The week in Braves Win Probability

The first week of the Braves’ season went about as well as any fan could have hoped.  They opened with a sweep of the Phillies and came home to take two of three against the Mets.  Let’s take a look at it from the viewpoint of Win Probability Added, a fantastic stat that is tracked by a fantastic site: Fangraphs.  First, let’s have a quick refresher on Win Probability, and then I’ll go into one more stat I’m tracking on top of what Fangraphs does.

Introduction

The Win Probability numbers I’ll be referencing are based upon a series of tables which were developed for the most part from empirical data over a large number of major league seasons (I’m not sure which ones, or I’d tell you).  Some of the numbers have been derived theoretically because of sample size issues (i.e. how many times do you actually see a situation where the home team is batting with the road team ahead 13-1 in the bottom of the first with bases loaded and two out?).

That may or may not mean anything to you, so here’s the real point: from these tables, we can take each situation in a game and find out what a team’s win probability is at that point.  We can also find out how much win probability a player adds for a particular at bat or inning pitched.  Thus, we have Win Probability Added, or WPA.

As a further point of explanation, let me say that a team will have +.500 in total WPA for a win and -.500 for a loss, since each team starts the game with a 50% chance of winning.

Next, it’s important to understand the concept of leverage.  Some situations in a game are clearly more important than others, and Leverage Index (LI) addresses that issue.  It’s easier to add win probability if every time you come to the plate, the bases are loaded in the bottom of the ninth of a tie game.  Leverage Index considers an “average” situation to be worth 1.00 of LI, while more crucial situations have higher numbers (sometimes upwards of 5.00), and less crucial situations have an LI of less than 1.00 (sometimes almost zero).

Fangraphs tracks both WPA and Leverage Index (they use pLI, which is the average LI per plate appearance or batter faced), but I’m going one step further on here.  If you read my WPA posts last year, you might remember a stat for pitchers that I called WPTP.  Basically, I measured the amount of win probability added each inning and divided it by the amount of WPA possible for that inning, with the effect being that it neutralized the pitchers’ WPA stats based on their usage (in crucial situations or not).  I’ll be doing the same thing this year, but since Fangraphs tracks LI instead of P (the stat I used for my work), I’m going to call it Leverage Neutral Win Probability, or just LWP.

LWP should work the same for both hitters and pitchers, and you can actually take it one step further and calculate LWP per plate appearance or per batter faced.  I’ll show those numbers (multiplied by 100 to make them a bit easier to read) and call them pLWP.  Now, let’s get to the stats.

April 2-8

Like I said before, this was about the best week a Braves fan could have hoped for.  Since they went 5-1, or four games over .500, there was a total of +2.000 in WPA to be handed out.  Here’s how it all broke down:

Hitting PA AVG OBP SLG OPS BRAA WPA pLI LWP pLWP
Brian McCann 24 0.364 0.417 0.727 1.144 3.35 0.752 1.33 0.564 2.349
Edgar Renteria 26 0.333 0.385 0.583 0.968 1.97 0.452 1.32 0.343 1.319
Scott Thorman 13 0.167 0.154 0.500 0.654 0.75 0.254 1.07 0.238 1.831
Matt Diaz 13 0.333 0.308 0.583 0.891 1.50 0.246 1.38 0.179 1.375
Chipper Jones 27 0.227 0.370 0.273 0.643 -1.47 0.082 0.99 0.083 0.306
Pete Orr 3 0.333 0.333 0.333 0.667 -0.12 0.031 0.40 0.077 2.567
Chuck James 2 0.000 0.500 0.000 0.500 0.05 0.012 0.83 0.014 0.723
Kelly Johnson 24 0.150 0.292 0.450 0.742 0.05 0.008 1.21 0.007 0.028
Chris Woodward 8 0.167 0.250 0.167 0.417 -0.90 -0.027 1.39 -0.019 -0.242
Jeff Francoeur 24 0.292 0.292 0.500 0.792 -0.80 -0.033 1.45 -0.023 -0.095
Tim Hudson 2 0.000 0.000 0.000 0.000 -0.35 -0.039 0.71 -0.055 -2.746
Mark Redman 2 0.000 0.000 0.000 0.000 -0.48 -0.058 1.18 -0.049 -2.468
Kyle Davies 2 0.000 0.000 0.000 0.000 -0.49 -0.062 1.81 -0.034 -1.713
John Smoltz 4 0.000 0.000 0.000 0.000 -0.83 -0.077 0.74 -0.105 -2.619
Brayan Pena 2 0.000 0.000 0.000 0.000 -0.90 -0.175 3.45 -0.051 -2.536
Ryan Langerhans 13 0.100 0.308 0.100 0.408 -1.85 -0.208 1.18 -0.176 -1.352
Craig Wilson 14 0.091 0.286 0.091 0.377 -1.38 -0.262 1.26 -0.208 -1.484
Andruw Jones 26 0.130 0.231 0.217 0.448 -2.20 -0.440 1.42 -0.309 -1.190
Pitching G GS IP ERA WHIP BRAA WPA pLI LWP pLWP
Bob Wickman 4 0 4.0 0.00 1.25 2.12 0.598 2.65 0.225 1.325
Rafael Soriano 5 0 3.7 0.00 0.27 2.90 0.435 1.56 0.278 2.318
Tim Hudson 1 1 7.0 1.29 0.86 2.70 0.223 0.79 0.281 1.080
Chuck James 1 1 5.0 1.80 1.40 1.64 0.124 0.99 0.125 0.570
Chad Paronto 4 0 2.3 0.00 1.71 -0.29 0.120 1.94 0.062 0.563
Macay McBride 3 0 2.0 13.50 4.50 -3.14 0.117 0.34 0.340 2.269
Kyle Davies 1 1 6.7 2.70 1.05 1.25 0.116 0.91 0.127 0.471
John Smoltz 2 2 12.0 3.75 1.67 1.34 0.011 1.30 0.008 0.015
Mike Gonzalez 4 0 3.7 2.45 3.00 -0.40 -0.018 0.96 -0.019 -0.094
Tyler Yates 2 0 1.7 10.80 3.60 -1.73 -0.021 0.18 -0.118 -0.986
Mark Redman 1 1 5.7 7.94 1.76 -0.28 -0.080 0.93 -0.086 -0.318
Oscar Villarreal 2 0 3.3 2.70 1.50 -0.96 -0.103 0.42 -0.244 -1.625

Even when you strip out the context of the Braves’ relatively close games (the team’s LI is 1.17), Brian McCann has been a force.  He’s already contributed 1.5 wins and is sporting a 1.144 OPS.  While he won’t keep up that pace, he may well be the best catcher in baseball.  He’s certainly the best in the NL, at the very least.

As a byproduct of the close games, the best relievers are getting in a ton of work already.  Rafael Soriano (he of the surgically repaired arm) has appeared in five games, while Wickman, Paronto, and Gonzalez (who has pitched a total of 147.3 innings in his last three seasons of work) have each appeared in four.  That pace can’t continue, but you have to hope that there will be fewer close games for them to appear in.

Big plays of the week:

  1. Scott Thorman’s go-ahead homer off Ryan Madson in the 11th inning of Wednesday’s 3-2 win (+.375)
  2. Edgar Renteria’s go-ahead 2-run homer off Madson in the 10th inning of Monday’s 5-3 win (+.340)
  3. Brian McCann’s 2-run tying homer off Tom Gordon in the 9th inning of Wednesday’s win (+.333)
  4. McCann’s game-tying double in the 8th inning of Sunday’s 3-2 win (+.286)
  5. Renteria’s game-tying blast in the 8th inning of Monday’s win (+.253)

Other notes:

  • Mike Hampton’s recent setback will send him for more tests, so it will once again be foolish to expect anything from him this year.  I’m just hoping the Braves’ front office will spend his salary on other players when he’s off the books.
  • Pete Orr, at 1-for-3 with a single, has been the most productive Brave on a per-plate-appearance leverage-neutral basis.  Beware of the small sample size.
  • Despite a 13.50 ERA, Macay McBride has been the second most productive pitcher.  He did well in his only crunch-time appearance, adding +.138 in win probability for throwing one pitch to Ryan Howard on Wednesday.

Upcoming Schedule: Tue-Thu at home against the Nationals, Fri-Sun against the Marlins, all at home.

Check back next week for another update.

UPDATE: Maybe it won’t take a week.  Hampton was shut down for 6-9 months today. 

Advertisements

3 thoughts on “The week in Braves Win Probability

  1. I’ve always been a mild Mike Hampton fan, but this is getting ridiculous. The Braves should see if they can trade him for something useful, like a couple of Gatorade coolers or something.

  2. So here’s something I’ve wondered about WPA that I never got around to asking you:

    It seems like there could be a problem with the fact that win probabilities for most game situations are calculated empirically, but the win probabilities at the start of the game are set to .500. In actuality, the home team wins more than exactly 50% of the time (I think it’s like .540 or so), and this is going to be reflected in the empirically-calculated win probabilities for different game situations.

    For example: suppose the first pitch of the game is a ball. This is slightly good for the visiting team, but only very slightly. I bet the home team still probably wins like .537 of all games where the first pitch was a ball, so let’s suppose the empirical win probability for no score, 1-0 count, 0 outs, bases empty, top 1st is .537. Then that means that the home pitcher’s WPA for throwing a ball on the first pitch is +.037! That can’t be right.

    So why not start it at .540 (or whatever) for the home team and .460 (or whatever) for the visiting team?

  3. Doug, you’ve hit the core of a very long debate that’s been going on at Fangraphs. Tangotiger, who came up with the tables they use, has been arguing for the status quo. User “Okay Fine” has argued for a version of WPA that incorporates home field advantage.

    So far, I’ve tended to side with Tangotiger, mainly for the reason that teams play an equal number of home and road games, and the effect mostly washes out. Okay Fine came up with some numbers showing just how big the HFA effect is, though, and if he’s right, there’s a ton of error included in the tables used for WPA.

    I suggest you read the whole argument on the Fangraphs forums and come to your own conclusion about it. Maybe you can help me understand their arguments.

    And Luke, am I correct in remembering that the Braves traded a case of balls and bats to the independent team they got Kerry Ligtenberg from a few years back? Perhaps this is payback for that little deal. I just hope it’s all over soon.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s