowed to be better.

And better he has been. After throwing a gem in Sunday’s win over the Mariners where he went eight innings allowing his only run on a solo shot by Franklin Gutierrez, Price lowered his season ERA to a still high 4.24 and had his eighth straight quality start. Over those eight starts, Price has been much better, allowing 16 runs over 58.1 innings for an ERA of 2.47. During this stretch, he has a 3.88 FIP and 8.6 K/9 and has only allowed hard contact around 27 percent of the time. Although his strikeouts have gone down and his FIP went up due to his decrease in strikeouts to go with an increase in home runs allowed, Price has limited the amount of hard contact he has given up. This can be seen in the BABIP over the two stretches. In his first seven starts, his BABIP against was around .370, while in this current eight-start stretch it is hovering around .230.

This in turn, has allowed him to be very successful while pitching to contact. His biggest issue remains his ability to keep the ball in the park. Over his last eight starts, Price has allowed at least one home run in seven of them. So while he has limited hard contact against him, the few mistakes that he makes each game are punished. Despite this increase in home runs allowed, he continues to pitch well and go deep into games, allowing the Red Sox bullpen a chance to recover after the consistently shaky starts from their 4th and 5th starters.

There are a few main reasons to this improvement. The first was his ability to regain his velocity. Looking at his velocity each month thanks to data from Brooks Baseball, there is a small but steady increase in his average four-seam and sinker velocity. Before May 8th, his velocity was low by his standards. Typically a pitcher averaging 94 to 95 MPH with his fastball, he had been sitting 93 MPH.

Although just a small dip in velocity, it made him much more hittable.

Since May 8th, his velocity has been back on the rise.

After the mechanical change, his four-seam has been averaging around 94 MPH and his sinker has been averaging around 93 MPH, but still slightly up from what it was before. Although it is a small increase, this added velocity has helped Price dominate hitters, gain confidence, and re-establish himself as an ace.

Another key factor in this improvement has been his pitch usage. Using pitch data from Brooks Baseball, I was able to look at Price’s pitch usage. In his first seven starts, Price relied on mixing different types of fastballs with his main offspeed pitch being a change-up while also displaying the occasional curve.

His four-seam was used around 28 percent of the time yet it lacked the movement displayed by his cutter and sinker. The high four-seam usage to go with decreased velocity spelled trouble for Price.

However, since May 8th, Price has made an adjustment displayed by the fact that he is now using his sinker as his primary pitch while also using his four-seam far less frequently.

His sinker is now used around 36 percent of the time compared to his four-seam being used around nine percent of the time. With this added movement and velocity, Price has been able to be more effective while keeping the use of his curve, cutter, and changeup around the same. This simple switch from a four-seam to a sinker has allowed him to go on a tear.

Looking forward, the Red Sox need Price to continue to be the pitcher that he has been over his last eight starts. They are paying him ace money and he is expected to pitch like one down the stretch as Boston hopes to continue their great turnaround year. If Price continues to have outings like these, the Sox should like their chances come October with him taking the mound with their season on the line.

The last time Matt Shoemaker allowed more than two runs in an outing was June 19. Since then, he’s pitched 37 1/3 innings, allowing just seven earned runs. He has 35 strikeouts compared to just 11 walks, leading to a 2.88 FIP. He’s been even better when just isolating the numbers in his three starts since the All-Star break, with 27/6 K/BB and a 1.36 FIP, although, to be fair, that is an incredibly small sample. For comparison’s sake, his FIP through June 19 was 4.70.

So has there been a change in Shoemaker’s game, or has his streak been a fluke? Well, I wouldn’t be writing this if it was the latter, as I’m sure you could’ve guessed (although if you weren’t able to guess who the article was about after the first paragraph, perhaps I’m overestimating you). There’s been a significant change in the way Shoemaker has approached batters. Take a look at his pitch type chart through June 19, courtesy of Baseball Savant:

And then take a look at the data since then:

Through June 19, Shoemaker threw his fastball (four-seam and two-seam) 51.6% of the time. Since then, it’s been 56.9% of the time. Comparing these two proportions with a two-tailed Z test yields a p-value of .034, significant at the .05 level, showing that there has indeed been in a difference in the amount of fastballs he’s thrown.

Of course, throwing more fastballs doesn’t translate to a drop in FIP of over 3 points. That is, unless, those fastballs are of higher quality. And, class, what’s the most important aspect of a fastball? Hopefully you were at least able to guess this one: the velocity. Which, naturally, is the next thing I looked at.

Again, I used Baseball Savant’s PITCHf/x data. Narrowing the results to just fastballs, here are the velocities of Shoemaker’s pitches this year:

At the beginning of the season, Shoemaker’s average fastball velocity hovered right above 88 mph. Since then, it’s steadily risen, and there’s a clear jump about two-thirds of the way into the season (note that this time would be remarkably near June 19). After the jump, his average velocity has hung closer to the 92 mph range, further away from Jered Weaver status. FanGraphs data shows the same thing:

Note, this data also shows Shoemaker’s average velocity from 2014, when he had a 3.04 ERA and a 3.19 SIERA. This image confirms the steady increase in velocity of Shoemaker’s fastball, as it has recently resided at or even above its value from last year’s productive season. There have been clear results from this change, especially in the form of whiff rate, and predictably, strikeouts. Through June 19, Shoemaker’s whiff rate sat at a mediocre 10.5%.

Since the All-Star break, this is what that breakdown looks like:

You might notice that his whiff rate sits at 13.7%, which would be top-5 among starters if he managed it for an entire season. Now, I’m not naive enough to think that number is where is true value lies after just 3 games, but he’s certainly improved off his 10.2% mark he had earlier in the season.

I’m not suggesting Shoemaker is the next coming of Clayton Kershaw. I’m not even sure if he’s the best pitcher on his own staff. But one thing is for sure: Matt Shoemaker is throwing the ball harder than he has in the past, and it’s working. And while it may not continue at this level, there’s no reason it should stop.