In selecting a game to review this week, I had a hard time choosing between two Texas/Seattle games. Friday night’s game was a wild one, featuring six lead changes in an eventual 9-6 Mariner win. I opted to cover Saturday’s game however, which the Rangers won 7-6. I hope Mariner fans forgive me.
I’ll be referring to Win Probability Added, or WPA throughout the article. Click on the link if you’re not sure what that is.
This game pitted 41-year-old Jamie Moyer against the youngster Pedro Astacio (just 34 years old). It was Astacio’s first Major League appearance since last May. Both starters pitched well in the early innings, with Moyer doing his crazy change-up thing and Astacio looking sharp.
WPA is a funny thing. After Alfonso Soriano led off the game with a double and Moyer walked Hank Blalock, the Rangers had gone from a 50% Win Probability (WP) to 59%, and Moyer’s WPA total already looked bad. However, Moyer got Michael Young to foul out and he struck out the next two batters, so his WPA at the end of the half-inning was .047.
In fact, no one scored in the first three innings. At that point, Astacio had a WPA of 0.145, having pitched a perfect three, while Moyer’s was 0.116. A nice pitching duel was evolving, though neither pitcher would likely last longer than 100 pitches.
In the fourth, the Rangers scored two runs on three hits, with all hits coming after two outs. Because there were two outs, the first two hits didn’t add much to the Rangers’ WP (.05). That’s because the probability of scoring runs with two outs, even with two men on base, is relatively low. But Sandy Alomar’s two-run single beat the odds, gave him a .205 WPA and the Rangers a WP of .716.
Ichiro led off the bottom of the fourth with a double, WPA of .077. Leadoff hitters have larger WPA opportunities because leadoff men who reach base are more likely to score than batters who reach base with one or two outs. Jeremy Reed next contributed a “productive out” by batting Ichiro over to third on a groundout. The trouble with productive outs, of course, is that they actually lower the team’s chance of winning. On that play, for instance, Reed’s WPA was a negative .022.
However, Adrian Beltre batted in Ichiro with a single to make the score 2-1, for a WPA of .069. Notice that Ichiro’s double was actually the bigger contribution to the probability of a win (.077 to .069), even though Beltre was credited with the RBI. That’s one of the powerful insights you can get from WPA.
In the top of the sixth, Alomar singled in another run (does he own Moyer or what?), for a 3-1 Ranger lead. At that stage of the game, his WPA was .354, and Astacio’s was .157. Given that a total WPA of .500 equals a win, those two guys had won a game all by themselves already. But things change.
Shigetoshi Hasegawa came in with one out and two on to stop the Rangers from increasing their lead beyond 3-1. He racked up a WPA of .061 in 2/3 of an inning, placing him second on the team behind Beltre at that point. At the end of the sixth, the Rangers led 3-1 and had a Win Probability of .824.
Beginning of the End
The Mariners were not about to concede, however. In the bottom of the seventh, Randy Winn knocked in a runner (WPA of .133) to make the score 3-2 and lower the Rangers’ WP to .656. Astacio then struck out Scott Spiezio with runners on first and second and two out, which raised Texas’s WP to .752 and gave Astacio a WPA of .096. It was his biggest play of the day.
Top of the eighth, and Hasegawa was ably assisted by his defense. Kevin Mench drew a leadoff walk but was thrown out by Dan Wilson when trying to steal second. Then both Wilson Valdez and Beltre made nice plays on groundballs to finish the inning. One of the key things I try to do when logging a game is divide proper credit between pitching and fielding. In this case, the inning was a 0.051 pickup by the Mariners, and the fielders accounted for a .066 WPA themselves, while Hasegawa’s WPA was -.015 due to the walk and less credit on the fielding plays.
Then the bottom of the eighth. Valdez and Ichiro singled with none out and Jeremy Reed came to bat for what would be the most important at bat of the game. We use something called “P” to measure the criticality of a situation, as discussed in the Closer article and the THT Bullpen Book. The P of this situation was .392, which is a very high score and would in fact be the highest of the entire game. In other words, what happened next occurred at the most critical point of the game, according to P.
Reed sacrificed, but Brian Shouse (who had relieved Astacio) threw the ball away. Valdez scored, and Reed and Ichiro wound up on second and third, respectively. Score tied 3-3 and, with none out, the Mariners’ WP was .855. That single play was worth .317 WPA.
My conundrum was… who should get credit? First of all, this was one of those cases in which the sacrifice was the correct call. A successful one would have raised the Mariners’ WP slightly. But Reed didn’t really do anything positive himself to further the Mariners’ WP; he was the beneficiary of Shouse’s error. In the end, I choose to give the total WPA credit to “None” of the Mariners, and the debit to Shouse of the Rangers.
Anyway, the Mariners scored three more runs, and finished the inning with a 6-3 lead and a .968 WP. At that point, “None” led the Mariners in WPA, with Beltre second at .223 and Ichiro third at .156. For the Rangers, Shouse had a negative .552.
With one out in the top of the ninth, Bret Boone (who had made a several fine plays earlier) made an error that allowed Soriano to reach first. Not a biggie: .021. Next up, Blalock hit a home run to make it 6-5, Mariners. However, one out and one single later, the Mariners still had a .924 Win Probability. Even Blalock’s home run hadn’t made a significant dent in the Mariners’ chances of winning. Things seemed well in hand with one out to go, despite this ninth-inning lapse of Eddie Guardado’s.
How do you measure the potential impact of a situation? With a runner on first and a one-run lead, the situation was obviously volatile. But how volatile? How can we measure such a thing?
I’ve already mentioned P, which is defined as the difference between the WP of the current situation and the WP if the pitcher logs outs to the end of the inning with no runs scoring. For instance, if Guardado gets the out, he’ll have an impact of .076, or 1.000 minus .924.
On the other hand, if Richard Hidalgo hits a home run in this situation, he will change the Rangers’ WPA from .076 to .812. Seems to me that this makes it a pretty critical situation as well. Tangotiger’s Leveraged Index is actually based on this kind of approach, and it probably measures impact better than P. The point was emphatically driven home by Hidalgo’s dramatic homer over the left-center fence, giving the Rangers an improbable come-from-behind lead of 7-6.
Francisco Cordero entered the game in the bottom of the ninth and got two quick outs before the Mariners put runners on first and third with two out. Their WPA at this point was .194, but an extra-base hit would win the game for them. Another critical situation, and their man was at bat: Adrian Beltre. But he lofted an easy fly to center. End of game. A remarkable game, really. Check out the wild, late swings in Win Probability:
Here is a list of the cumulative WPA of all players, including breakouts of WPA by offense, pitching and fielding:
Rangers Off Pitch Field WPA Hidalgo 0.636 0.000 0.010 0.646 Alomar 0.336 0.000 0.000 0.336 Cordero 0.000 0.196 0.000 0.196 Astacio 0.000 0.194 0.000 0.194 Young 0.066 0.000 -0.013 0.053 Matthews 0.038 0.000 0.000 0.038 Allen 0.028 0.000 0.000 0.028 None 0.021 0.000 0.000 0.021 Blalock -0.004 0.000 0.000 -0.004 Soriano -0.024 0.000 -0.003 -0.028 Mench -0.098 0.000 0.000 -0.098 Brocail 0.000 -0.102 0.000 -0.102 Teixeira -0.230 0.000 0.001 -0.229 Shouse 0.000 -0.552 0.000 -0.552 Total 0.769 -0.264 -0.004 0.501 Mariners Off Pitch Field WPA None 0.317 0.000 0.000 0.317 Ichiro 0.202 0.000 0.004 0.206 Valdez 0.127 0.000 0.002 0.129 Winn 0.088 0.000 0.000 0.088 Hasegawa 0.000 0.081 0.000 0.081 Reed 0.029 0.000 0.000 0.029 Beltre 0.016 0.000 0.012 0.029 Ibanez 0.004 0.000 0.000 0.004 Boone -0.044 0.000 0.016 -0.028 Wilson -0.086 0.000 0.056 -0.030 Olivo -0.052 0.000 0.000 -0.052 Dobbs -0.066 0.000 0.000 -0.066 Spiezio -0.096 0.000 0.000 -0.096 Sexson -0.171 0.000 0.004 -0.167 Moyer 0.000 -0.189 0.000 -0.189 Guardado 0.000 -0.755 0.000 -0.755 Total 0.268 -0.863 0.094 -0.501
I’ll check in next week with another game.
References & Resources
If you’re interested in WPA, I will be tracking the WPA of all Reds’ games this year and posting the results at the RedsZone Forum. Also, Baseball Graphs has a WPA Central blog up and running for those of you who are interested n the nasty details of Win Probability Added.