OOPSY! Baseball's newest - and most optimistic - player projection system

If you're a fantasy baseball nerd like me, you likely spend some time examining FanGraphs player pages. Over time, you might have noticed that the number of projections for each player multiplied.

The forecasts on the individual player pages are highlighted in green below the players' yearly historical performance records. The projections often occupy more rows than the annual output.

There are now eight projection forecasts for each player, and that's just at FanGraphs, which doesn't include the dozens of other forecasts floating around the internet.

Don't get me wrong, I love forecasts. They're useful tools, and a lot of work goes into creating them. But you may have noticed that many of them produce similar results.

Consider Aaron Judge's forecast for 2025 (shown above). The following are his projected home runs, runs, RBIs, steals, and batting average - the five rotisserie baseball categories at FanGraphs:

The first seven lines are powered by four systems: Steamer, ZiPS, ATC, and The Bat, and variations of those systems adjusted by playing time or slightly different inputs. The near uniformity applies not just to Judge but to many players.

However, there's an outlier in the newest projection system published by FanGraphs called OOPSY. It projects Judge's output at significantly higher rates than the others.

When its creator, Jordan Rosenblum, introduced his system through a blog post on FanGraphs earlier this year, one commentator described it as "the crazy uncle" of projection forecasts.

"I thought that was pretty funny," Rosenblum told me in February.

He wasn't insulted. He wanted his system to be different.

Rosenblum became interested in fantasy sports as a competitive outlet with his friends while in high school in New Jersey. He was always interested in statistics, earning his business degree at the University of Maryland before picking up a master's degree in public policy at Syddansk University in Denmark. That's where Rosenblum met his wife, who he followed to Finland.

He jokes that he really moved to Finland, where he's pursuing a doctorate in social policy at Åbo Akademi University, to gain a time-zone advantage for free-agent pickups in his Yahoo! fantasy league.

To understand why OOPSY is different, let's understand how other projection systems converged.

Rosenblum explained that most systems have the same underlying ingredients. That makes sense because those ingredients are tried, tested, and useful.

Some assumptions and inputs will vary, but systems generally consider a player's historical performance, weighing recent performance more heavily, and then compare Player X to players like him at that age in baseball history. Some systems may apply an aging curve model to assess how performance may grow or decay. Playing time is estimated, ballpark factors are often added, and numbers are regressed to the league average.

The last idea helps draw the projections into a narrow range. "People get annoyed when there is too much regression to the mean," Rosenblum said.

These systems rely on major-league performance history and can adapt slowly to new information.

Systems generally have more trouble projecting rookies and foreign professionals making their MLB debuts. And when new data emerges, projection creators typically want time to study it to determine whether it's predictive.

Rosenblum wondered if there was a way to make a projection system that was more nimble and quicker to adjust to meaningful changes, even if those changes came within small samples.

"My preferred approach is to look for variables that capture signal more quickly," he said.

Creating such a system would benefit not only season-long projections - which are important for fantasy owners in their draft preparation - but also weekly projections - which are critical for guidance when setting lineups.

Like other systems, OOPSY uses players' historical performance and aging curves, regresses to league averages, and adjusts for ballparks. Rosenblum researched other systems and dove into the work of analytics pioneer Tom Tango. One Tango finding stuck with him: aging effects are generally twice as steep on the way up as they are on the way down.

To create a more responsive, dynamic projection, Rosenblum searched for "sticky" skills that don't need a large sample to become meaningful.

He began statistically testing his assumptions and how much impact they had on performance. One big difference with OOPSY is that it includes swing speed, a new skill measure released last year by MLB's Statcast system.

OOPSY's inclusion of swing speed helps explain, in part, projections like Judge's, as well as more polarizing ones, such as Giancarlo Stanton's (before his spring injury).

"The good thing about adding in these really sticky measures like barrel rate and swing speed, which only take a few swings to show its true self or stabilize, is they allow you to be more bullish at the extremes," Rosenblum said. "OOPSY can kinda stick its neck out more. If you look at (Shohei) Ohtani, (Ronald) Acuña, and Yordan Alvarez, I am kind of more bullish on all of these guys because of the bat speed, because of the barrel rate.

"The more we can capture that signal, if the projections account for stuff like that, then they can better differentiate between really good and average players. … It's not like I'm assuming a higher league average. My average wRC is the same as Depth Charts or Steamer. (OOPSY) seems to be better able to distinguish between the elite and average dudes."

OOPSY gets its first public test this season. In the introduction post on FanGraphs, titled "Yet Another Projection System," Rosenblum explained why the baseball world needed another forecast and how he arrived at a name.

"OOPSY is a fake acronym, an homage to making errors while playing the infield growing up, and to making mistakes more generally, the fallibility of humankind," Rosenblum wrote. "It aims to stay humble, to be transparent about the fact that projection systems are built upon different assumptions."

Projections will never be perfect. What we hope for is that forecasters keep refining their processes and assumptions, try to be a little less wrong, and inch closer to predicting the future.

Travis Sawchik is theScore's senior baseball writer.