As the Astros move forward in a playoff race over the next 49 days, the FanGraphs playoff odds become increasingly interesting. As shown below, the Astros are neck-and-neck with the Rangers in the odds of winning the division, but the odds of winning a playoff spot (whether division champion or wild card) are quite good. The playoff odds change daily.
Fangraphs Playoff Odds
(Win division / playoff odds / projected wins / ROS strength of schedule)
Astros Division 46.3% / Playoffs 86.1% / 91.6 Wins / SOS .500
Rangers Division 48.7% / Playoffs 90.0% / 92.1 Wins / SOS .510
Mariners Division 4.9% / Playoffs 26.9% / 85.9 Wins / SOS .499
As of today, the division race is expected to be a barnburner. The Rangers are expected to win the division by less than one game. For the Rangers and Astros, the consolation prize is expected to be a wild card appearance. However, the Mariners have surged forward and have a reasonable chance of making the playoffs. The Mariners’ surge prevents either the Rangers or Astros from enjoying greater than 50% odds of winning the division at this point. The Rangers have a somewhat higher strength of schedule in the future, as compared to the Astros.
For the rest of the season, the Rangers will play 7 more road games than home games. So far, the Rangers win 51% of road games and 67% of home games. The Astros will play 2 more games at home than on the road. Currently, the Astros have a better winning percentage on the road (58% vs. 55%).
Rest of Season Projections
The Fangraphs odds expect some degree of regression by the Rangers, as indicated by a decrease in the Rangers’ current division lead (3 games) to a closer— less than 1 game— lead at season end. The Rangers’ current W% of .593 is expected to regress to a .513 W% for rest of season. (The Astros’ W% of .566 is expected to regress only slightly, ROS .562). This leads me to wonder why the regression is occurring. Player projections are a starting point, though not necessarily the sole cause.
The data source for the Fangraphs’ odds is the depth chart, which is based on projections of rest of season player performance and playing time. The player performance projections are a combination of ZIPS and Steamer results.
I downloaded the Rangers and Astros depth chart projections and attempted to analyze the regression or reversion produced by the ROS projections In this case, I consider “reversion to the mean” to be the same as regression in the direction of projected ROS performance. Four Excel files later, I produced some estimates based on weighted averages of the change in wRC+ and ERA compared to current performance.
The Fangraphs model implies fairly significant regression in the Rangers’ position player wRC+ and slightly positive reversion by Astros’ position players. (For this calculation, I used projected playing time to weigh players’ current performance, so that previous downtime due to injury is not counted as “regression.”)
Weighted Team wRC+: ROS vs. “Current”
(Regression in wRC+ and Percent Change in wRC+ )
Rangers -16.6 / -15%
Astros +1.19 / 1.01%
The Fangraphs model indicates both the Astros’ and Rangers’ pitching will experience ROS regression in the direction of a higher ERA. But the regression is small for the Astros and significantly higher for the Rangers. (For this calculation, I used projected playing time to weigh players’ current performance, and calculated trade deadline pitchers’ current performance based on their ERA with their current team.)
Weighted Team ERA: ROS vs. “Current”
(Regression in ERA “runs” and Percent Change in ERA )
Rangers +.37 earned runs / +10.3%
Astros +.13 earned runs / +3.4%
More details of the expected ROS trend:
- Every Rangers’ position player except for Robbie Grossman experiences a downward regression in wRC+. (Grossman’s ROS wRC+ is expected to increase by 10%.) This is another way of saying that the ROS projections believe the Rangers’ offense has been overperforming to this point in the season. Percentage regression for some notable extreme examples: Seager (-24%); A. Garcia (-20%), Duran (-19%); Jankowski (-21%).
- The Astros’ position players are evenly split between expectations of positive and negative regression—which is probably more normal than the Rangers’ uniformly negative regression. Examples of expected positive regression: Bregman (+16%), Abreu (+36%), Pena (+10%), Dubon (+13%). Examples of expected negative regression: McCormick (-25%), Altuve (-10%), Meyers (-7%), and Diaz (-6%).
- Given that traded players have only played a short time for their current team, regression is not a meaningful term. For your information: the ROS ERA projected for notable traded pitchers: Verlander (3.61), Graveman (3.95), Scherzer (4.06), Montgomery (4.01), Chapman (3.71).
- The Fangraphs model believes that the Rangers’ starting pitchers have over-performed to a significant degree. The starters’ ROS projected percentage increase to ERA: Eovaldi (39%), Gray (13%), Heaney (2%), and Dunning (41%). Perez is expected to improve by 9%. Examples of percentage increases projected for reliever ERAs: Will Smith (26%), LeClerc (36%), Burke (38%).
- For Astros’ starting pitchers, expected positive or negative percentage changes in ROS ERA: Valdez (+2%), Brown (-11%), Javier (-6%), France (+39%), Urquidy (-35%), Bielak (+17%). Examples of relievers expected to increase ROS ERA: Abreu (+29%) and Neris (+69%). Examples of relievers expected to decrease their ROS ERA: Montero (-41%) and S. Martinez (-29%).
My feeling is that the model correctly presumes that the Rangers have overperformed on both offense and pitching so far this season. I don’t know whether the magnitude is as large as the Fangraphs model indicates, though. In particular, it seems unusual that almost every member of the Rangers’ offense has over-performed so far. But this may be something that happens when teams have extraordinarily good seasons. And for all we know, this may be one of those seasons, and the projected regression will not occur.
As for the Astros, they should be able to keep the division race close as the season goes along But we already know some of the question marks which might determine the course of the season. Will Bregman’s offense result in a finale to the season as productive as his 2022? Is Jose Abreu capable of producing positive regression as projected? Will Yainer Diaz receive more playing time in the future? Can Chas McCormick avoid the offensive regression projected by the model?