clock menu more-arrow no yes

Filed under:

Predictive game scores: Opening Week

New, comment

The live implementation of the predictive Bill James game score.

Frank Victores-USA TODAY Sports

It is one thing to grab a sample of past games and see how a predictive formula does. It's a wholly different experience to put that predictive method against real games before they happen.

A couple months ago, I started working on a method to predict a game's outcome by running the statistics of the pitcher against that specific opposing lineup. It culminated in a three-part series published right here at Beyond the Box Score. You can read the conclusion here but, in short, it didn't work out the way I had hoped. That being said, it represented a step in the right direction for predictive analysis. At least for one person's research purposes.

Without much more delay we will get to the way the formula expected Opening Day to play out. But first, this is a very laborious exercise of data gathering and in a very compressed amount of time, because lineups are posted fairly late. I point this out to show you that a) this formula isn't very practical for actually predicting a game, and b) a better iteration could be used as more of a macro prediction (to predict a pitcher against a certain division's worth of players, for instance). In fact, it was so difficult to gather lineup data before the game happened that I didn't get it done in time. That is why this article will predict today's games instead. However, since I have the data collected anyway, let's take a look at how the formula expected Opening Day to play out.

Pitcher pBJGS Actual
Jon Lester 62 39
Adam Wainwright 78 68
Masahiro Tanaka 75 38
Drew Hutchison 27 63
David Price 80 79
Phil Hughes 50 45
Kyle Lohse 33 9
Kyle Kendrick 33 69
Cole Hamels 63 44
Clay Buchholz 61 79
Chris Archer 90 47
Chris Tillman 66 63
Max Scherzer 76 71
Bartolo Colon 66 69
Henderson Alvarez 66 59
Julio Teheran 78 55
Yordano Ventura 60 61
Jeff Samardzija 78 38
Felix Hernandez 83 78
Jered Weaver 66 41
Johnny Cueto 75 78
Francisco Liriano 69 69
Clayton Kershaw 109 55
James Shields 58 58
Dallas Keuchel 62 72
Corey Kluber 80 69
Sonny Gray 84 82
Yovani Gallardo 70 38
Josh Collmenter 65 27
Madison Bumgarner 90 63

Some stray observations before we get to tonight's predictions. First, the formula actually got one right! And it actually got two right! Francisco Liriano and James Shields' performances were spot-on the projections. Second, the formula clearly loves Kershaw way too much. This is likely a product of it depending on pitching trends. There's an implicit assumption in the formula that says 'hot' pitchers will stay 'hot' and 'cold' pitchers will stay 'cold.' It isn't weighted that much, but in Kershaw's case, it doesn't have to be on account of his machine-like performance last season.

It also likes Chris Archer way too much. That one is weird to me, but it probably has a lot to do with his splits. He faced 6 right-handers yesterday and has a career 0.263 wOBA against right-handed hitters. In other words, it might not be the best idea long-term to continue putting a lot of right-handed hitters in against Archer. The opposite can be said about Drew Hutchison. He faced 8 left-handed batters in yesterday's Yankees lineup and has a career 0.353 wOBA against left-handed hitters. I digress, it will be wrong often.

Without further delay, ladies and gentlemen, place your bets.*

* - please don't use this to gamble unless you plan on losing your college fund.

Pitcher pBJGS
Mat Latos 86
Alex Wood 78
Nate Karns 20
Wei-Yin Chen 58
Matt Garza 53
Jordan Lyles 37
Rubby De La Rosa 30
Ryan Vogelsong 61
Jesse Hahn 79
Colby Lewis 42
James Paxton 83
C.J. Wilson 57
Zack Greinke 82
Tyson Ross 64

. . .

All statistics courtesy of FanGraphs.

Michael Bradburn is a Featured Writer for Beyond the Box Score. You can follow him on Twitter at @mwbii. You can also reach him at michaelwbii@gmail.com