Baseball Toaster was unplugged on February 4, 2009.
Well, almost everything went according to plan in the first round. The Mets swept the Dodgers just as I predicted. Oakland and St. Louis won though in one fewer game than I predicted. However, I was way off on the Yankees-Tigers series, and I am not the only one.
When the series started, the broadcast crew was hailing the Yankees as a modern "Murderers' Row and Cano" (as McCarver dubbed them). Meanwhile, the Tigers were reeling as the playoffs approached, losing five straightnot to mention the division on the last day of the seasonand being swept in their final series by the execrable Royals.
The series opened as expected, with an 8-4 Yankee win, but after rain delayed game two the next night, the Yankees never seemed to be in the final three games. After going ahead 3-1 in the fourth inning of the rescheduled game two, the Yankees went the next 20-2/3 innings without scoring a run. Their next run came when they were already down two games to one and were losing 8-0 in game four, with a little under three innings remaining.
So what happened?
Obviously, the Yankees offense was the most salient issue. Of course, not scoring runs in the playoffs will end with the other team drinking the bubbly, but anyone who watched the Yankees this season, or over the last couple of seasons, knows that their problems have been pitching. Indeed it's been the problem at least since the ill-advised signings of Carl Pavano and Jaret Wrightwho lost game four? And their middle relief has been worsening for years, from a great asset about a decade ago to an Achilles heel today. Pitching should be their focus in the offseason.
Some will say that the Yankees were doomed because of the adage that pitching always beats hitting. The great Yankee offense was held at bay by the Tiger staff while the sub-par Yankee pitching staff could not overcome the Detroit offense. That's what conventional wisdom would say.
But I have to ask whether we actually know this to be true, as in can it be tested by the historical record. So I tried
For each playoff series, I compared the winning percentage of the winner with a variety of statistics to determine which had he biggest connection to winning. I looked at the regular season winning percentage of both teams as well as the expected winning percentage based on runs scored and allowed. I took each team's ERA to test the pitching and the batting average, on-base percentage, slugging average, and OPS (on-base plus slugging) to test offense. For each, I took the ratio of the playoff series winner's value to the loser's.
First, let me say that none of these values correlate to postseason success very well, which is blanket indictment for any theory generalizing why some teams beat other teams in the postseason. However, looking at how well these various statistics correspond to postseason success, I think the adage about pitching beating hitting is on less sure footing than almost any other one can dream up.
The stats that correlated best to postseason success were, not surprisingly, regular-season winning percentage (coefficient of .280) and expected winning percentage (.242).
Actually, OBP is slightly ahead of expected winning percentage (for now), but the other stats aren't even close:
ERA:PCT |
.069 |
BA:PCT |
.062 |
OBP:PCT |
.251 |
SLUG:PCT |
.113 |
OPS:PCT |
.180 |
From these results, ERA, our representative for pitching prowess, correlates worse to postseason success than anything except batting average, which it beats out by a hair. In this series, though, the Yanks came in with a major-league leading .363 OBP while Detroit was 24th in the majors at .329. Keep in mind that Detroit's ERA (3.84) led the majors while New York was 12th (4.41). There must be something more to this correlation.
Could it be that taking the stats at face value misrepresents their actual meaning? Remember that for decades the postseason consisted solely of a World Series between two leagues that had totally separate sets of data. The stats would then be taken out of context.
I reran the numbers with the stats adjusted for the league average. The results weren't much better:
ERA:PCT |
.069 |
BA:PCT |
.016 |
OBP:PCT |
.198 |
SLUG:PCT |
.059 |
OPS:PCT |
.122 |
The offensive stats did a little worse, but OBP was still the victor. However, it became clearer that regular season winning percentage and expected percentage were by far the best means to predict a series winner.
I think the Yankees' woes had more to do with the shortcomings of the playoff system itself, and that's something I will be exploring in a piece with Baseball Prospectus.
Any indication how K\9 or K\BB correlates?
But they lost the series because they let that A-hole Kenny Rogers make them look like a bunch of amatuer chumps.
You lost me here, "Other Baseball Toasters have been talking about fastball speed as coming from some book. . ." Huh?
I can run WHIP, K/9IP, K:BB, etc.
redking,
I can do WHIP, but unless you can get MLB to change the historical record, I don't have enough data GB/FB.
Schteeve,
Yep, but whaddya gonna do? The Yankees reacted in game five by swinging at the first pitch--I think Bonderman had thrown something ludicrous like 35 pitches through 5--and dug even deeper.
I wonder what post season starters' ERA (or whatever is your favorite metric) is for the regular season vs. the post season. Does it go down - even though they theoretically are facing better hitters; stay the same; or go up - due to better hitters.
1. A power pitching staff, as measured by normalized strikeout rate (EqK9), which accounts for park and league differences
2. A good closer, as measured by Reliever Expected Wins Added (WXRL), which accounts for the degree of difficulty (runners inherited and run margin) of a reliever's appearance
3. A good defense, as measured by Fielding Runs Above Average (FRAA)
1. I don't think the W-L% of series winners is a very useful outcome measure. Do we really want to analyze the differences between teams that win 4-1 and teams that win 4-3? Or between teams that win and teams that lose?
2. What happened to statistical significance? I can't run the numbers, but I'd be surprised if the highest correlation you report - .28 - is significant. So unless I'm missing something, you're essentially analyzing random variation rather than any meaningful differences.
I'm no expert, so I'm more than willing to be set straight.
Comment status: comments have been closed. Baseball Toaster is now out of business.