MLB

How Our MLB Forecast Is Changing For 2022

By Neil Paine and Jay Boice and Neil Paine and Jay Boice

Apr. 6, 2022, at 6:00 AM

After an extra-long offseason of doubt and acrimony, baseball is finally back. Thursday’s slightly belated opening day will feature 18 teams in action — including many of the ones we think will be the most interesting to watch throughout the year. To help you prepare for the new season, we have booted up our MLB prediction model once again — with a few twists this time around. Some are due to new format changes in the sport, while others involve our own improvements to the forecast. Read on to find out what’s new in the model and what it says about the favorites in the season to come.

The first major alteration is a structural one to the sport itself. As part of MLB’s new collective bargaining agreement, the playoff field is expanding this season from 10 teams to 12, giving the American and National Leagues one new wild-card slot each. The change has come with criticism that it might affect teams’ behavior or inject even more randomness into an already chaotic sport. But in the big picture, the new format changed surprisingly little in our World Series forecast. Here are the favorites’ odds for 2022, under both the old and new playoff systems:

Although a number of the top teams saw a reduction in their championship probability under the expanded format, the changes were mostly very small. (The Milwaukee Brewers were the biggest losers, with 0.5 percentage points shaved off their World Series odds; everyone else lost fewer than 0.4 percentage points.) The biggest effects came in terms of playoff chances, where six teams saw their odds of making the postseason rise by double-digit percentage points — a group headlined by the Boston Red Sox and their 11.3-point boost in playoff odds.

The change primarily helps talented teams that play in difficult divisions — like Boston, but also its AL East rivals in Tampa Bay and New York (or the Padres and Giants in the NL West, to use another example). These clubs are dangerous if they make the playoffs, but they would have been more likely to miss out on a postseason spot under the old system. By contrast, the new format hurts teams like Milwaukee and the Chicago White Sox, who are favorites in comparatively weaker divisions — and therefore need the extra wild-card slot less, while simultaneously being hurt by the increased probability of facing more dangerous teams in the postseason (particularly since only the two top-seeded division winners in each league receive byes out of the wild-card round now).

Though these changes may not make a huge difference to the overall World Series picture, it will be intriguing to see who uses them to their advantage — and whether the road to either league’s pennant goes through a team that wouldn’t have made the playoffs at all just last year. 

The other seismic change for this season is the adoption of the designated hitter in both leagues, forever eliminating the presence of light-hitting pitchers at the bottom of NL lineups. (There’s also a special rule tweak made for two-way players such as Shohei Ohtani, allowing them to stay in the lineup at DH even after being replaced as a pitcher.) But while the ripple effects of the universal DH will be wide-ranging, they don’t necessarily show up in our model beyond how the extra firepower in NL lineups affects their ability to score runs and win games.

However, we did introduce a new wrinkle to the model to handle another specific type of player: the opener. Under normal circumstances, each starting pitcher carries a pitcher rating — based on a rolling average of his performance over time — which is then used to modify his team’s chances of winning in any given start (by comparing his rating to the performance the team typically gets out of its starters). This works well when pitchers are used conventionally, which remains true in the vast majority of games, but not so well when teams use a relief pitcher to “open” a game before bringing in another pitcher (or pitchers) to handle the bulk of the subsequent innings.

The most high-profile instance of openers confusing our model came in last year’s National League Division Series, when the Los Angeles Dodgers opted to start reliever Corey Knebel in their pivotal Game 5 contest against the Giants. Knebel’s low pitcher rating artificially depressed L.A.’s chances of winning in our forecast, especially when compared with the rating of eventual “follower” Julio Urías. So in response, we’ve changed how our model handles the pitching adjustments in games involving openers.

First, we use a combination of previous statistics to identify cases where an opener is likely being used.¹¹ Then, in games featuring an opener, we use the average pitcher rating of available starters on the roster when calculating a team’s pitching adjustment, rather than the rating for the opener. (This effectively estimates the expected performance of the follower who will enter the game after the opener is done.) Finally, we also ignore the opener’s pitching performance in the game for the purposes of updating the team’s rolling pitcher rating, to avoid misleading stats affecting the team’s forecast going forward.

We hope this change will improve the model’s performance — even if only slightly in the grand scheme of the season — and prevent cases where a team’s odds are being unfairly penalized when they use an opener, particularly in the playoffs. With an expanded postseason field this year, there will be more opportunities than ever to look for small edges like the opener strategy, and our model will help you keep track of who’s inching ahead in the World Series race all season long.