Archive for Research

You want your ace on the mound when all the marbles are in the fire. When the stakes are high, the big cheese needs to have his hands on the wheel. Are those real metaphors? No, but I bet you knew what I was talking about in each case, because the playoffs are a perfect time to discuss the importance of star pitchers. But do you want to deploy them as early in the series as possible? Or do you want them for the decisive game? Does it even matter?

In our game chats and article comments — across the internet really — it’s hard to find a consensus. When I wrote about Chris Sale earlier this week, there were comments to the effect of “It’s a bummer he pitched out of the ‘pen in the Wild Card, now he won’t get to pitch in NLDS Game 5,” as well as “Don’t save Sale for Game 7 of the NLCS, there might not be a Game 7!” Those are exact opposite opinions, and yet they both make a lot of sense when I hear them. But they can’t both be right, so let’s do some math. Even though there’s no real adage to prove or disprove here, this feels like a job for Mathemadage.

Here’s the question, as I see it: For a given set of pitchers making postseason starts, which way would you like those starts arranged in a seven-game series? In other words, if you have an ace, a second banana, and some bullpen games, would you prefer to have your ace for Games 1 and 5 or Games 3 and 7? There are good intuitive reasons to prefer either. Games 1 and 5 happen in almost every seven-game series; Game 7 is a much less frequent occurrence. More innings for my best guy, please! But the stakes can also never be higher than they are in a decisive game. In a big spot? Why wouldn’t I want my best guy in? Read the rest of this entry »

Approaching the end of a 2026 season that didn’t match his stratospheric standards, Paul Skenes attributed some of his problems this year with velocity to his participation in the World Baseball Classic.

Usually, that time of year, you’re kind of ramping up and it’s a little bit slow. You can kind of get onto the slope and find your body. One live [batting practice], you’re topping at 97 [mph], and then the next you’re 98, 99. And that wasn’t the case this year. Looking back, I don’t think that really helped.

This notion from Skenes did not come out of nowhere. A wide assortment of people has addressed this question on some level. Our very own Michael Baumann touched on it in 2023, though he was looking more at teams as a whole than players.

I’m a natural skeptic when it comes to claims of causation, which tend to be fueled by anecdote. From the supposed sophomore slump to the suspected swing-ruining impact of the Home Run Derby, there’s a lot of loosey-goosey if-then statements going around. But as these things go, the theory that WBC participation in a given year negatively affects player performance in the corresponding campaign has a lot of plausibility. The WBC is a very different environment than typical spring training games, and it comes at a time when players are normally focused on getting ready for the regular season. It doesn’t sound crazy to think these changes in preparation have some kind of deleterious effect on a player. However, we should test these things. Read the rest of this entry »

Back in August, the baseball savants at Baseball Savant bestowed upon us nerds yet another helping of Statcast data, this time in the form of a first base receiving leaderboard. Mike Petriello and Tom Tango wrote explainers describing how the new metric works and what exactly it means. Want to know who’s the best at scooping the ball, or corralling bouncers, or jumping up and snaring high throws and then landing on the bag before the runner? Now you can learn that information without having to watch all that bothersome baseball.

In the simplest terms, receiving works like other Statcast metrics. Each throw to the first baseman has a catch probability, which then maps onto out expectancy. Say a throw has a catch probability of 75%. A first baseman who makes the play earns 25% of an out, while the first baseman who muffs it loses 75%. With this data, I can tell you that this season, Alec Burleson has been the best at this skill by a wide margin, while Bryce Harper has been the worst by the same amount. That last part’s not necessarily a surprise, considering that six weeks before the new metric dropped, I wrote an entire article detailing ways in which Harper needed to improve at the cold corner, and that four weeks after that, the Phillies moved Harper off the position altogether (maybe they should’ve tried some of my fixes).

Within his explainer, Petriello noted that, as is so often the case in the game we love, the numbers tell us that receiving is not quite as important as we might want it to be. In our hearts, scooping the ball is the soul of the first base position. If you can’t pick it, you don’t belong there. But in reality, opportunities to scoop the ball come so infrequently that its value is in no way commensurate with the premium we place upon it. Spencer Torkelson and Jonathan Aranda sit atop the scooportunity leaderboard with 18 apiece, which means that they don’t even see one per week. The next time you attend your Saturday night Moneyball singalong and you shout along with Ron Washington that first base is incredibly hard, feel free to voice your own addendum: “But it doesn’t matter much anyway.”

Nevertheless, I haven’t been able to keep away from this leaderboard. Tango and Petriello explained a whole lot, but I still found myself overflowing with questions that this new data can finally help us answer. Scoops and bouncers and wide throws may not be as important as I want them to be, but that doesn’t make them any less interesting. I have a bunch of article ideas about this topic, but today, we’re starting small. Read the rest of this entry »

Baseball fans are often unhappy with projections, and they seem to enjoy sharing that unhappiness with me. That’s hardly surprising, of course — projections have large error bars (which we try to express when discussing them), and on some level, they represent a computer saying mean things about players you might like. You probably aren’t shocked by this, though you might be surprised to learn that the time of year I get the most complaints is during the stretch run rather than the preseason. And while some of the complaints I hear are about projections that missed the mark, the majority are about how little the in-season projections have changed since the start of the year. It can be frustrating when a player who is slugging .500 despite a preseason SLG projection of .350 only has a rest-of-season SLG projection of .400. Clearly, the conventional wisdom says, the projections are too slow to change.

Ben Clemens has written a number of pieces auditing our playoff odds, and while they aren’t as meticulously perfected as a 1996 Lexus, they’re pretty darn good. These calculations rely heavily on our in-season projections, which at their gooey, caramel center are an attempt to combine what we knew about a player or team before the season (in the form of preseason projections) with what we’ve learned since.

The relevant question then becomes how much a few months of performance should change what we thought about a player in March. And as it turns out, it’s less than the folks who find projections frustrating might think. When it comes to ZiPS, the preseason projections, both for teams and individual players, are more predictive of rest-of-season performance than season-to-date numbers. That holds true even very late in the season, when there’s been a lot of cumulative in-season performance to look at. I’ve tested preseason projections versus in-season performance as late as September 10 each year, and the preseason projections never clearly do worse than the in-season stats. Since we’re right at the start of September, I thought this would be a good opportunity to look at how the ZiPS preseason projections stack up against March-through-August performance when it comes to projecting what happens in the final month of the season.

And to be clear, I don’t think this preseason edge is exclusive to ZiPS; I expect that Steamer and PECOTA perform similarly well. It’s just that I have a rather unique level of access to ZiPS. But data is more important than a simple claim, so let’s dig in. Read the rest of this entry »

Behold, a table. Does anything strike you as odd about this leaderboard?

Some of those guys are just the best relievers in baseball and have been for a while. We’ve also got a couple breakout relief aces from this season, including Brent Headrick, Erik Sabrowski, and Samy Natera Jr., who is apparently not the lead singer from Chicago. Don’t feel bad, it’s an easy mistake to make. Read the rest of this entry »

Welcome to Mathemadage, a new series where I investigate time-honored baseball adages using whatever data I can get my hands on and explore whether there’s any validity to them. I’ll apply my normal Ben Clemens mixture of a bit of math, a bit of intuition, and plenty of healthy doubt in declaring strong solutions to complicated problems. I hope to demonstrate that some of these true and others false, but I’m sure that many will end up in a suggestive but unprovable middle ground. And true to the premise of the series, we’re starting off with a fun adage that resists easy conclusions.

If you watch a baseball game from start to finish, you’re pretty much guaranteed to see a hitter get tied up on an inside fastball. “Got jammed,” the announcer might say, or maybe, “He just couldn’t get around on that one.” It’s a visual reminder of one of the pitching truisms that I learned as a kid and haven’t forgotten since: hard in, soft away.

I intuitively look for this pitching pattern when I’m watching a game. Hitters look incredibly uncomfortable when they’re trying to get their bat around a fastball in on their hands. Now that we have a few years of bat tracking data, though, I can do better than just vaguely searching for this effect. So this week, when I saw someone get jammed, I winced in sympathy and then started furiously downloading spreadsheets. Read the rest of this entry »

Today at FanGraphs, we’re introducing an updated approach to prospect valuation. You can read the announcement here, and also see the new Farm System Rankings for 2026 on The Board. This post is a detailed methodological examination of how we’ve produced our new estimates. It goes over each step of the process in order, and concludes with a sensitivity analysis. If you’re interested in the broad strokes of our new approach, the introductory post will likely suffice. But if you want to see how the sausage is made, read on.

Prospect Classes
We began with Baseball America’s annual Top 100 prospect lists for each year from 2005-2016, plus FanGraphs’ lists for 2017 and 2018. The BA lists serve as a publicly accessible bridge to the current era of FanGraphs prospect writing, and provide a nice through line with Craig Edwards’ earlier research. We took all instances of a prospect being ranked, including duplicates of the same prospect in multiple years. We converted those ordinal rankings into Future Value grades using a two-step process. First, we separated the rankings into pitchers and hitters and created two separate ordinal lists for each year. Second, we adjusted those ordinal rankings between years by a regressed factor based on that class’ major league production. This allowed us to differentiate between classes – without some type of delineation between years, every top overall hitter would receive the same grade, which is contrary to the way we grade prospects.

This method introduces some potential bias. Judging prospects based on how they turned out inherently brings some information from the future into the mix. We decided that this was the best possible way to systematically introduce varying year-over-year quality to an otherwise ordinal-only set of values, and that it also did a good job of replicating the way that grades might have actually been assigned in the past. The top pitching prospect on the 2010 list was Stephen Strasburg. The top pitching prospect on the 2011 list was Julio Teheran. It’s important to differentiate between the likely grade that they would have received. There’s some volatility in relative value assignment at the very top end of the scale based on this methodology, which is addressed in the sensitivity analysis. Read the rest of this entry »

Seven years ago, Craig Edwards published a landmark study on prospect valuation. Craig’s work built on previous studies by Victor Wang, Scott McKinney, Kevin Creagh, Steve DiMiceli, and our own Jeff Zimmerman, as well as a few prior ad hoc attempts here at FanGraphs; subsequent work on the subject was done by the team at Driveline Baseball. These studies have been hugely important both for FanGraphs’ own evaluation of prospects — among other things, Craig’s work has helped to feed the Farm System Rankings over on The Board — and for the broader public study of the minor leagues.

The reasoning behind these studies is clear and simple. If you want to evaluate a prospect-for-big-leaguer trade, you’ll need to know the expected value of the prospect in the trade. If you want to evaluate how much help is waiting in a given team’s farm system, a quantitative assessment of the talent there is necessary. Even if you’re just wondering how likely your team is to find the next big thing, again, you’ll need some type of framework to understand how often that’s happened in the past.

The previous studies of prospect valuation are still excellent, but they’re all very much of their time. Since Craig published his study in November 2018, the league has changed significantly. The COVID-abbreviated 2020 season changed minor league timelines across the board. The league contracted the number of minor league franchises significantly in 2021. A new CBA, signed before the 2022 season, changed compensation structures and competitive balance tax levels, and introduced the Prospect Promotion Incentive. The cost of a win in free agency has skyrocketed; league-wide payrolls are up more than 30%, and free agent salaries are up by more than that. Read the rest of this entry »

If you clicked through to read this post, you’ve probably visited the ABS Challenge Leaderboard on Baseball Savant at some point this season. While you were there, you may have sorted by Won% to see which players have been the most successful with their challenges. And if you, like me, are a bit of hater, you also reverse sorted to see which players are now considered a fire hazard because of how rapidly they burn through challenges. In that case, you know that James Wood has won just 20% of his 15 challenges, that Josh Naylor owns a 25% success rate on 12 challenges, and that Jazz Chisholm Jr. has a 27% hit rate on his 15 challenges. Players this bad at picking their spots probably shouldn’t be allowed to challenge at all, right? Well the truth is, those samples probably aren’t large enough to definitively signal an inability to consistently win ABS challenges. Or maybe they are large enough, but it’s tough to say for sure because the ABS challenge system hasn’t been in place long enough to generate the volume of data needed to determine an appropriate sample size.

But even if there were absolute certainty about which players lack the eye for challenging ball/strike calls, sitting a player down and telling him he’s not allowed to challenge anymore because he sucks at it isn’t exactly the best strategy. It runs the risk of damaging the relationship between the player and the team and it shuts down the opportunity to improve with additional reps. And let’s say that player is in the box for a pitch that absolutely should be challenged — given the short window to challenge following a call, a batter paralyzed by self-doubt or concern over potential reprimand is set up to fail. It’s also much easier to communicate and get buy-in on a single, team-wide philosophy than it is to devise a bunch of player-specific exceptions to the rule.

The good news is that there’s a straightforward method for eliminating many of the most infuriating failed challenges, a method independent of any given player’s ability to judge whether a pitch was in the zone. Because there’s more to challenging than assessing whether a ball/strike call is correct and then assigning a level of certainty to that assessment. If you’ve ever watched a batter on your favorite team spend a challenge on an 0-0 count in the first inning, you know that it’s vexing on multiple levels. Even a successful challenge in that scenario doesn’t offer a significant swing in advantage, since it’s just flipping an 0-1 count to a 1-0 count (a swing in run expectancy of about a tenth of a run, depending on the base-out state). And to make things even more maddening, it also tightens the calculus around future challenges, since an additional failed challenge risks leaving the team unable to act on a potential missed call in a late-and-close situation. Read the rest of this entry »

On April 15, Zach Neto was at the plate with one out and nobody on in the top of the fifth inning of the Angels’ game in the Bronx, where his team trailed the Yankees, 3-2. The first two pitches, a low changeup and a high slider, were nowhere near the zone, and Neto laid off easily. The 2-0 pitch from Luis Gil was another slider, this one about belt high and bending away from the right-handed Neto, who kept the bat on his shoulder and watched as the pitch appeared to clip the outside edge of the zone. Home plate umpire Lance Barksdale held up his hand. Strike one. Neto tapped his helmet immediately to challenge the call.

The graphic on the gigantic video board in center field showed that the pitch had missed by 0.4 inches. The call was overturned; the count was now 3-0. Neto walked on the next pitch. Mike Trout stepped in, took a fifth straight ball from Gil, then let a four-seam fastball over the heart of the plate get deep on him. He unloaded, clobbering the cookie 383 feet into the right field seats for a go-ahead two-run blast.

The no-doubt Trout clout would have been the decisive blow in an Angels win if not for a misplayed popup and a Jordan Romano meltdown. The Yankees walked it off on a José Caballero single, relegating Neto’s challenge to a footnote in that night’s game story, if it was mentioned at all. Even so, the gamble was an early example of how the new automated ball-strike challenge system can make the difference between winning and losing a game. Read the rest of this entry »