Suffice it to say, the 2023 season has not gone the way either the New York Yankees or their fans had hoped. The team’s current nine-game losing streak is their longest in 41 years. And while the team’s 60-65 record isn’t on the same grim level as those of the Athletics or Royals, it’s still awful by the franchise’s typical standards. New York has teetered on the edge of .500 a few times recently, including being outscored in three of four seasons from 2013 to 2016, but you have to go back to 1992 to find the last time they crossed that negative line. Rather than tear everything down to the foundation when things go wrong, the Yankees tend to be a team that reloads and tries again next time. But can they do that this offseason?

The Yankees have had some bad breaks this season, but blaming everything on that would be a mistake. I’m not going to wax poetic about why this season has been so miserable — other writers have already laid out the club’s tale of woe — but we still need to review the basics to get a good view of where things truly stand. The pitching bears quite a lot of the blame. In detailing how the preseason PECOTA projections for the Yankees diverged from what has actually happened, Patrick Dubuque didn’t mince words at Baseball Prospectus:

Two of the Yankees’ seven starters have met expectations so far, and it’s their two worst ones. Injuries have pressed those sixth and seventh (and eighth) starters into service, even more so than our depth chart team anticipated. But when you imagine a collapse like the Yankees have had, you assume that it’s injuries. You envision Aaron Judge’s plate appearances replaced by Billy McKinney’s, like the world’s most unprepared Broadway understudy. While Brito and Randy Vásquez haven’t bailed the team out, they also didn’t make the hole. And at this point, it’s more hole than boat.

Read the rest of this entry »

Padres fans in 2023 don’t have a ton to be excited about. The Friars have been in win-now mode for the last four seasons and are staring down their second losing campaign during that span. One of the two winning seasons was kind of ruined — for everyone in the world — by a raging pandemic, leaving fans with only one year that was both normal and an enjoyable experience since 2010. Unlike a lot of teams with a similar performance record, it’s not for lack of investment in the team. Just a few months after the gigantic trade that brought Juan Soto to town, the team signed Xander Bogaerts to a $280 million contract and kept Manny Machado from opting out with an even spicier $350 million pact. The Padres also agreed on a $100 million contract for Joe Musgrove and locked up Yu Darvish for $108 million. That’s more than $800 million, so we’re not talking about the case of, say, the White Sox having issues in part because they couldn’t be bothered to fill giant holes in the lineup because that would have required money.

As gloomy as the season feels right now, there are still legitimate reasons to think the Padres are a good baseball team. Their 68–54 Pythagorean record is 10 wins above their actual record, and records derived from run differential are more predictive than win-loss record. The projections all still agree there’s a lot to like and similarly have a good record, relatively speaking, of predicting the future. And this holds true even when talking about teams with the largest disagreement between the projections and the record. Looking at the 25 teams that FanGraphs like better than their seasonal winning percentage the most, coin flips missed their rest-of-season winning percentages by an average of 86 points, season to-date records by 81 points, and FanGraphs records by 65 points. Those 25 teams had played .396 ball through August 14 of their seasons; FanGraphs projected a .476 RoS winning percentage, and the actual RoS winning percentage for those teams was .458. We weren’t imagining things.

But the fundamental problem the Padres face is that it’s simply far too late to be the team they hoped they were. Our projections still believe they are a .572 team, but that’s only good enough for a 19% chance of making the postseason with a divisional probability that rounds to zero; the ZiPS projections have it at 15%. While those are still pretty good odds, especially compared to how the season has felt, it’s still far more likely than not that this year ends up being a dark companion to the 2021 season that also ended in stunningly bleak fashion.

And here’s the problem: the Padres project to be worse in the future than they are now. You could say that about most teams, but the Padres are also a team that has a massive amount of payroll already tied up in a declining roster, an unsigned Soto approaching free agency, and probably not a lot of room left to grow in a payroll sense. Complicating things even further is the financial collapse of Bally Sports, as the team has not yet figured out how to replace that revenue. Forbes estimated the Padres lost $53 million in 2022, and things are likely to get worse from there. Peter (Seidler) actually saw a wolf.

Running some up-to-date projections for players signed long-term demonstrates the enormity of the team’s challenge. I’m going to start with the Padres’ core of six players who have guaranteed contracts with annual salaries of at $10 million or more through at least the 2026 season. Whatever happens elsewhere on the team, these six are almost certainly going to be part of the foundation.

You may cringe looking at the end of Machado’s contract, but ZiPS already expected that before the season. Machado put together a strong enough July — though he’s slumped since then and is nursing a sore hamstring — and experienced a clear return to defensive form to cause his 2024-and-on projections to tick up slightly. While ZiPS didn’t like the deal, it doesn’t like it any less than it did in February.

Bogaerts was mired in a deep slump in May and June, aided by a sore wrist, but has hit a more Xanderian .290/.351/.413 since the start of July, in-line with preseason expectations. As with Machado’s recent deal, the Padres go into this contract knowing that they’re paying for Bogaerts to decline.

This is a bit of an awkward projection because it highlights an assumption in team construction that turned out not to be true. Typically, a decent defensive second baseman who can credibly fake playing shortstop will usually fare well at first base, but that just has not happened, at least so far, with Jake Cronenworth. With Ha-Seong Kim 김하성 firmly entrenched as a starter, the Padres have a player with value but not a logical place to play him in order to get that value. The difference is extreme enough that ZiPS thinks that Cronenworth is now more than a win per season less valuable at first than second base.

Add in the fact that Cronenworth is having a down year (and a pretty odd one in terms of Statcast data), and there’s just a lot less reason to like his future than there was before.

Where the positional gods punished the Padres with Cronenworth, they were far kinder here. Tatis’ bat isn’t quite where it was, but he’s actually turned out to be an excellent defensive outfielder, at least so far. Given his age, he’s the one player who projects to finish out his contract as a plus contributor to the team.

ZiPS was always fairly pessimistic about the Darvish extension, and without him reversing the continued slow decline in his peripherals, it hasn’t changed direction since the start of the season. While I always say “hitters age, pitchers break,” Darvish is at an age where cliffs do, in fact, beckon. ZiPS didn’t even want to project the final two seasons of his extension; I’ll be kind and not force it to, so we’ll call those zero-WAR seasons, which is sunnier than what ZiPS would say if I made it.

The performance projections of Musgrove have stayed about the same — hardly surprising considering that when he was healthy, he was having a similar season to last year. But his return this season remains up in the air, and new injuries create new risk for a pitcher, so his projected innings totals have dropped considerably.

OK, let’s throw everybody into one table, complete with their salaries.

If the projections hold true, these six will make up less than a third of the WAR needed to be a 90-win team as soon as 2026, when they combine for $120 million in salary. Unless the team continues to spend more and more money, it’s going to get harder and harder to use dollars to patch holes, which means that the farm system has to get back to producing very quickly. The Padres aren’t likely to be able to win on the backs of these six players for very long, which means that they likely have to come up with a whole new core of talent around these players.

The risk here is one of dynastic failure. I’m not calling the Padres a dynasty in terms of baseball success, but more how the term has been used historically. Lots of warlords in history managed to get a throne, but to establish long-term rule, they had to survive the transition to the next rulers. The Astros are an example of a team that has avoided dynastic failure; only a handful of the players on the team that won the World Series in 2017 were still on the roster when Houston won the World Series in 2022. Only Jose Altuve and Alex Bregman remain among the Astros’ offensive contributors today, and the only pitcher still in Houston, Lance McCullers Jr., won’t pitch again until 2024 [and Verlander who I forgot about for some very odd reason -DS]. They basically came up with nearly an entirely new team in five years.

Baseball history is riddled with successful teams that were unable to transition to the next era without a significant interregnum, such as the Utley-Howard Phillies and the Tigers during the peak Miguel Cabrera years. But those Phillies won a World Series, and while the Tigers didn’t, they had more playoff success than these Padres teams have had. To achieve that success, the Padres are going to have to be extremely creative over the next five years, lest they end up as one of the great “what ifs” in baseball history. Spending money and having a few big stars won’t be enough.

Two months ago, if you asked me to name the most disappointing member of the phenom class, I’d have said Bobby Witt Jr. With barely a month of professional games past high school under his belt, he was invited to spring training in 2021 and hit three homers and put up an .851 OPS, creating chatter around baseball that he might start the season with the parent club. That was a bit premature, though he did spend the next six months terrorizing minor league pitchers into thinking long and hard about their choice of occupation. But in 2022 and early on in ’23, brevity was no longer the soul of Witt, as his whirlwind professional progress slowed to become one of those inevitably anemic breezes on an unpleasantly muggy July day.

Things appeared to reach their nadir in late June, when his OPS almost dipped under .700 once again. Since then, however, Witt has been on a tear, hitting .350/.385/.662, not only bringing his OPS safely over the .700 line but also getting it over .800. Since the morning of June 30, he’s been one of the absolute best players in baseball, providing a rare highlight for the 2023 Royals:

I’ve included BABIP here for a very good reason: when players are having hot streaks, BABIP is usually a big reason why. After all, players playing at their peak are more likely to be playing above their abilities than below. Witt is no exception here, with his numbers fueled in part by a .370 BABIP over that period. But I include that figure not to defuse my thesis, but to reinforce it. While a BABIP that high is hard to sustain over the long haul, ZiPS’ zBABIP thinks that .370 mark only barely outperforms what he’s actually done in the last month and a half. Read the rest of this entry »

The Astros currently rank third in the American League in runs scored — not an uncommon sight for a franchise that has only been outscored by the Dodgers and Red Sox over the last decade. But they’ve done it with a lineup with some pretty big holes, with half of their eight players with at least 300 plate appearances this season posting an OBP under .300. The team’s offense has been driven this year mainly by four players: Kyle Tucker, Yordan Alvarez, Alex Bregman, and Chas McCormick, with an assist from Yainer Diaz. A year ago, Houston signed Alvarez to a six-year, $115 million contract extension that ensured he would remain in town until the end of the 2028 season. Tucker, though, does not have a long-term deal and is scheduled to hit free agency after the 2025 season. What would a possible deal look like?

There’s certainly interest from the Astros’ side, as there ought to be. The team has discussed an extension with Tucker in the past, though there are no active talks right now. But general manager Dana Brown did use his weekly radio spot in part to discuss making Tucker an “Astro for life,” so some kind of deal coming together is hardly implausible. Read the rest of this entry »

One of the strange things about projecting baseball players is that even results themselves are small samples. Full seasons result in specific numbers that have minimal predictive value, such as BABIP for pitchers. The predictive value isn’t literally zero; individual seasons form much of the basis of projections, whether math-y ones like ZiPS or simply our personal opinions on how good a player is. But we have to develop tools that improve our ability to explain some of these stats. It’s not enough to know that the number of home runs allowed by a pitcher is volatile; we need to know how and why pitchers allow homers beyond a general sense of pitching poorly or being Jordan Lyles.

Data like that which StatCast provides gives us the ability to get at what’s more elemental, such as exit velocities and launch angles and the like — things that are in themselves more predictive than their end products (the number of homers). StatCast has its own implementation of this kind of exercise in its various “x” stats. ZiPS uses slightly different models with a similar purpose, which I’ve dubbed zStats. (I’m going to make you guess what the z stands for!) The differences in the models can be significant. For example, when talking about grounders, balls hit directly toward the second base bag became singles 48.7% of the time from 2012 to ’19, with 51.0% outs and 0.2% doubles. But grounders hit 16 degrees to the “left” of the bag only became hits 10.6% of the time over the same stretch, and toward the second base side, it was 9.8%. ZiPS uses data like sprint speed when calculating hitter BABIP, because how fast a player is has an effect on BABIP and extra-base hits.

And why is this important and not just number-spinning? Knowing that changes in walk rates, home run rates, and strikeout rates stabilized far quicker than other stats was an important step forward in player valuation. That’s something that’s useful whether you work for a front office, are a hardcore fan, want to make some fantasy league moves, or even just a regular fan who is rooting for your faves. If we improve our knowledge of the basic molecular structure of a walk or a strikeout, then we can find players who are improving or struggling even more quickly, and provide better answers on why a walk rate or a strikeout rate has changed. This is useful data for me in particular because I obviously do a lot of work with projections, but I’m hoping this type of information is interesting to readers beyond that.

Yesterday, I went over how pitchers zStats for the first two months of the season performed over the last two months. Today, we’ll look at the updated data, through the games on August 10. Read the rest of this entry »

One of the strange things about projecting baseball players is that even results themselves are small samples. Full seasons result in specific numbers that have minimal predictive value, such as BABIP for pitchers. The predictive value isn’t literally zero — individual seasons form much of the basis of projections, whether math-y ones like ZiPS or simply our personal opinions on how good a player is — but we have to develop tools that improve our ability to explain some of these stats. It’s not enough to know that the number of home runs allowed by a pitcher is volatile; we need to know how and why pitchers allow homers beyond a general sense of pitching poorly or being Jordan Lyles.

Data like that which StatCast provides gives us the ability to get at what’s more elemental, such as exit velocities and launch angles and the like — things that are in themselves more predictive than their end products (the number of homers). StatCast has its own implementation of this kind of exercise in its various “x” stats. ZiPS uses slightly different models with a similar purpose, which I’ve dubbed zStats. (I’m going to make you guess what the z stands for!) The differences in the models can be significant. For example, when talking about grounders, balls hit directly toward the second base bag became singles 48.7% of the time from 2012 to ’19, with 51.0% outs and 0.2% doubles. But grounders hit 16 degrees to the “left” of the bag only became hits 10.6% of the time over the same stretch, and toward the second base side, it was 9.8%. ZiPS uses data like sprint speed when calculating hitter BABIP, because how fast a player is has an effect on BABIP and extra-base hits.

ZiPS doesn’t discard actual stats; the models all improve from knowing the actual numbers in addition to the zStats. You can read more on how zStats relate to actual stats here. For those curious about the r-squared values between zStats and real stats for the offensive components, it’s 0.59 for zBABIP, 0.86 for strikeouts, 0.83 for walks, and 0.78 for homers. Those relationships are what make these stats useful for predicting the future. If you can explain 78% of the variance in home run rate between hitters with no information about how many homers they actually hit, you’ve answered a lot of the riddle. All of these numbers correlate better than the actual numbers with future numbers, though a model that uses both zStats and actual ones, as the full model of ZiPS does, is superior to either by themselves. Read the rest of this entry »

12:00

Dan Szymborski: It’s a chat!

12:00

Dan Szymborski: Probably not in serious jeopardy. ZiPS projections for Singleton aren’t great, either

12:01

Dan Szymborski: some unsalted peanuts I happen to have here

Read the rest of this entry »

One of the strange things about projecting baseball players is that even results themselves are small samples. Full seasons result in specific numbers that have minimal predictive value, such as BABIP for pitchers. The predictive value isn’t literally zero — individual seasons form much of the basis of projections, whether math-y ones like ZiPS or simply our personal opinions on how good a player is — but we have to develop tools that improve our ability to explain some of these stats. It’s not enough to know that the number of home runs allowed by a pitcher is volatile; we need to know how and why pitchers allow homers beyond a general sense of pitching poorly or being Jordan Lyles.

Data like that which StatCast provides gives us the ability to get at what’s more elemental, such as exit velocities and launch angles and the like — things that are in themselves more predictive than their end products (the number of homers). StatCast has its own implementation of this kind of exercise in its various “x” stats. ZiPS uses slightly different models with a similar purpose, which I’ve dubbed zStats. (I’m going to make you guess what the z stands for!) The differences in the models can be significant. For example, when talking about grounders, balls hit directly toward the second base bag became singles 48.7% of the time from 2012 to ’19, with 51.0% outs and 0.2% doubles. But grounders hit 16 degrees to the “left” of the bag only became hits 10.6% of the time over the same stretch, and toward the second base side, it was 9.8%. ZiPS uses data like sprint speed when calculating hitter BABIP, because how fast a player is has an effect on BABIP and extra-base hits.

ZiPS doesn’t discard actual stats; the models all improve from knowing the actual numbers in addition to the zStats. You can read more on how zStats relate to actual stats here. For those curious about the r-squared values between zStats and real stats for the offensive components, it’s 0.59 for zBABIP, 0.86 for strikeouts, 0.83 for walks, and 0.78 for homers. Those relationships are what make these stats useful for predicting the future. If you can explain 78% of the variance in home run rate between hitters with no information about how many homers they actually hit, you’ve answered a lot of the riddle. All of these numbers correlate better than the actual numbers with future numbers, though a model that uses both zStats and actual ones, as the full model of ZiPS does, is superior to either by themselves. Read the rest of this entry »

One of the strange things about projecting baseball players is that even results themselves are small samples. Full seasons result in specific numbers that have minimal predictive value, such as BABIP for pitchers. The predictive value isn’t literally zero — individual seasons form much of the basis of projections, whether math-y ones like ZiPS or simply our personal opinions on how good a player is — but we have to develop tools that improve our ability to explain some of these stats. It’s not enough to know that the number of home runs allowed by a pitcher is volatile; we need to know how and why pitchers allow homers beyond a general sense of pitching poorly or being Jordan Lyles.

Data like that which StatCast provides gives us the ability to get at what’s more elemental, such as exit velocities and launch angles and the like — things that are in themselves more predictive than their end products (the number of homers). StatCast has its own implementation of this kind of exercise in its various “x” stats. ZiPS uses slightly different models with a similar purpose, which I’ve dubbed zStats. (I’m going to make you guess what the z stands for!) The differences in the models can be significant. For example, when talking about grounders, balls hit directly toward the second base bag became singles 48.7% of the time from 2012 to ’19, with 51.0% outs and 0.2% doubles. But grounders hit 16 degrees to the “left” of the bag only became hits 10.6% of the time over the same stretch, and toward the second base side, it was 9.8%. ZiPS uses data like sprint speed when calculating hitter BABIP, because how fast a player is has an effect on BABIP and extra-base hits.

ZiPS doesn’t discard actual stats; the models all improve from knowing the actual numbers in addition to the zStats. You can read more on how zStats relate to actual stats here. For those curious about the r-squared values between zStats and real stats for the offensive components, it’s 0.59 for zBABIP, 0.86 for strikeouts, 0.83 for walks, and 0.78 for homers. Those relationships are what make these stats useful for predicting the future. If you can explain 78% of the variance in home run rate between hitters with no information about how many homers they actually hit, you’ve answered a lot of the riddle. All of these numbers correlate better than the actual numbers with future numbers, though a model that uses both zStats and actual ones, as the full model of ZiPS does, is superior to either by themselves.

And why is this important and not just number-spinning? Knowing that changes in walk rates, home run rates, and strikeout rates stabilized far quicker than other stats was an important step forward in player valuation. That’s something that’s useful whether you work for a front office, are a hardcore fan, want to make some fantasy league moves, or even just a regular fan who is rooting for your faves. If we improve our knowledge of the basic molecular structure of a walk or a strikeout, then we can find players who are improving or struggling even more quickly, and provide better answers on why a walk rate or a strikeout rate has changed. This is useful data for me in particular because I obviously do a lot of work with projections, but I’m hoping this type of information is interesting to readers beyond that.

As with any model, the proof of the pudding is in the eating, and there are always some people that question the value of data such as these. So for this run, I’m pitting zStats against the last two months and all new data that obviously could not have been used in the model without a time machine to see how the zStats did compared to reality. I’m not going to do a whole post for this every time, but this is something that, based on the feedback from the last post in June, people really wanted to see the results for.

Starting with zBABIP, let’s look at how the numbers have shaken out for the leaders and trailers from back in June. I didn’t include players with fewer than 100 plate appearances over the last two months. Read the rest of this entry »

Well, that’s the end of that. The trade deadline has come and gone, and whatever teams have is, well, whatever they’re going to have for the stretch run. My colleague Ben Clemens has already done the traditional look at the winners and losers of the deadline, so now it’s the turn of ZiPS, as I do every year afterwards. This is a very targeted look, in that ZiPS isn’t really looking at whether teams did a good job on a general level, only how the deadline affects their 2023 chances. So a team like the Mets ranking at the bottom isn’t a reflection on their competence, but how the deadline changed their postseason probability.

The methodology is simple. I project the rest of the season (after the games on August 1) both with the current rosters and if I undid every single trade made in the final two weeks of the trade period. I’ve included the projections for playoffs, division, and World Series for each of the 30 teams, with the default sort being playoff probability. Read the rest of this entry »