Position: Washington Nationals Baseball Research & Development Analyst

Location: Washington DC

Description:
The Washington Nationals are seeking a data analyst to join the organization’s Baseball Research and Development team. This role will focus on using data science to support Baseball Operations in player evaluation, roster construction and in-game tactics. This includes using machine learning techniques and predictive modeling to draw insights from our in-house baseball datasets. The analyst will work closely with the existing R&D team, including other analysts and developers, to design and build decision-support systems and tools for use throughout the organization.

Responsibilities:

• Statistical modeling of baseball data.
• Build automated solutions to import, clean, and prepare baseball datasets for downstream analyses.
• Create informative visualizations of our baseball datasets.
• Assist in the maintenance of our data warehouse.
• Build information systems to support Baseball Operation’s efforts to improve player health and performance.
• Review of public research in both baseball and data science.

Read the rest of this entry »

Someone submitted something to my chat last Friday:

CapnZippers: Seth Lugo’s curveball averaged 3300 RPM last night. That spin rate is almost 10% higher than the next guy. Holy Moly! Mets maybe found a gem?

The same day, Mike Petriello sent out a relevant tweet:

Seth Lugo has a 2.51 ERA for the @Mets. He’s also wayyy ahead of everyone on the #statcast era curve spin list: pic.twitter.com/5CPs40o3EK

— Mike Petriello (@mike_petriello) August 26, 2016

I made a note to do some digging. To be honest, I don’t have a lot to add. As has been measured, Seth Lugo has thrown an outstanding curveball, in terms of its spin rate. It’s outstanding not because it’s amazing; it’s outstanding because it stands out. The average spin rate is way higher than anyone else’s. As a different way to demonstrate that, I pulled up all the individual games with the highest-spin curveballs, from Baseball Savant. Lugo, in the majors, has appeared in 10 games in which he’s thrown at least one curve. A leaderboard:

Lugo dominates the spin charts in the way that Aroldis Chapman dominates the velocity charts. This isn’t a one-off fluke — Lugo has an exceptional breaking ball. The question is whether Lugo himself will become a quality pitcher, but this should at least get you to raise your eyebrows.

It’s really too soon to say how much this means. It’s not an easy thing to analyze an individual pitch, and Lugo won’t go anywhere if his other pitches don’t play well, too. Everything works together. There is some evidence that high spin is correlated to reduced slugging. And there’s evidence that high spin is correlated to increased whiffing. The evidence isn’t strong, but it makes sense intuitively, and again, this is all complex. It’s fair to say that Lugo’s curveball is interesting, without going any further. I don’t know how interesting this should make Lugo as a player, but I know that now I’ll keep my eye on him. I didn’t have any reason to think about him before.

By velocity and movement, the best comparison for Lugo’s curve in the PITCHf/x era is Brett Myers‘ curve. Myers threw a phenomenal curve for an entire decade. Garrett Richards has another comparable curve, but he’s never thrown it much. Jake Arrieta’s curve also compares well, so that’s promising. Lugo’s curve appears to be a major-league pitch. A major-league out pitch, even. We’ll see about the other pitches.

Lugo picked up his first-ever big-league strikeout on a curve. It’s with a clip of that curve that I’ll leave you today.

How interesting an arm is Seth Lugo? I don’t know. “More” would be one answer.

A year ago, when the Pirates were alive in the playoff race, they made what felt like a fairly uninspiring deadline trade for J.A. Happ. Happ was almost a giveaway, and no one really batted an eyelash about the Pirates’ tiny upgrade, but then they made some very minor tweaks and Happ pitched the rest of the season like one of the better starters in the league. There was, at one point, an actual conversation about whether Happ should start the one-game playoff opposite Jake Arrieta. Things were weird.

This year, with the Pirates alive in the playoff race, they made what felt like a fairly uninspiring deadline trade for Ivan Nova. As I recall, news broke after the actual deadline had passed, and it was a small story because Nova didn’t have a lot of value. Nova, also, was almost a giveaway. The move drew criticism, with many saying the Pirates weren’t doing enough. Ivan Nova, after all, is no Chris Sale.

Guess what? Nova has started five games for Pittsburgh, and in those games the Pirates are 5-0. He’s run a sub-3 ERA, and while his strikeouts haven’t spiked, he’s sitting on one walk. One walk, out of 121 batters faced. Nova walked three of 23 batters in his final start with New York. All of a sudden, the Pirates have turned Ivan Nova into a strike machine, and it’s funny what happens when you have a pitcher who consistently gets ahead. The batters, you see, do worse.

As with Happ, the Pirates haven’t had to do anything drastic. Nova’s repertoire looks mostly the same. Nova’s delivery looks mostly the same. Ray Searage himself has said that Nova’s been easy to work with because there’s just not much to do. If the Pirates have done anything, it’s just encourage Nova to pitch with more confidence. Through July of this year, Nova ranked in the 15th percentile of all pitchers in rate of pitches thrown while ahead in the count. In August, Nova ranks in the 88th percentile. Where he was consistently behind, now he’s consistently ahead. This is all very fundamental.

We can look at Nova’s rolling zone rate over time:

The Pirates have Nova working in the zone more often. As for a rolling-average plot of first-pitch-strike rate:

First-pitch strikes more often. And the differences here aren’t huge. I’m going to show you pitch-location heat maps, comparing Nova through July to Nova in August. You can tell that the heat maps are, I don’t know, siblings? They’re just definitely not twins.

The Pirates have Nova working up a little more, and they’ve shifted him a bit, over the plate. Where Nova, previously, was a nibbler with his fastball, now he’s less focused on trying to stay on the edges. Very generally, Nova still has a familiar-looking pitch pattern, but there’s just more confidence there, so there’s more aggressiveness, too. He’s not a strikeout pitcher, and he’s not going to be a strikeout pitcher, but there’s nothing wrong with a low-walk pitcher who can work in the 90s. The Pirates can generate outs behind him.

The explanation might be obvious. Nova no longer is pitching in the American League, and he’s no longer working in AL East ballparks. PNC Park is very forgiving, and maybe Nova just needed to believe that not every fly ball is a threat. This doesn’t necessarily have to be Ray Searage magic. Maybe the Pirates simply identified the right guy to add. Nova hardly cost them anything. Now he’s working to cost some other team a potential playoff spot.

Here is one of my favorite clips of the season:

That’s Ronald Torreyes, attempting a delayed and perfunctory swing at a pitch-out to try to protect the running Aaron Hicks, who ends up in a heap on the ground after getting jarred in the marbles. Torreyes swings for no reason other than he’s always been told to swing in these situations, so the decision was entirely out of his hands. You can see that he’s temporarily overruled by his own brain, which properly identified that a swing would come with no upside. But then the training kicked in, and Torreyes whispered the bat in a vaguely forward direction while Hicks sprinted like the dickens, unaware the situation would end with teammates discussing his sterility.

Pretty obviously, no swing has been attempted this season at a more-outside pitch. Yet I don’t know if that should really “count,” since Torreyes didn’t swing because he wanted to. The swing was mandated by the hit-and-run play. So let’s take that off the table. Now the most-outside swing attempt of the season belongs to Jose Abreu, as of Thursday night. Abreu should probably be ashamed of himself.

Though I looked at everyone, the swings at the very most-outside pitches have been attempted by righties. Allow me to read off to you the top three:

1. Ronald Torreyes, June 30, swinging pitch-out
2. Jose Abreu, August 25, swinging strike
3. Jose Iglesias, May 24, swinging pitch-out

The only worse swing was at a pitch-out. The next-worst swing was at a pitch-out. The next-worst swing at a non-pitch-out was at a pitch more than five inches closer to the plate. That swing was also with two strikes, attempted by Javier Baez. Baez will do that sometimes. So, evidently, will Abreu.

Exclaimed Mariners announcer Dave Sims, after Abreu’s strikeout with runners in scoring position:

Swing and a miss, he got him! What a big pitch.

It’s easy to get fooled on the fly. Strikeouts are strikeouts, and when the batter swings, that implies a pitch could have been only so bad. Abreu chased this slider from Steve Cishek; therefore, it must have been a good slider from Steve Cishek. Yet it’s not hard to see how that could have been a disastrous slider from Steve Cishek. You don’t want a pitch in that situation to get away. And Abreu had never before swung like this. I went to Baseball Savant. I plotted all of Abreu’s career swings. The swing above is highlighted below.

I mean-

Eleven inches. The difference between that pitch and the next-most-outside pitch Abreu had chased is 11 inches. Nearly a whole damn foot. There’s really no excuse for that kind of swing. The easy explanation is “Abreu was trying to do too much,” but trying to do anything with that pitch is trying to do too much. It’s a brain fart. It has to be a brain fart. I don’t know what else it would be unless, as of Thursday night, in the seventh inning, Jose Abreu suddenly became, on camera, the single worst hitter in Major League Baseball.

By the way, the Baez swing? The one that’s the next-worst of the season?

That swing was also against a Steve Cishek slider. It’s probably just a coincidence. But, maybe I’m the one who doesn’t get it.

Jorge Alfaro’s physical tools have put him in the prospect conversation for years — he’s cracked Baseball Prospectus’ top-101 prospects in each of the past five seasons, for example — but his on-field performance has always left something to be desired. He hit a respectable-for-a-catcher .253/.314/.432 in an injury-shortened season at Double-A level last year, but his plate discipline was poor. Although he demonstrated enticing power, his 4% walk rate and 29% strikeout rate hinted at serious issues with his approach. 

He’s seemingly begun to make the right adjustments this year, as he’s hacked six points off of his strikeout rate without sacrificing much power. In just under 400 plate appearances in Double-A, he slashed a more-encouraging .279/.322/.444. Whatever development has occurred, it seems to have satisfied the Phillies, who will promote the catcher today according to Yahoo’s Jeff Passan.

Alfaro’s future looks brighter than it did five months ago, but he’s still far from a slam-dunk prospect. Though his strikeout and walk numbers are trending in the right directions, they’re still cause for concern. And though he’s only 23, Alfaro has been playing professionally since 2010, so he may not have a ton of improving left to do.

While he’s improved at the plate, Alfaro’s biggest strides seem to have taken place behind it. According to Baseball Prospectus’ pitch-framing data, Alfaro’s framing was nearly a run worse than average last year, but has been over 14 runs better than average this season. Clay Davenport’s data tell a similar tale: +1 last season and +12 this year.

Read the rest of this entry »

Last night, Mike Trout went 3-for-6 with a double, a homer, two runs, and an RBI. The Angels won, 8-2, over the Blue Jays, but let’s get back to that home run. Here’s what Mike Trout’s body looked like just moments before making contact with a pitch that went over the left field fence:

We all know what a typical home run swing looks like, and it sure doesn’t look like that. We also know what a normal baseball player looks like, and it sure doesn’t look like Mike Trout.

Read the rest of this entry »

A 13th-round pick last year, infielder Max Schrock — received by Oakland today from Washington in exchange for left-handed reliever Marc Rzepczynski — has made something of a name for himself by putting up strong offensive numbers in the lower levels. He’s hitting .333 between two levels of A-ball this season, largely due to an 8% strikeout rate. That’s encouraging coming from a middle infielder with speed and decent power. As a result, he’s become a regular on Carson Cistulli’s Fringe Five column.

Despite his strong performance, KATOH isn’t a huge fan of Schrock. My system pegs him for 2.9 WAR over his first six seasons by the traditional method and 2.8 WAR by the method that integrates Baseball America’s rankings. That puts him in the #150-#200 range in terms of prospects. To help you visualize what his KATOH projection entails, here is a probability density function showing KATOH+’s projected distribution of outcomes for Schrock’s first six seasons in the major leagues.

While Schrock’s hitting has been very good, KATOH dings him for being just 5-foot-8, and also for playing second base rather than shortstop. Second baseman with good numbers in the low minors don’t pan out all that often. There are some obvious exceptions to that statement, but it’s worth pointing out that those exceptions all provide defensive value, while Schrock has been eight runs below average at second base, according to Clay Davenport’s numbers.

To put some faces to Schrock’s statistical profile, let’s generate some statistical comps for the undersized second baseman. I calculated a weighted Mahalanobis distance between Schrock’s performance this year and every A-ball season since 1991 in which a batter recorded at least 400 plate appearances. In the table below, you’ll find the 10 most similar seasons, ranked from most to least similar. The WAR totals refer to each player’s first six seasons in the major leagues. A lower “Mah Dist” reading indicates a closer comp.

Please note that the Mahalanobis analysis is separate from KATOH. KATOH relies on macro-level trends, rather than comps. The fates of a few statistically similar players shouldn’t be used to draw sweeping conclusions about a prospect’s future. For this reason, I recommend using a player’s KATOH forecast to assess his future potential. The comps give us some interesting names that sometimes feel spot-on, but they’re mostly just there for fun.

As Dave Cameron pointed out, Rzepczynski is a mediocre left-handed reliever, and a month of his services probably could have been had for next-to-nothing. Schrock probably won’t win any MVP awards, but there’s a pretty decent chance he’ll be a useful role player in a couple of years. That’s demonstrably more than next-to-nothing.

One of the recurring themes over the last year has been the high price the market is putting on relief pitchers. Ken Giles and Craig Kimbrel brought back monster returns in trades last winter, while even decent middle relievers were getting two or three year deals as free agents. At last month’s trade deadline, Aroldis Chapman got the Yankees a terrific haul. It’s clearly a good time to be selling relief pitching.

And today, it looks like those rising prices have trickled down to mediocre lefty specialists, as the Nationals gave up a legitimately interesting prospect for a five week rental of Mark Rzepczynski. The guy nicknamed Scrabble is a useful situational reliever, having held lefties to a .270 wOBA in his career — don’t put any stock on his 2016 reverse split, which is all BABIP driven — but he’s useless against righties (.351 career wOBA allowed) so he’s effectively a one out guy.

Of course, the Nationals already have exactly this kind of pitcher in Oliver Perez, who they signed for $7 million over two years last winter. Perez hasn’t been as good as they hoped this year, and has been particularly bad lately, getting torched for a .473 wOBA in August. But he’s still held lefties to a .316 wOBA this year after keeping them to just a .230 wOBA last year, and is at .306 for his career; you’d have to really overreact to five bad innings in August to think that the Nationals needed to give up real value to improve their lefty specialist for the playoffs.

For the right to marginally upgrade their LOOGY, a guy who might face four or five batters in an entire playoff series, the Nationals surrendered a 21 year old currently running a 131 wRC+ in high-A ball. Max Schrock isn’t an elite prospect or anything, but he’s been a fixture on Carson’s Fringe Five all year, based on his strong contact rates and at least a little bit of power. As a diminutive second baseman, it’s easy to look at Schrock as a limited upside guy, but with Jose Altuve on the verge of getting AL MVP votes, we should remember that the idea of firm upside ceilings are much less concrete than is often thought.

Schrock’s more likely future is as a bench bat, but even if he’s just Alberto Callaspo 2.0, giving that up for a marginal gain in lefty specialists seems weird.

So why did the Nationals do this? Well, the deal was announced as Rzepczynski and cash for Schrock, so presumably, Oakland is paying some of the remaining ~$800K or so left on his contract this season. But according to sources in the game, Rzepczynski actually cleared waivers before this trade was completed, so the Nationals could have simply had him for just a the waiver fee, and kept a legitimately interesting prospect in their system.

So, effectively, the Nationals just sold a decent prospect for some cash savings in order to bolster the least important part of their bullpen. I know the value of relievers is going up, but deals like this still seem silly to me. Maybe Scrabble will get a big out or two in October and it will seem worth the long-term cost, but Rzepczynski seems like the kind of guy the Nationals should have gotten for next to nothing. When the market is inflating reliever prices to the point you have to give up a legitimately interesting 21 year old for a five week rental of a LOOGY, maybe it’s just time to stop paying market prices for relievers and go with what you already have.

Andrew Miller made a batter look silly last night. Andrew Miller makes batters look silly most nights, but last night, a batter looked particularly silly. One surefire way to identify a batter who just got done looking silly is to check whether he’s laying down in dirt right after he swung. Let’s see.

Q: Was This Major Leaguer Laying Down in the Dirt Right After He Swung?

A: Sure was

That there batter sure looked silly. The good news for Khris Davis is, he’s not alone. He didn’t even take the season’s most ill-advised swing against Miller. As our own Jeff Sullivan pointed out back in May, maybe the thing Miller does very best is force hitters to take ill-advised swings, at least certainly relative to the ones they don’t take. I’ll explain. When Jeff wrote his article in May, Miller’s O-Swing% against was higher than his Z-Swing% against, making him the only pitcher in baseball with such a distinction. To translate that into English: batters were swinging at would-be balls from Miller more frequently than they were swinging at would-be strikes. That’s not how hitting is supposed to work.

Read the rest of this entry »

The Reds are bad, and the Dodgers just clobbered them. As a part of that clobbering, Adrian Gonzalez socked three dingers. Here, you can watch them all! Each is impressive in its own way, I suppose. They’re most impressive as a group.

There was a time when Gonzalez’s power was absent. He hit three home runs in April. He hit three home runs, combined, in May and June. The fairly obvious culprit was a back injury, and it takes no imagination at all to figure out how an achy back could affect a swing. Gonzalez got some treatment. He insisted his back felt better. The numbers still didn’t quite show up.

Now they’re showing up. From the looks of things, Gonzalez is indeed healthy again, and it simply took him a short stretch to re-find what previously had made him successful. Adrian Gonzalez is looking like Adrian Gonzalez again. The most compelling evidence is probably what follows, as drawn from Baseball Savant. Here are nearly two years of Gonzalez’s batted balls, as tracked by Statcast. This plot shows rolling average launch angles.

That dip is impossible to miss, as an injured and compromised Gonzalez plummeted toward 0. He’s reversed that, again successfully putting batted balls in the air. And he’s once more hitting with authority. Here’s a rolling-average plot of hard-hit rate, and this doesn’t include Monday’s game, not that it would make a huge difference:

Gonzalez is getting his swing back, and he’s also showing a better ability to hit the ball hard to right. Of course, at his best, he’s been more of an all-fields threat — and Monday, he pulled two homers while sending another the other way. I know it was a day game in Cincinnati against the Reds, but big-league homers are big-league homers. Gonzalez can hit them again, and he can do other things, and suddenly he’s up to a 123 wRC+. That’s in line with where he’s been for a while.

The Dodgers are still playing without Clayton Kershaw. The Dodgers are still winning without Clayton Kershaw. The Dodgers, of course, would love Kershaw back, but since the All-Star break, Gonzalez has been a key part of what’s been the best offense in the National League. Kershaw’s important. He’s pretty far from being everything.