Thursday, October 30, 2014

The Last Strike Zone of the Year

Last night's Game Seven was taut, with the outcome in doubt until literally the last at-bat (the Royals had the tying run at third with two outs in the ninth) and will be remembered for yet another amazing performance by Giants pitcher and Series MVP Madison Bumgarner

Fortunately, it won't be remembered for the umpires, other than the (correctly) overturned call on the acrobatic double play Joe Panik started in the third inning (aided by Eric Hosmer's dumb slide into first base - do sprinters slide headfirst across the finish line?)

Specifically, home plate umpire Jeff Nelson had a pretty good game. Here's a map of his ball-strike calls with a right-handed batter at the plate:

That's from the umpire's point of view, so the batter's on the left side. Called strikes are in red, called balls are in green. The solid line is the rulebook strike zone, while the dashed line is the strike zone as typically called by major league umpires. (They expand the strike zone a bit). There were a three calls that Nelson missed, calling a strike on the Royals that should have been a ball and a couple strikes on the Giants that were balls (one away, one down), but that's OK.

Here's his strike zone for left-handed batters, whom you should imagine standing on the right side of this diagram:

Lefties get sort of hosed by major league umpires, as their strike zone is a good bit wider than it should be, and Nelson called a lot of strikes on pitches that were outside the rulebook zone, though they were inside the zone lefty swingers usually see. So, all in all, a nice job.

Wednesday, October 29, 2014

Reality Check: Game Seven

Tonight's the last game of the 2014 season. Does history tell us who's going to win?

Yesterday (I'm sorry I didn't read it until today), ESPN's Jayson Stark had a piece about how the Royals, as the home team for the last two games of the Series, have an advantage. Specifically, here are two of his datapoints:
Starting in 1982, the home team has gone 22-3 in Games 6 and 7. And yes, you read that right. We said 22-and-3. That's an .880 winning percentage. Which makes no mathematical sense at all. If you subtract those 25 games, the winning percentage of home teams in all other World Series games, including this year, is just .540. Amazing.No home team has lost a Game 7 in 35 years, since the Pirates won their last title in Baltimore. Since then, road teams have lost eight straight Game 7s.
Now, my purpose here isn't to call out Jayson Stark. He's a great journalist. His coverage of the game is passionate and humorous, and he's an old-school print journalist who's not afraid to bring on new concepts. His latest book is on my reading list. And the statistics he cited have the virtue of being, you know, correct.

However, they're also selective. True, the home team, since 1982, is 14-3 (after last night, 15-3) in World Series Game Six and 8-0 in Game Seven. But that's only since 1982. What happened before then?

  • From 1905 (when the World Series became a best-of-seven tournament, excluding 1919-21, when it was best of nine) to 1982, the home team was 27-18 in Game Six.
  • From 1905 to 1982, the home team was 10-18 in Game Seven. Won ten, lost eighteen.
  • So from 1905 to 1982, the home team was an unimpressive 37-36 in Games Six and Seven.
  • Adding that to the Stark's figures, the home team, in all of baseball history, is now 60-39 (.606 winning percentage) in Games Six and Seven, comprised of a 42-21 (.667) record in Game Six and just 18-18 (.500) in Game Seven.  
Did something significant happen in 1982 that changed the way the game's played, making the post-1982 record more relevant to today's game? We'd already gone to the modern bullpen: In 1982, reliever Bruce Sutter got a win and two saves for the champion Cardinals, while Brewers reliever Bob McClure had two losses and two saves. It was the tenth year of the designated hitter, so that wasn't a new thing, though the practice at the time was that the DH used was used throughout the Series in even-numbered years and not at all in odd-numbered years. (The current rule of using the DH only at the American League park began in 1986). It's hard to see that minor DH change making the home team suddenly invincible. Stadiums have gotten a little smaller (shorter fences) and the fields slower (grass replacing artificial turf), but that shouldn't change the home-field advantage. In short, the game didn't really change. So neither should the outcome of the last two games of the Series.

So in looking at tonight's game, I don't think it's any less accurate, nor less predictive, to say "the home team is 18-18 in Game Sevens" than it is that the home team's 9-0 in the past 35 years (8-0 since 1982, and a winner in 1982 as well), the last home team Game Seven loss being the Orioles to the Pirates in 1979.


On another Game Seven topic, a friend asked me, "Rob, I'm looking game 7 starters and thinking to myself 'whatever happened to aces that pitched games 1,4,7?'" My quick answer was, "That went the way of the four-man rotation." Turns out there's more to it than that.

There have been, including this one, 37 World Series that went seven games. That means there have been 74 opportunities for a pitcher to start games 1, 4, and 7. That happened only 14 times, just 19% of the time: Boston's Smoky Joe Wood and the Giants' Jeff Tesreau in 1912 (the Giants' Christy Mathewson started games 2, 5, and 8--the second game was an 11-inning tie); Washington's Walter Johnson in 1925; Brooklyn's Joe Black in 1952; Pittsburgh's Vern Law in 1960; St. Louis's Bob Gibson in 1967 and 1968; the Mets' Jon Matlack and Oakland's Ken Holtzman in 1973; St. Louis's John Tudor in 1985; the Mets' Ron Darling in 1986; Minnesota's Frank Viola in 1987 and Jack Morris in 1991, and Arizona's Curt Schilling in 2001. The golden era was 1967-2001, when there were 13 Series that went seven games, and eight of them had a pitcher go in the first, fourth, and last games, but that's still only 35% of the total (nine pitchers, including two in 1973, divided by 26 teams).

Part of the reason for the infrequency is that the current pattern of two games, off day, three games, off day, two games didn't emerge until 1957. Still, there have been 23 seven-game series since then, and pitchers started 1-4-7 only ten times (22% of possible). Prior to 1957, there were seven pitchers who got three starts in a seven-game series, but they weren't in games 1, 4, and 7. Since then, there have been 18 such pitchers, including three pitchers in each of the 1962 and 1965 World Series and the Cardinals' Chris Carpenter in 2011.

But I was wrong about the four-man rotation. The only pitchers among those to have started at least 39 games in a 162-game season or 37 in a 154-game season--reasonable indicators of a four-man rotation--were Wood and Holtzman. (Black was a reliever who'd started only two games in the regular season when he was called upon to start three games in the 1952 Series.)

So I suppose the answer to the question is that aces pitching games 1, 4, and 7 was never really all that common in the first place.


And one other Game Seven note: Yesterday on SiriusXM, Steve Sax was speculating on whether Madison Bumgarner might pitch in Games Six or Seven after throwing 117 pitches in a complete game shutout on Sunday. Sax said that Bumgarner could, or even start one of the games. After all, said Sax, Babe Ruth, when a pitcher, would sometimes start both games of a doubleheader. 

Babe Ruth started 147 games in his career. On Tuesday, July 11, 1916, he started the first game of a doubleheader against the White Sox. The box score shows that he pitched to one batter and didn't record an out. I assume he was pulled after just a pitch or two. (The story is that the Red Sox starter that day, Rube Foster, wasn't ready for the start of the game, so Ruth threw some pitches while Foster completed his warmup.) Anyway, he started the second game as well, pitching a complete game in a 3-1 Red victory. That's the only game in which he started both games of a doubleheader, and it really doesn't count. 

Tuesday, October 28, 2014

What's Bugging the Statheads About This World Series

Every postseason, it seems, engenders a de rigueur response from someone in the media about how one team or the other or the style of play or something infuriates the stats guys, sabermetricians, Moneyball crowd (my favorite term of disrespect, because it misses the point of the book, which is exploiting market inefficiencies, not statistical analysis), whatever. This year's Series brought out a larger-than-usual number of such columns, given that both teams don't hit a lot of home runs, strike out much, or draw a bunch of walks. The Giants were 17th, 17th, and 21st in the majors in home runs, strikeouts, and walks, respectively, while the Royals were famously last in all three categories, the first such team to make it to the Series. Add the Royals' penchant for stolen bases (153, first in the majors, 11% more than the second-ranked team) and sacrifice bunts to extend the narrative.

Not that any of this is really all that true. I mean, yes, the Royals were last in those categories during the regular season, but, as I pointed out after Kansas City won the ALCS, they made it to Series by hitting an above-average number of homers, walking at a much higher-than-average rate, and striking out about as frequently as everyone else. As for the speed, well, the Royals have attempted only five stolen bases in the ALCS and World Series, and were successful only twice, so the stolen base has been more a detriment than a weapon for them.

But there's one thing that both managers have done that's fair game for criticism from he analytic community: Sticking with their starters too long. The times through the order penalty (TTO) refers to what happens to starting pitchers each successive time they face the opposing team's lineup. The longer a pitcher goes, not only does he get tired, but the hitters get used to his delivery and timing, making it easier to make contact.

In 2014, for example, in batters' first plate appearance against a starting pitcher, they batted .246 with a .304 on base percentage and a .377 slugging percentage. They hit home runs in 2.3% of plate appearances, walked in 7.1%, and struck out in 21.1%.

In the second plate appearance, batters' slash line improved to .256/.313/.395. They homered in 2.3% of plate appearances, walked in 7.0%, struck out in 19.1%.

In the third plate appearance, the slash line was .268/.327/.421. Per plate appearance, batters homered 2.7% of the time, walked 7.4%, and struck out 17.7%.

The pattern is pretty clear: The more times a pitcher faces a batter, the more hits, home runs, and walks he gives up, and the fewer strikeouts he gets. 

This has played out in the Series:

  • In Game Two, the teams were tied at two runs apiece after five. The managers kept their pitchers in, even though both were facing batters for the third time. In the top of the sixth, Royals starter Yordano Ventura allowed two singles and was lifted. Reliever Kelvin Herrera put out the fire. In the bottom of the sixth, Giants starter Jake Peavy allowed a walk and single, was pulled, but both runners scored, and by the time the inning ended, the Giants were losing 7-2. Both starters ran into problems the third time through the order, though only the Giants were burned.
  • In Game Three, the Royals were up 1-0 after five. In the top of the sixth, Royals manager Ned Yost let starting Jeremy Guthrie bat, and he grounded out. Giants starter Tim Hudson then faced Royals shortstop Alcides Escobar for the third time. He singled, left fielder Alex Gordon doubled him home, and a groundout and a pitching change later, first baseman Eric Hosmer drove in Gordon. In the bottom of the inning, Yost's decision to let Guthrie bat came back to bite him, as Guthrie allowed a single and a double and then, facing center fielder Gregor Blanco for the third time, a walk. Again, the Royals bullpen put out the fire, as the Royals won 3-2. Again, both starters ran into problems the third time through the order.
  • Game Four was a blowout, with the Giants winning 11-4, but the Royals had a 4-2 lead going into the fifth inning. Starting the bottom of the fifth, facing Giants second baseman Joe Panik for the third time, Royals starter Jason Vargas gave up a double. Yost relieved him, so there wasn't much damage, though Panik came around to score in an inning which the Giants tied the game. Vargas gave up a third-time-through-the-order hit that at the time was costly.
I'm not including Games 1 and 5, because the Giants jumped to a first-inning 3-0 lead in the first game, and besides, Madison Bumgarner, who started both games for the Giants, has been invincible in the postseason (six starts, 1.13 ERA, 0.67 WHIP). He also has had a minimal TTO penalty this year, with batters compiling a .706 OPS against him in their first plate appearance, .603 in the second, .675 in the third. (The league average OPS in 2014 was .694.)

Tonight, the starters are Jake Peavy for the Giants and Yordano Ventura for the Royals. Here's how batters did against them in their starts this year:

First plate appearance258205912272363.229.290.372.6622.4%24.4%8.0%
Second plate appearance263285913181754.224.276.373.6482.8%18.9%5.9%
Third plate appearance220347117482240.323.387.545.9333.1%15.7%8.6%
Provided by View Original Table
Generated 10/28/2014.

First plate appearance245115210312054.
Second plate appearance241336013182657.249.326.411.7373.0%21.1%9.6%
Third plate appearance18118485142042.265.340.370.7102.0%20.5%9.8%
Generated 10/28/2014.

Peavy has been pretty good through the first couple times through the order but he's collapsed the the third time through. That .933 OPS that he's allowed--well, Mike Trout, who will be the AL MVP, had a .939 OPS this year. Giants manager Bruce Bochy really shouldn't let him pitch after he's gone through the Royals lineup twice.

As for Ventura, he's been nails the first time through the order, but below average thereafter: the .725 OPS he's allowed the second and third time through the order compares to an AL average of .706.

Presumably, both pitchers will be on short leashes tonight, with the Giants one win from the championship and the Royals one loss from elimination. But if we see, say, Peavy come out in the sixth to face the Royals lineup a third time, clinging to a one- or two-run lead, we could see more sixth-inning fireworks (12 of the 42 runs scored this Series, nearly 30%, have come in the sixth inning, during which starters are typically facing the opponents' batters a third time), courtesy of a failure to understand the times through the order penalty.

Monday, October 27, 2014

Oscar Taveras, 1992-2014

You have probably heard that Oscar Taveras, St. Louis's talented outfield prospect (ranked third in baseball by Baseball Prospectus, Baseball America, and before the season) died in a car accident in the Dominican Republic yesterday. He was only 22.

Two things about Taveras. First, his rookie season, listed below, was viewed as somewhat of a disappointment. The column OPS+ measures his on base plus slugging relative to the league average, adjusted for his home park. His score of 65 means that he was 35% worse than the league average, which of course isn't very good. But his manager, Mike Matheny, seemed unwilling to commit to him. Taveras was called up at the end of May and started nine games in June, 15 in July, 24 in August, and 12 in September. The Cardinals were 36-25 in his starts, a .590 winning percentage, compared to .534 in games he didn't start. His minor league statistics--.320/.376/.516 slash line in 436 games spread over six seasons, including .313/.358/.485 at AAA Memphis--portended a strong, middle-of-the-order bat.

80 248 234 18 56 8 0 3 22 0 1 12 37 .239 .278 .312 .590 65 73 10 1
Generated 10/27/2014.

Second, he died in an auto accident. So did 35,303 Americans in 2011 (the last year for which data are available). That's almost as many people who died of breast cancer (41,374). It's 26% more than died of prostate cancer, more than twice as many who died of homicide, 4.6 times as many who died of HIV. It's 9% more than people who died from firearms (the majority of which, 19,990 of 32,351, were suicides). You're about 40 times more likely to die in a car accident than a boat or airplane accident. Just a reminder that cars are dangerous. Drive defensively, drive under control, always wear seat belts and shoulder harnesses, and catch a cab if you're at all impaired. 

Sunday, October 26, 2014

The Unassailable Wisdom of Dodger Fans

(Programming note: This post ran yesterday at FanGraphs Community. You can read it here. The format there is slightly more reader-friendly there than it is here.)

Another exit from the postseason deprived the nation of tales of Dodger fandom and their proclivities–-Dodger Dogs, Vin Scully, and, of course, leaving the game early. Why they leave early, beats me. Maybe they have premieres to attend. Maybe they’re going to foam parties. Maybe they’re trying to beat the traffic. Me, I don’t know. Like most FanGraphs readers, I’d guess, I have never been invited to a premiere. Or, for that matter, a foam party. (And I’m still not entirely clear as to what one is.) As for beating the traffic, yeah, I get it, average attendance at Dodger Stadium was 46,696 this year, highest in the majors, so I imagine that’s a lot of cars. But Dodger games took an average of 3:14 last year, which means that night games ended well after 10 PM, so one would assume that traffic on the 5 and the 10 and the 101 and the 110 would have eased by then, though I don’t live in a part of the country in which highways are referred to with articles, so what do I know.

Aesthetically, of course, the argument against leaving a game early is that you might miss something exciting–an amazing defensive play, a dramatic rally, last call for beer. That would seem to trump the concerns of early departers.

Especially a rally. A late-innings comeback is one of the most thrilling pleasures of baseball. But that made me wonder: Are they becoming less common? If so, wouldn't that be an excuse, if not a reason, for leaving early?

During the postseason, you may have heard that the Royals have a pretty good bullpen. (It’s come up a couple times on the broadcasts.*) With Kelvin Herrera often pitching the seventh, Wade Davis the eighth, and Greg Holland the ninth, the Royals were 65-4 in games they led after six innings. Of course, a raw number like that requires context, so here is a list of won-lost percentage by teams leading after six innings:

TeamWL Pct.
Red Sox52985.2%
Blue Jays611184.7%
White Sox511083.6%

Sure enough, the Royals did very well. The major league average was 87.7%. Kansas City, at 94.2%, easily eclipsed it. But, as you can see, so did the Dodgers. We certainly didn’t hear about their lockdown bullpen in their divisional series loss to the Cardinals. Presumably, the Dodger bullpen’s 6.48 ERA and 1.68 WHIP over the four games of the series had something to do with that. But during the regular season, the Dodgers held their leads.

How about the other way: What teams were the best at comebacks? Shame on Dodger fans if they were leaving the parking lot just as the home team was launching a rally, turning a deficit into victory. Here’s the won-lost record of teams that were trailing after six innings:
TeamWL Pct.
White Sox106413.5%
Blue Jays86011.8%
Red Sox87210.0%

Whoa. Ignoring for now the late-inning heroics of the Nationals, who were able to come from behind to win over one of every five games that they trailed after six innings, look who’s at the bottom of the list! The Dodgers trailed 56 games going into the seventh inning this year, and won only two.

So maybe the Dodger fans who left games early are on to something. I devised a Forgone Conclusion Index (FCI) by combining the two tables above. It is simply the percentage of games in which a team leading after six innings wins the game. For example, the Royals led after six innings 69 times and, by coincidence, trailed after six innings an equal number of times. Their Forgone Conclusion Index is 65 Royals wins when leading after six plus 58 opponents’ wins when the Royals trailed after six, divided by 138 (69 plus 69) games in which a team led after six innings. The Royals’ FCI is thus (65 + 58) / 138 = 89.1%. The team leading Royals games going into the seventh inning wound up winning just over 89% of the time. A Royals fan wishing to leave a game after six innings did so with 89% certainty that the team in the lead would go on to win. (Yes, I know, I should do a home/road breakdown, but this is a silly statistic anyway.)

Here’s the Foregone Conclusion Index for each team last year.

Padres92.9%Indians88.1%Blue Jays86.4%
Reds90.1%Red Sox87.9%Angels85.5%
Rays90.1%Yankees87.6%White Sox85.2%

And there you have it. The Dodger patrons leaving the game early weren’t being fair-weather or easily-distracted fans. Rather, they were simply exhibiting rational behavior. They follow the team for which the team leading after six innings was the most likely in the majors to hold on to win. They were the least likely fans to deprive themselves of the excitement of a late-inning comeback by leaving early.

I know what you’re thinking: Single-season fluke. There have to have been more comebacks in Dodger games in recent years, right? As it turns out, yes, but not a lot. The Dodgers were eighth in the majors in Foregone Conclusion Index in 2013 (87.8%) and seventh in 2012 (90.4%). Maybe 2014 is an outlier in which there were an extremely small number of comebacks in their games, but over the 2012-2014 timeframe, only the Braves (91.8% FCI) and Padres (90.9%) have played a higher proportion of games in which the team leading entering the seventh inning has gone on to win than the Dodgers (90.8%).

So keep it up, Dodger fans. Get into your cars during the seventh inning, turn on Charlie Steiner and Rick Monday on the radio, and drive on your incrementally less crowded highways on the way to your premieres and foam parties. You probably won’t be missing a comeback, and by leaving early, you’re expressing your deep understanding of probabilities.

*TBS managed to botch a fun fact about Kansas City’s bullpen. At one point, they posted a graphic stating that the Royals are the first team to have three pitchers-–the aforementioned Herrera, Davis, and Holland-–to compile ERAs below 1.50 in 60 or more innings pitched. They forgot the key qualifier: Since Oklahoma became a state. The 1907 Chicago Cubs featured three starters with ERAs below 1.50: Three-Finger Brown (1.39), Carl Lundgren (1.17), and Jack Pfiester (1.15). The Cubs’ team ERA was 1.73.