As anyone who does a lot of work with projections could likely tell you, one of the most annoying things about modeling future performance is that results themselves are a small sample size. Individual seasons, even full ones over 162 games, still feature results that are not very predictive, such as a hitter or a pitcher with a BABIP low or high enough to be practically unsustainable. For example, if Luis Arraez finishes the season hitting .350, we don’t actually know that a median projection of .350 was the correct projection going into the season. There’s no divine baseball exchequer to swoop in and let you know if he was “actually” a .350 hitter who did what he was supposed to, a .320 hitter who got lucky, or a .380 hitter who suffered misfortune. If you flip heads on a coin eight times out of 10 and have no reason to believe you have a special coin-flipping ability, you’ll eventually see the split approach 50/50 given a sufficiently large number of coin flips. Convergence in probability is a fairly large academic area that we thankfully do not need to go into here. But for most things in baseball, you never actually get enough coin flips to see this happen. The boundaries of a season are quite strict.

What does this have to do with projections? This volatile data becomes the source of future predictions, and one of the things done in projections is to find things that are not only as predictive as the ordinary stats, but also more predictive based on fewer plate appearances or batters faced. Imagine, for example, if body mass index was a wonderful predictor of isolated power. It would be a highly useful one, as changes to it over the course of a season are bound to be rather small. The underlying reasons for performance tend to be more stable than the results, which is why ERA is more volatile than strikeout rate, and why strikeout rate is more volatile than the plate discipline stats that result in strikeout rate. Read the rest of this entry »

As anyone who does a lot of work with projections could likely tell you, one of the most annoying things about modeling future performance is that results themselves are a small sample size. Individual seasons, even full ones over 162 games, still feature results that are not very predictive, such as a hitter or a pitcher with a BABIP low or high enough to be practically unsustainable. For example, if Luis Arraez finishes the season hitting .350, we don’t actually know that a median projection of .350 was, in fact, the correct projection going into the season. There’s no divine baseball exchecquer to swoop in and let you know if he was “actually” a .350 hitter who did what he was supposed to, a .320 hitter who got lucky, or even a .380 hitter who suffered misfortune. If you flip heads on a coin eight times out of ten and have no reason to believe you have a special coin-flipping ability, you’ll eventually see the split approach 50/50 given a sufficiently large number of coin flips. Convergence in probability is a fairly large academic area that we thankfully do not need to go into here. But for most things in baseball, you never actually get enough coin flips to see this happen. The boundaries of a season are quite strict.

What does this have to do with projections? This volatile data becomes the source of future predictions, and one of the things done in projections is to find things that are not only as predictive as the ordinary stats, but also more predictive based on fewer plate appearances or batters faced. Imagine, for example, if body mass index was a wonderful predictor of isolated power. It would be a highly useful one, as changes to that over the course of a season are bound to be rather small. Underlying reasons for performance tend to be more stable than the results, which is why ERA is more volatile than strikeout rate and why strikeout rate is more volatile than plate discipline stats that result in strikeout rate.

MLB’s own method comes with an x before the stat, whereas what ZiPS uses internally has a z. I’ll let you guess what it stands for! I’ve written more about this stuff in various places such as here and here, so let’s get right to the data for the first two months of the MLB season. Read the rest of this entry »

The Blue Jays started a short road trip off in style on Monday night with an 8–0 shellacking of the Royals. It was a feel-good kind of an evening — at least if you’re not a resident of Kansas City — in that the Jays got homers from two young players falling short of lofty expectations (Vladimir Guerrero Jr. and Bo Bichette) and that a recently struggling bullpen (4.70 FIP, -0.5 WAR over the last 30 days) one-hit the Royals over four innings in closing out the rout. It was one of the 10 games in which the Jays play teams at the bottom of the American League before heading to New York to square off against the first-place Yankees.

One of the questions I was asked on one of my radio hits yesterday, this one with Blake Murphy of Sportsnet 590 in Toronto, was if the next two weeks were especially important for the Jays, coming against the weakest teams and with a seven-game divisional deficit. I believed it was, but it did make me curious just how important it was. And since I can’t run a ZiPS projection in my own head but need a computer, here we are! Just how important are the next two weeks for Toronto?

It would be hard to characterize the Blue Jays as a struggling team. While they were hovering just above the .500 mark in mid-May, they’ve been on a bit of a tear lately, winning 14 of 19 games. A 32–22 record stands at a 96-win pace, comfortably above the 88-win median that ZiPS projected for each of the top four AL East teams back in April. The problem is that the Yankees have been even better at 39–15, or a 117-win pace.

Still, Toronto’s record is impressive, and even more so when you consider how tough a schedule it’s had. The team’s average opponent this year has had a .544 winning percentage, which translates into an 88–74 record. In other words, a typical game for the Jays has seen them face off against a team projected by ZiPS to be as strong as the average non-Orioles AL East team. From a projection standpoint, ZiPS believes that they have had the toughest schedule in baseball so far.

A seven-game deficit is a significant one. Even if you thought the Jays were as solid as the Yankees at the start of the season, winning one more game over 162 games is an easier task than winning eight more over 108. Even if New York plays just .500 ball the rest of the season, Toronto has to maintain its pace (62–46, .574) in order to finish with a one-game lead (avoiding a tiebreaker now that MLB has killed off game 163).

Opening up ZiPS, I ran some experiments on the team’s fate over the next two weeks. Let’s start with the current projections. Read the rest of this entry »

It seems like almost yesterday when, amidst the lockout’s flurry of recriminations and constantly shifting arbitrary deadlines, we weren’t quite sure if there was even going to be a 2022 season at all. But Opening Day arrived after a short delay, and now we’re just about a third of the way through the season. The trade deadline is just two months out and as we saw last year, the elimination of the August waiver-trade period served to increase the stakes. While we don’t know the exact contours of what the pennant races will look like or which destinations make the most sense for potential trade candidates, the basic outlines of the season have been drawn. Short of some major surprises, we can start speculating about a few of the more interesting players likely to be available.

Juan Soto, Washington Nationals

The official position of the Washington Nationals is that they aren’t trading Juan Soto, no way, no how. I’m not sure I actually believe them. Soto will be a free agent after the 2024 season and has already turned down a 13-year, $350 million extension offer from the team. Plus, the longer they hang onto him, the less mega of a mega-package they’re likely to net in return for their superstar. It’s tempting to compare Soto’s situation with Bryce Harper’s, but as he approached free agency, the Nationals were fielding a team they had reason to think was competitive. This year’s squad is looking up at the Reds, and the farm system doesn’t have anywhere near the talent needed to quickly salvage the situation. The possibility of a sale and Soto’s age complicate the calculus – if Washington was able to convince him to stick around, he’s young enough that he’s likely to still be very good the next time they are. Soto isn’t posting his normal numbers, but ZiPS sees little reason to worry; it thinks that Soto’s hit data should have resulted in a BABIP closer to .320 and a slugging percentage well in excess of .500, similar to his xBA and xSLG. It would stink for Nationals fans, and putting together a deal worthy of netting Soto is its own challenge, but a trade could be a possibility come August. Read the rest of this entry »

12:02

Dan Szymborski: Welcome to the start of the Dan Szymborski Birthday season.

12:03

Dan Szymborski: I’d personally avoid unless the next option is really poor – Coors is still Coors.

12:03

Dan Szymborski: Possible, but I’m still optimistic on some of those guys turning it around

12:06

Dan Szymborski: What’s kinda awkward is that Profar and Hosmer have been steadily fading but the guys struggling aren’t really bouncing back

Read the rest of this entry »

As you may recall, just as the season was getting underway, Aaron Judge, who is set to become a free agent after the 2022 season, rejected a contract extension proffered by the Yankees; the deal would have been worth $230.5 million over eight years (seven years at $30.5 million per year, plus $17 million for this year), keeping Judge in pinstripes for most of the rest of his career. Instead, Judge decided to play out his final season under team control and then hit the free agent market with as much leverage as he is ever likely to have. Judge gambled on himself, and while two-thirds of the season remains, the early returns are pointing in his direction.

In a year that has seen offense largely disappear — just as a number of power hitters have seen their performance evaporate — Judge has bucked the trend. After his home run in Sunday’s loss to the Rays, he’s already up to 18 on the season, nearly half of his total (39) from his impressive 2021 campaign. That number even outstrips the pace of his 2017 season, during which he hit 52 round-trippers in an offensive environment far more conducive to crushing pitchers’ dreams. Judge might not have the big contract he’s looking for yet, but he’s done about as much to improve his standing as anyone could in two months.

Judge’s season line stands at a spicy .303/.371/.657, numbers that would count as superlative even in Coors Field during the era’s highest-offense seasons. In 1968: The Next Generation, that’s enough for a 192 wRC+ and 2.8 WAR; spicy may actually undersell just how dangerous he’s been. Read the rest of this entry »

An old friend returned to the big leagues yesterday in a relatively unexpected place. After 11 seasons in the majors, all with the Cardinals, Matt Carpenter found himself searching for a new team this winter; in the end, he only managed to snag a minor league deal with the Rangers. Now, though, he’s found himself in New York with the Yankees, signing a major league deal with them that was announced on Thursday and hitting eighth in the starting lineup for their game against the Rays that same day. Does Carpenter have a second wind, or has too much time passed since he was an effective contributor?

That Carpenter found little interest in his services this winter was hardly surprising. Some players age gracefully, but he fell off a cliff after the 2018 season, dropping from a wRC+ of 140 to 96. If that had been the extent of his collapse, he’d still have a role in the majors; he still managed to collect 1.7 WAR in 492 plate appearances in 2019, thanks to not being awful at second or third base. But after hitting .176/.313/.291 combined over 2020 and ’21, even that saving grace didn’t provide quite enough grace. What rope remained after the COVID-shortened 2020 rapidly ran out of slack the following year, and his role was reduced to the extent that only 11 of his 53 games after the All-Star break were as a starter.

It strikes me as likely that Carpenter only survived on the roster because of his long history with the franchise; infielders who can’t hit are a dime a dozen, and he didn’t have an exploitable platoon split advantage the way a steeply declining Albert Pujols did. And while much has been made of Carpenter’s struggles against the shift, and while he’s been worse throughout his career relative to a traditional infield configuration, it’s not sufficient to explain the collapse. He hit just fine overall as the approach against him shifted (no pun intended) yearly toward all-shift after 2015; by the time 2018 rolled around, when he was still a dangerous offensive player, he was almost exclusively hitting against a stacked right side of the infield. Read the rest of this entry »

12:03

Dan Szymborski: Welcome to THUNDER CHAT

12:04

Dan Szymborski: OK, just regular chat, but adding Thunder to things always makes them more exciting

12:04

Dan Szymborski: I think ther’es some bad luck

12:04

Dan Szymborski: xStat stuff is having *calibration* problems, but it should be an issue of *bias* more than accuracy really

12:05

Dan Szymborski: ZiPS stats are more predictive (with slightly differing methodology) and is similarly overexpecting results

Read the rest of this entry »

Last week, I discussed some of the young, pre-free agency players who teams should be trying to sign to long-term contract extensions. I hadn’t been planning for there to be a part two, but you guys had so many additional players you wanted to talk about, and I can’t remember the last time I got more DMs about a piece than that one — well, about a piece for which everyone isn’t mad at me, at least!

So, let’s oil up and turn the crank on the ol’ ZiPS-o-Matic and get this projection mill hopping for seven more players.

Walker Buehler, Los Angeles Dodgers: Eight years, $204 million

Buehler is currently in the second and final year of an extension with the Dodgers that pays him $4 million a year. His next deal will be a tad more pricey. Clayton Kershaw is still around in Dodger blue, but his injury history and mild decline resulted in 2021 being the year that Buehler became The Man in the rotation, reducing Kershaw to the role of deuteragonist. And while Los Angeles still has a rocking rotation, the depth isn’t quite what it was in recent years, so there should be more than slight concern that the franchise’s most valuable pitcher is unsigned. With Buehler two years from free agency, the Dodgers aren’t likely to get any massive discounts, but this is the best time to sign him if you don’t want to pay him Gerrit Cole money later. The Dodgers don’t necessarily have to stop at this figure, either; what’s the fun of being wealthy if you don’t use that cash to pay for cool things?

There may be some concern in some places about the dropoff in Buehler’s strikeout rate, but while strikeout rate changes do tend to stick very quickly, they stick far more when the underlying stats support the drop-off than when they don’t. In this case, the contact rates and swinging-strike rates haven’t worsened at all, nor has his velocity fallen off a cliff, suggesting that it’s a blip rather than a plunge. You can make a similar argument for the Dodgers signing Julio Urías to an extension, likely for a significantly lesser haul, but given the workload Buehler has shown he can handle, he’d be my priority. Read the rest of this entry »

Last week, I talked about a few young players who teams should be itching to sign to long-term contracts. Due to public demand, another set of projections is on the way, but I’ll admit I intentionally omitted one pitcher, Pablo López, from the first look because I wanted more space to talk about him.

Unlike a lot of pitchers the Marlins have accumulated during their various fire sales, López wasn’t a highly touted arm in the minors. Prior to 2017 — the season during which he and three other players were traded by the Mariners to the Marlins for reliever David Phelps — he was basically a non-entity among prospect-watchers. He didn’t receive an official ZiPS projection that year, but if he had, it would have been similar to the projection he received before the 2018 season, which essentially saw him as a below-average innings-eater at his peak. At no time did he rank on a ZiPS Top 100 prospect list.

His first couple of campaigns with the Marlins featured decidedly mixed results. While López was essentially a league-average pitcher thanks to exit velocities that ranked towards the top of the league (the good kind of top of the league), he lacked the ability to finish off batters. From 2018-19, he basically threw four pitches: a relatively straight fastball, a sinker, a curve, and a changeup. None them were whiff-makers, and none of them had even a 20% put-away rate, resulting in a mediocre 7.5 K/9 combined over those two seasons. Read the rest of this entry »