Forecasting October with Playoff Runs

If you’ve paid any attention to baseball over the past decade or so, you know that playoff results are governed by something other than logic. There is no formula that can identify a great playoff team. October baseball is driven by some combination of heart, guts, grit, and randomness- a little heavier on the last one.

That paragraph served as the introduction to my 2015 MLB playoff preview- an excuse, perhaps, for the inevitable failure of the playoff runs model I’d just created. As it turns out, Playoff Runs worked pretty well, whether by chance or due to some underlying validity. I correctly predicted the winners of both Wild Card games, three of four Division Series, and the Mets beating the Cubs in the NLCS, a result that probably surprised many observers. All three series I missed were the result of the Royals’ playoff magic.

Perhaps I should rest on those laurels, tell everyone I knew the Mets were better than they looked, and shut the whole thing down. But I’d rather try again and risk going 2-7 this year and losing both of my followers on Twitter.

Here’s a link to last year’s piece, which goes into more detail about the process than I will below.

Essentially, Playoff Runs assigns roles to 25 players on each playoff team, assumes a certain number of innings pitched, plate appearances, and defensive chances for each role, and uses Fangraphs’ Runs Above Replacement (for pitchers and hitters) and Runs Above Average (for fielders and baserunners) to determine how much value each player will provide over those innings and plate appearances.

Like any prediction system, and particularly one that tries to standardize each team’s roster construction, it has its flaws, but I prefer it to a simple observation of each team’s 2016 results because it gives full credit to the players on the roster right now, leaving off injured players while giving full credit to called-up prospects and trade acquisitions.

Rather than simply ranking the teams based on their total scores, let’s mock the playoffs by assuming the team with the better Playoff Runs score for the appropriate length series (1 game, best-of-five, or best-of-seven) wins every time. Here’s guessing the result won’t be as surprising as last year’s.

American League Wild Card Game

Blue Jays (3.43) over Orioles (3.06)

Why the Blue Jays Win This Simulation

The Blue Jays basically do everything one can do on a baseball field better than the Orioles do. Baltimore hits more home runs, but not enough to offset the Blue Jays’ major advantage when it comes to avoiding outs. Baltimore has the better bullpen, particularly at the back end, but that’s not enough to offset Toronto’s far better starting pitching. Neither team is using its best pitcher in the Wild Card game, but Marcus Stroman stacks up better than Chris Tillman, mostly thanks to a better walk rate.  Baltimore, by a miniscule margin, is the weakest team in the playoffs according to Playoff Runs.

What the Model Doesn’t Consider

Buck Showalter has gotten far more out of the Orioles than their talent would suggest all year. If Baltimore can keep the game close in the early innings, Brad Brach and Zach Britton could be the perfect antidote to the Donaldson-Bautista-Encarnacion murderer’s row.

National League Wild Card Game

Mets (3.85) over Giants (3.57)

Why the Mets Win This Simulation

This is where the Playoff Runs system is going to get accused of being biased in the Mets’ favor. No Harvey, no deGrom, no Matz, no Wright, no Walker, and these guys are supposed to beat Bumgarner? Well, as we learned last year, these Mets are good. Fangraphs is FIP-based, so Noah Syndergaard’s 218 strikeouts against 43 walks make him look like a pretty good option in an all-or-nothing game. New York has the better bullpen and a better offense (through these two teams hold the bottom two spots in offensive Playoff Runs). Home field advantage won’t hurt either.

What the Model Doesn’t Consider

I’m not sure any quantitative model would have predicted any of the three championships the Giants have won over the past six years. This tends to be the position they find themselves in when they start catching breaks and sneaking past better teams. If I knew how to assign points to Bruce Bochy, I’d give him some. The model also gives no credit to Bumgarner’s bat, which is certainly worth more than nothing.

American League Division Series

Blue Jays (13.84) over Rangers (11.92)

Why the Blue Jays Win this Simulation

The only thing the Rangers do well is win. That may seem like an excellent formula for playoff success, but based on the underlying results, this team should have been closer to .500 than the top seed in the AL. If they hadn’t added Jonathan Lucroy and Carlos Beltran at the trade deadline, they would be by far the worst contender by Playoff Runs, but that improved offense is enough for them to squeak past the Orioles. Meanwhile, the Blue Jays hit better, pitch better, and play better defense than Texas.

What the Model Doesn’t Consider

Like the Orioles, the Rangers tend to find a way to win. The model doesn’t love Cole Hamels because he walked a batter every other inning all year, but he prevented runs at an elite level and the lineup behind him is legitimate. There’s little chance this one matches the intensity of last year’s battle, but it’s likely to be close.

Red Sox (16.24) over Indians (12.55)

Why the Red Sox Win this Simulation

This was a truly great Indians team. Unfortunately, Playoff Runs doesn’t see it as a great team now. Carlos Carrasco and Yan Gomes are out, Corey Kluber isn’t likely to be ready for Game 1, and Danny Salazar is hoping to pitch a few innings out of the bullpen. Cleveland has the third-best offense among playoff teams, but they’re playing the team with the best offense by a crazy margin. Boston also has far better starting pitching and scores well on defense and in the bullpen.

What the Model Doesn’t Consider

Craig Kimbrel has completely lost the plate over the past week, and Boston’s strong bullpen is far less formidable if he’s not throwing strikes. Kluber could come back to pitch games two and five, while Eduardo Rodriguez and Clay Buchholz are question marks in Boston’s rotation. Andrew Miller and Cody Allen will make Terry Francona more comfortable in the late innings than he would be if he were still managing the Red Sox. This series won’t be the blowout the numbers suggest.

National League Division Series

Cubs (18.21) over Mets (13.24)

Why the Cubs Win this Simulation

The Mets are good, but they’re nowhere near Cub-caliber. Jake Arrieta, Jon Lester, and Kyle Hendricks are all among the best run-preventers in Major League Baseball, whether on their own merits or thanks to Addison Russell, Jason Heyward, and company on the field behind them. Chicago’s defense scores more than twice as many Playoff Runs as any other team still playing. Their offense, paced by likely unanimous MVP Kris Bryant, tops that of everyone but the Red Sox. The only weakness this team has is the letter on their caps.

What the Model Doesn’t Consider

That letter on the Cubs’ caps doesn’t stand for championship. And their opponent is certain to be either the team that swept them out of the playoffs last year or the team that always wins the World Series in even years. The scariest phrase to a Cubs fan right now is probably “anything can happen in baseball”.

Dodgers (17.90) over Nationals (13.33)

Why the Dodgers Win this Simulation

Like the Cubs, these Dodgers do everything well. The best pitchers in the playoffs by Runs Above Replacement per inning pitched are Aroldis Chapman, Kenley Jansen, Andrew Miller, Clayton Kershaw, Zach Britton, and Rich Hill. You may have noticed that half of them wear Dodger blue. Corey Seager, Joc Pederson, and Justin Turner pace the fourth-best offense left standing, while Seager, Turner, and two good catchers lead the third-best defense.

What the Model Doesn’t Consider

Max Scherzer may win the NL Cy Young award and Tanner Roark is right on his heels. If Stephen Strasburg can come back (the model considers him the fourth starter, giving him only three innings pitched in the series), this team can pitch with LA, who will depend on more health from Kershaw and Hill than they got all year. Trea Turner has emerged as a young star, and Bryce Harper’s 2016 numbers probably sell short his ability to dominate a playoff series.

American League Championship Series

Red Sox (22.51) over Blue Jays (19.02)

Why the Red Sox Win this Simulation

Boston’s offense is enough to carry them past just about anyone. They play excellent defense, particularly in the outfield, and Rick Porcello and David Price form a formidable one-two punch in the rotation. As much power as the middle of Toronto’s lineup packs, Boston’s is just relentless, with Sandy Leon, Jackie Bradley, Jr., and Andrew Benintendi forming the best bottom three in the playoffs by far.

What the Model Doesn’t Consider

Again, Boston’s bullpen will have to stare down three great hitters and survive Toronto’s own offensive attack. Aaron Sanchez, Marco Estrada, JA Happ, and Marcus Stroman form the deepest rotation on the AL side of the playoffs. The matchups in Games 3 and 4 in Toronto would likely favor the Jays.

National League Championship Series

Cubs (25.10) over Dodgers (24.73)

Why the Cubs Win this Simulation

Last year, the one surprise revealed by this model was that the Mets might have had the best team in the playoffs. If there’s a surprise this year, it’s that the Dodgers are this close to the Cubs. Kershaw, Hill, and Kenta Maeda would be a handful for any team, and the Dodger bats will be a tall order for the Cubs’ pitching and defense. But the model still prefers the Cubs by a hair, due entirely to their otherworldly defense. In reality, this would be an incredibly fun NLCS, with two great rotations and lights-out closers trying to lead their respective teams to the World Series for the first time in 28 or 71 years.

What the Model Doesn’t Consider
Immense emotional gravity could pull this series out of the spreadsheets and reward the iron will of whatever team doesn’t crack under intense pressure. By Playoff Runs, this is the closest matchup of all the series I’m predicting, capable of swinging on one slider just off the corner that the ump calls a strike or one ball that kicks up the tiniest bit of chalk down the left field line. Of course, if Kershaw and Hill aren’t healthy, it might just be a Cubs sweep, leading us to some more emotional gravity in the form of…

World Series

Cubs (25.10) over Red Sox (22.51)

Why the Cubs Win this Simulation

I wonder how many times pundits have predicted this World Series since the matchup last happened in 1918. The storylines would be endless: the team mired in a 108-year title drought against the team that broke an 86-year drought only to try for their fourth in 12 years. Theo Epstein, Jon Lester, John Lackey, David Ross, Anthony Rizzo (kind of), and throngs of former Orioles, Yankees, and Rays coming home to Fenway. Two wild fanbases in two ancient stadiums. The top-flight pitching and defense of the Cubs against the relentless offensive attack of the Red Sox…

But it’s not that simple. The Cubs can pitch and field, but they can hit too. They’ll have more confidence in Aroldis Chapman closing out games than Boston will in Kimbrel, the 2010-2014 version of Chapman. Chicago has the depth to crush Boston’s righties and lefties alike, and Jorge Soler (or Rizzo or Bryant or Ben Zobrist) will slot in at designated hitter without making the Cubs feel offensively disadvantaged at Fenway. It’s the Cubs’ year; everyone else is just along for the ride.

What the Model Doesn’t Consider

Randomness rules baseball far more than talent or drive or momentum. Just as the Red Sox would have at least a 40% chance to win this series despite the talent gap, this series is almost as likely to pit the Blue Jays against the Dodgers or the Giants against the Orioles. If you’re a Cubs fan, you just watched your team win 103 games and show ample evidence that it’s the team most likely to win the World Series. But “most likely” is something like a 20 to 25% chance. That’s baseball.

Let’s close with the ten playoff teams ranked by their aggregate Playoff Runs over a best-of-seven series:

1. Cubs (3rd in pitching, 1st in fielding, 2nd in hitting)

2. Dodgers (1st in pitching, 3rd in fielding, 4th in hitting)

3. Red Sox (7th in pitching, 4th in fielding, 1st in hitting)

4. Blue Jays (5th in pitching, 5th in fielding, 5th in hitting)

5. Nationals (4th in pitching, 9th in fielding, 6th in hitting)

6. Mets (2nd in pitching, 10th in fielding, 9th in hitting)

7. Giants (6th in pitching, 2nd in fielding, 10th in hitting)

8. Indians (10th in pitching, 6th in fielding, 3rd in hitting)

9. Rangers (8th in pitching, 7th in fielding, 7th in hitting)

10. Orioles (9th in pitching, 8th in fielding, 8th in hitting)

Click here for the companion piece, with player-level detail for every team at every position.

This entry was posted in Blue Jays, Cubs, Dodgers, Giants, Indians, Mets, Nationals, Orioles, Predictions, Rangers, Red Sox. Bookmark the permalink.

2 Responses to Forecasting October with Playoff Runs

  1. Pingback: Forecasting the 2016 Postseason – High Heat Stats

  2. Barrie Pollock says:

    Thanks, Bryan, for an interesting analysis.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s