May 21, 2002
Box Work: Rally Killers
We experience baseball through its box scores more than through any other medium. A box score is an analogue, a concise recreation of everything we need to know, a tidy answer to Lenin's question: Who is doing what to whom?
Didn't most of us develop our understanding of the game by reading box scores? Before the Internet, "Baseball Tonight," and Extra Innings, our access to the game outside our town came through the morning paper. Did we read for data or drama? A lively imagination could dramatize the data so that the box scores really were the next best thing to being there.
A good box score answers countless questions. Was it windy, was it wet, how many people were there? Did the wind blow in, did it blow right to left, did it cause home runs? Was it a day game? Who was the home plate umpire? Did he squeeze the strike zone? How many fans showed up, and when were they allowed to leave?
The box score tells us how many batters a pitcher faced, how many pitches he threw, and whether he pitched ahead or behind the hitters. Do mid-inning relief changes make games longer? Are pitchers more effective when they work faster? Does this manager use his relievers as specialists, while that manager lets his stretch their arms? Who abuses their young starters?
From a box score, we can learn if a team loses because it has poor fundamentals. Does it commit too many errors, passed balls, and wild pitches? Does it take needless risks on the bases? Can it take a pitch or throw a strike? Do its pitchers get rattled after two-out errors?
A good box score tells us which pinch hitters were used, when they were used, and whether the manager has a priority off the bench. Does he use platoons? Does he have a designated starting catcher against right-handed power pitchers? Does he let players work through slumps? Does he favor veterans?
The box score gives us the data we need to play at being manager, general manager, and in-house sabermetrician.
Baseball is reactionary, but it has progressed beyond the confines of the daily paper. Through the Internet, we can see box scores not merely the next day, but even as they develop during play. With a keystroke we can see the recaps, game notes, game logs, line scores, pitch-by-pitch accounts, and up-to-the-minute statistics. Schedule results, winning streaks, records in one-run games. Lefty/righty, day/night, inning-by-inning splits. There's nothing recordable that we can't find easily.
Sabermetrics and the information explosion have not revolutionized the game. The game has not changed. We just see more of it now, and have a better understanding of its causes and effects. Information technology is a tool that gives us instantaneous access to just about everything that stimulates our curiosity. Technology didn't revolutionize box scores so much as it allowed us to reorient our attention back to what mattered before the advent of fantasy baseball.
The goal of this piece is to find provocative items from the box scores, stuff we might miss when we're in a hurry or when we're merely trying to maintain our roto rosters. Its emphasis will be on the causes and effects of actual baseball.
From the boxscores of April 23:
Losing to Detroit 3-0, Michael Tucker led off the Royals' fourth inning with a double and was thrown out at third trying to get himself a triple. Baseball's "book" of common sense says that it's stupid to make the first out at third base. From third base, the runner can score on a passed ball, wild pitch, ground out, error, or balk, but it usually takes a base hit to get him home. With no outs, the slim scoring advantage to be gained isn't worth the risk of taking your runner off the bases altogether.
In The Hidden Game of Baseball, Pete Palmer and John Thorn did a study of potential runs for two dozen base-out situations. They analyzed a team's potential for run scoring, given the number of baserunners relative to outs in an inning. The study was published in the early 1980s, so it's not the freshest research, but it was based on more than 75 years' worth of data. Using the years 1961 to 1977 to represent the modern age, this is the probability chart they generated:
0 1 2
None .454 .249 .095
1st .783 .478 .209
2nd 1.068 .699 .348
3rd 1.277 .897 .382
1st, 2nd 1.380 .888 .457
1st, 3rd 1.639 1.088 .494
2nd, 3rd 1.946 1.371 .661
full 2.254 1.546 .798
With no outs and no one on base, a team would be expected to score a half run in the inning. We can apply this to Tucker's attempt to stretch his double into a triple. With no outs and a runner on second, the Royals would be expected to score 1.1 runs; with no outs and a runner on third they would be expected to score 1.3 runs. There's a slight advantage to having Tucker on third rather than second to start the inning. But what's the downside risk of taking that extra base? With one out and no one base the team would be expected to score .25 runs, a drastically lower likelihood of scoring. The potential gain is not worth the risk. The data corroborates common sense.
Later in the game, the Royals were still down 3-0 with the top of their lineup due. Chuck Knoblauch hit a leadoff single, but tried to stretch it into a double and got thrown out. If he had made it to second, he'd have improved the Royals' scoring probability by .29 runs. By getting himself thrown out, he reduced their likelihood by .53 runs. Again, the risk wasn't worth the reward. And Knoblauch had been turning his walks and singles into doubles: at the time he reached first base, Knoblauch was eight-for-eight in stolen base attempts. On paper, the gamble was reckless, especially for a team in need of base runners. The Tigers retired eight of the next nine Royals, and Kansas City lost 3-0.
Going into their April 23 game at Tampa, the Twins had lost seven straight games at Tropicana Field. Trailing 3-1 in the third inning, the Twins rallied. With no outs, Doug Mientkiewicz hit a two-run double. With one out and Mientkiewitz on second, Corey Koskiesingled to center, but was thrown out advancing to second. Mientkiewicz stopped at third on the play. If Koskie had been safe, the Twins would have increased their run scoring potential by .28 runs. Because he got caught, their probability dropped by .71 runs. Mientkiewicz was marooned on third when Brian Buchanan struck out swinging to end the inning.
In the seventh inning, the Twins ran themselves out of another rally. They were down 6-3 when Dustan Mohr led off with a double. He scored on A.J. Pierzynski's single to left, making the score 6-4 with no outs and the lineup about to turn over, but Pierzynski was thrown out trying to stretch his single into a double. The Twins didn't score again until the next day.
On April 22, Milwaukee lost 5-4 to Montreal, dropping the Brewers' record in one-run games to 1-7. Making the first out at third base is dumb, but book says that making the first out at home is the cardinal sin of baserunning. Eric Young led off this game with a triple to right field, but got nailed at home trying to get an inside-the-park homer off of Vladimir Guerrero, who may have the best arm in baseball. If Young had stayed at third base, the Brewers would have been likely to score 1.3 runs. With nobody on base and one out, they were likely to score .25 runs.
In the second inning, with two out and two runs already in, Raul Casanova was thrown out at home trying to score from second--again testing Guerrero's arm. With two outs, a runner on second will almost always be sent home since he's running on contact. If Casanova had stayed at third the Brewers would have been likely to score .45 runs from having runners on first and third. So if a runner has an even chance of being safe at home, the risk of sending him is arguably worth it, which is why runners--even those as slow as Casanova going up against an arm like Guerrero's--are rarely held at third in this situation.
With his team down 5-4 in the eighth, Jeffrey Hammonds got a one-out, bases empty single. He tried to stretch it into a double and got himself killed and the Brewers went on to lose 5-4. Montreal outfielders made three errors in the game, but the Brewers gave them three base running outs in trade.
We usually think of a team's record in one-run games as random, a matter of chance more than skill. Over at Diamond Mind, Tom Tippett recently published an article by Bill James analyzing the causes of one-run wins and losses. Following up on research done by Rany Jazayerli, Joe Posnanski, and Tom Ruane, James concluded that while one-run wins are likely a matter of luck, one-run losses are to some degree earned and have some tendency to carry over from one season to the next.
Last season, the Brewers had the National league's worst record in one-run games. At 2-14 this season, they again are arguably the worst (Colorado is 1-9). They were 1-6 in one-run games under Davey Lopes. On the day of Lopes's firing, Dean Taylor said, "This is Opening Day II." Under Jerry Royster the team is 1-8 in one-run games. We're due for Opening Day III.