A look at the most valuable players, and biggest plays, of the first half

I’ve had some time on my hands (just kidding – I’ve really had no more time than usual) and been studying more deeply some various baseball statistics. And the more I study, the more I become intrigued by win probabilities. So this is (hopefully) the first in a series of articles examining the Astros (and perhaps some of the rest of the league) through the lens of win probability.

Win expectancy

Win expectancy is a fairly simple concept. For any given combination of inning, outs, runners on base, and runs ahead or behind, a team has a statistical likelihood (based upon historical outcomes) of winning the game.

To take it to an extreme example:

Let’s suppose the visitors are down by two runs in the 9th, with two outs, and runners on first and second. From 1957-2013, there have been a total of 1294 games that were in this situation; the visitors won only 70 of them. Therefore, the visitors have only a win expectancy of 0.05%.

But what if the next batter homers? Well, then you have the visitors up by 1 with 2 out and nobody on. There have been 6694 games in that situation, with the visitors winning 5548 – a win expectancy of 82.88%.

Win probability added (WPA)

Given these win expectancies, or win probabilities, it then becomes a simple exercise to determine how much a given play added to, or subtracted from, a team’s win probability. This is referred to as “win probability added”, or WPA.

In the above example, the home run would be a win probability added of (82.88% – 0.05%) = 82.83%. As mentioned, this is an extreme, as the typical play in the course of a game will be less than 1% WPA (positive or negative).

Why WPA?

Others have written that, if forced to look at only one overall statistic, WAR (or some form thereof) is the best reference. I disagree for the purposes of answering the following questions:

1. Which player has contributed the most to the ends of actually winning games?
2. Which plays have been the most meaningful to the ends of actually winning games?

For the second question, I believe that this is a self-evident advantage. Other statistics – whether rate statistics like BA, SLG, OBP, RC/27, or traditional counting statistics – do not differentiate for the end result of the play. Counting singles or strikeouts tells you nothing of what those plays contributed to the game; all context is removed.

For the first question, we should consider several factors about WPA:
1. It is objective. There is no consideration of a theoretical “replacement player” as used by VORP or WAR; win expectancies are generated entirely according to the record of all results since 1957.
2. It properly reflects the zero-sum nature of the game. For any given play, either the offense or the defense just moved closer to a win. Theoretically, the league should have a mean WPA of zero, because the pitcher is credited with the opposite WPA of the hitter for each play.
3. It properly rewards situational play. A strikeout with a runner on third and less than two outs IS different than a strikeout with nobody on; as fans, we all recognize this, so relevant statistics should recognize this as well.

With that preamble, let’s count down the 10 most valuable Astros of the first half, as ranked by WPA:

10 (tie) – Robbie Grossman and Jason Castro, 0.33%

Going into the season, Castro was viewed as a lynchpin of the team, counted on to give middle-of-the-lineup production from the catcher position. Robbie Grossman was going to hopefully be an everyday left fielder. Neither of these have occurred, but the fact that Castro is barely in the upper half of the squad is as disappointing as Grossman having a positive WPA is surprising.

9 – Collin McHugh, 0.34%

We now start to run into a flaw within WPA: because win probability will change more quickly in later innings, relievers tend to have disproportionately higher WPA than starters.
Regardless, we I think we can agree that nobody would trade McHugh straight up for…

7 (tie) – Darrin Downs, 0.44% …
and you certainly wouldn’t hesitate to trade Downs straight up to get…
7 (tie) – George Springer, 0.44%

Could a relief pitcher possibly be as valuable as one of the most exciting rookies in the game? Well, let’s consider a few factors:

1. Springer’s outs frequently occur with runners (cough, ALTUVE, cough) on base, and are therefore a “worse” out under WPA.
2. Downs has been in multiple “high-leverage” situations, as he is one of the only relievers Porter can depend upon.
3. WPA has no idea how far a #GeorgeGorge can travel.

6 – Dexter Fowler, 0.76%

Get well soon, Dex. This team is a whole lot better with you out there.

5 – Tony Sipp, 0.84%

The high-leverage relief pitcher rears its head again. That being said, it sure is nice to see Sipp in a game; good things have usually happened.

4 – Dallas Keuchel, 0.99%

Now we move into the true core of the current Astros squad. Keuchel has come out of seemingly nowhere, and has managed to overshadow the next individual and seem to be the ace of the staff.

3 – Jarred Cosart, 1.18%

With the emergence of Keuchel and McHugh, Cosart has somehow become the forgotten man of the Astros’ rotation. That’s a shame, because as evidenced by his WPA, he’s done a great job of going out every five days and giving the team a fighting chance to win.

2 – Jose Altuve, 1.33%

/does double take
//checks math
///reconsiders entire premise of article

2? 2?!?!?! How can ANY player be more valuable to the Astros than the man who set a franchise record for hits and steals prior to the All Star Game?

1 – Mr. Chad Qualls, 1.39%

A perfect storm of circumstance:

a. Qualls has been highly effective in his appearances, posting a 1.89 ERA with a 1.08 WHIP
b. Qualls has a front office that believes in the leverage model of relief pitching, encouraging their best pitchers to be used when it matters most. What, you thought they were just indecisive about naming a “closer”?
c. Because Qualls is not the “closer”, Porter has been free to use him when it matters most – for example, the decision to use him against Detroit’s 3-4-5 hitters. (The less said about Jerome Williams’ performance in the 9th, the better)

When you get a good reliever pitching in the highest leverage innings for a bad team, WPA will tell you that he’s the most valuable player on that team – because those wins they’ve had, he’s been able to finish the job, or get them much closer to that end.

As discussed initially, each play has its own WPA, and so we can determine the plays that have had the most impact on the games to date. With that said, I present to you the…

PLAYS OF THE (HALF) YEAR

10. July 4: Mike Trout homers off Tony Sipp in the 9TH – (36.4%)
9. May 10: Delmon Young hits a bases-loaded single in a 4-3 game with 2 out in the 9th off Anthony Bass – (38.5%)
8. April 11: Robinson Chrinos breaks the scoreless deadlock with 2 out in the 12th on a single off Brad Peacock – (39.5%)
7. June 27: Jason Castro’s walk-off homer in the 11th off Hardy – 41.9%
6. April 12: Michael Choice homers in the 9th off Chapman to make it the 5-5 tie – (44.1%)
5. May 21: Albert friggin Pujols homers with 2 out in the 6th to make it 2-1, where it would end – (45.8%)
4. April 19: Alberto Callaspo singles off Chad Qualls to tie the game in the bottom of the 9th – (50.2%)
3. May 10: Mighty Altuve singles home Keuchel(!) and Villar in the 9th to take the lead 4-3 (WPA doesn’t care that they lost!) – 61.8%
2. April 23: Kyle Seager hits a 3-run job, down 2, off Josh Fields with 1 out in the 9th – (67.7%)
1. June 28: Jerome Williams loses the game, and his job, on a 2-out, 3-run homer to Kinsler – (70.8%)

Well, that’s depressing. 8 of the 10 biggest swings have gone against the hometown 9.

Can we at least look at the 8 other plays in FAVOR of the good guys?

8. May 8: Altuve doubles in Corporan and Hoes to take a 3-2 lead in the 5th off Smyly – 26.2%
7. May 9: Carter grounds into a DP with runners on the corners in the 9th to cut the lead to 4-3. Yes, a GIDP in a losing effort makes the Astros’ top 10 plays for the first half. – 26.8%
6. May 10 (again!) Marc Krauss doubles off Hunter to put the lead run on second – 27.5% – note: this play and Altuve’s ensuing single combine for a 89.3% swing in that game – and they still lost!
5. July 12: Qualls gets a game-ending double play from Pedroia as his only batter faced – 28.2%
4. May 14: Dominguez delivers an RBI single with 1 out in the 9th, for a 5-4 win – 29.1%
3. July 9: Springer homers with Altuve aboard in the 7th inning with a 6-4 lead – 30.2% note: check out the difference between a two run and four run lead!
2. April 12: Grossman homers in the 4th with 2 on to take a 5-2 lead – 31.2%
1. May 25: A 6th-inning #SpringerDinger off Iwakuma to take a 2-1 lead – 31.4%

May was a nice month, wasn’t it?