Taking the long roadby Shane Tourtellotte
September 12, 2012
We are close to the start of another great experiment in baseball. Major league baseball has had Wild Cards, themselves an innovation, for nearly 20 years. From 1995 to 2011 the Wild Card playoff teams had the same path toward the World Series as the division champions, through the League Division and League Championship Series. Now, with two Wild Cards per league, those teams must play a single winner-take-all game for the privilege of advancing to the LDS.
This change has sat well with many fans, perhaps surprisingly so with the more traditionalist set among them. Despite adding another layer of postseason play to a sport that once went straight from the regular season into the World Series, the assumption is that it favors division winners. The Wild Card survivor, having spent another day playing and having expended its best available starting pitcher to win that game, is expected to be at a disadvantage against its well-rested division-winning foe and its hand-picked starting pitcher. The Wild Card survivor would have a diminished chance of advancing to the LCS and World Series, which some fans think fitting for a second (or third!) place club.
The theory looks logical and sound on the surface. As to the practice, we're going to have to see how the Wild Card knockout survivors fare over the years before we have even a partial picture. Unless, that is, we can draw some lessons from earlier playoff history.
Over the decades, we've had a lot of baseball teams put at a disadvantage going into a postseason series. Their Division or Championship Series went longer than that of their upcoming opponent; a tie forced them into a playoff game, or two or three, while their future foes rested and set up their rotations. The exact circumstances may not line up perfectly, but the general ones do.
This article will look for some lessons in the experiences of teams that have gone into a postseason series with that kind of tilted playing field.
On a tilted field
There are two ways we can define a playoff team's disadvantage. One is in having to play more games than its future playoff foe before the series begins. The other is in having fewer days of rest before the start of the series. For our purposes, meeting either definition qualifies a team for the pool, although the great majority of qualifiers (59 of 76) fall into both categories. I searched all the "playoff era," from 1969 on, as well as earlier years with a league playoff, for eligible squads.
Our roll of disadvantaged teams goes all the way back to 1946 and the first National League playoff. The St. Louis Cardinals swept out the Brooklyn Dodgers in two straight to win their best-of-three series, and went on to defeat a long-rested Boston in the World Series. The most recent teams are both from last year, and both had to face the Texas Rangers at a disadvantage. Detroit gave away one extra game and two days rest to the Rangers, and fell four games to two in the ALCS. Then St. Louis, due to a staggered LCS schedule, faced Texas in the World Series one day down in rest despite also having won its LCS in six. The Cardinals matched their 1946 forebears, overcoming the deficit to win it all in seven.
In total, 76 teams went into a postseason series so handicapped. Given our assumptions, their overall performance is surprisingly strong. They won 37 of the series and lost 39. In individual games played, they come out even stronger, with a 212-211 record. Going just by those numbers, one can say there is no apparent pattern, no apparent cost to taking the longer route into a postseason series.
This assumes, though, that the long-route and short-route teams would have been evenly matched to begin with, and this is not the case. I tallied up the regular-season winning percentages of both sets of teams. (These figures excluded postseason play, along with the playoff games that MLB now calls "tiebreakers" that get counted in regular season standings.) The long-road teams come out to .5826, and the short-road teams to .6003.
Using the Log-5 method invented by Bill James, I calculated the expected results between teams of these winning percentages. The long-road teams should play at .4817 against their luckier foes, but in real life it came out at .5012. Their 212 wins is more than eight games ahead of the expected number. Eight-plus out of 423 is not massive, but considering that we (or at least I) anticipated they would underperform the number due to the disadvantage of extra play and shorter rest, it looms that much larger.
(As a check, I also did the expected record calculations year-by-year. The total results differed by about half a win, narrowing the over-performance slightly but not substantially.)
I will hold off on making a blanket declaration for the time being, as I break the numbers down a little further, looking for the elusive pattern that might tell us something more.
Deep in the weeds
2007 was baseball magic in Colorado. A tremendous late-September hot streak got the Colorado Rockies into a Wild Card tiebreaker with San Diego, which they won in dramatic extra-inning fashion. They then tore up the National League playoffs, sweeping Philadelphia and Arizona. As the Red Sox and Indians fought down to the wire in the ALCS, Colorado had the gift of an amazing eight days rest before the World Series began. With that plus a spent Boston to face, Rockies fans could taste the champagne.
Instead, what they got for this early Christmas was coal in the stocking: a four-game sweep out of the Series.
Anecdotal evidence holds precious little weight in statistics, but some things can't avoid drawing the eye. The six-day difference in rest days between Colorado and Boston is the highest ever for a postseason series, but it has a bit of company. This table shows the records of long-route teams based on how many more days off their opponents had.
DayOffDifference 0 1 2 3 4 5 6 SeriesRecord 1-3 21-20 8-8 2-5 3-3 1-0 1-0 GameRecord 12-15 117-113 41-41 16-23 18-19 4-0 4-0
No real pattern emerges except for the final two columns. The five-game difference was for the 2006 World Series, when the Tigers got a long time to prepare for the 83-78 Cardinals, and still got routed in five. Eight games is an awfully tiny sample, but a combined 8-0 record is a pretty strong result. If you want evidence that baseball teams can go stale waiting to play games, you've got some here, if not anything decisive.
The trouble is, the pattern doesn't quite carry over when just tracking days off for the short-road team. (Records, as in all these tables, are for the long-road team.)
ShortRoadDaysOff 1 2 3 4 5 6 7 8 SeriesRecord 3-2 7-10 13-14 6-5 4-5 3-3 0 1-0 GameRecord 14-11 44-53 74-76 34-29 25-25 17-17 0 4-0
You can manipulate the numbers a bit and get a good record for the long-road teams when their opponents have four or more days off (14-13 series; 80-71 games), but manipulation is all it really is. The numbers aren't that much better than the overall ones, so without the eye-catching perfect record, all you have is a small sample.
Much the same thing happens with days off for the long-road team—but there may be a little something once you combine the two. First, days off for the long-route team:
LongRoadDaysOff 0 1 2 3 4 SeriesRecord 4-5 13-15 14-11 5-5 1-3 GameRecord 21-24 75-80 77-67 29-26 10-14
The record's a little low when you get thrown promptly into the next series, but not in a statistically significant way. A long layoff doesn't look good either, but that sample is hopelessly small. The most interesting thing I found, and this may be surprising, was the record when the long-road team has two days off before its next series. Yes, it's the best series and game record on the table, but does it really say all that much?
Possibly, if you combine the short-road team's numbers. When the short-route team has two days off, that's when the long-route team has its worst performance, the lowest series and game records on that table. There is a hint here—and with this sample size, a hint is probably all it can be—that the optimum rest time for a baseball team is two days.
My apologies in advance if this gets written into the next CBA and we get cut down to a 60-game season.
We've been looking at the days off, so let's give some equal time to the added games the long-road teams have to play. Again, the sample sizes mean that when we get a bit of an interesting number, it's only a bit of one:
AddedGames 0 1 2 3 Series Record 4-9 20-17 10-11 3-2 GameRecord 30-46 107-95 58-61 17-9
Interesting that the long-route teams suffer most when they don't play any extra games, just getting fewer days off because of staggered schedules. Even with all our small sample sizes, it's the only time the long-road teams end up below .400 for a subset. Tough to believe, though, that one fewer day of rest (and it is one day in all 13 cases) could cause such a shortfall. I'd say this is random chance rather than an underlying pattern.
It's another oddity that having to play three extra games did not hurt the long-roaders. Two of those series we've seen before: the Cardinals and Red Sox sweeping teams that spent a week or more idled. The third was the 1988 Dodgers, so I'll just name-drop Kirk Gibson and we'll take that miracle as a given.
The other two instances, the two losses, are curiously intertwined in baseball history. It was 1951 and 1962: the Giants (never mind which city) coming from behind with a four-run rally in the ninth to win the rubber game of a three-game NL playoff against the Dodgers (never mind which city), starting the World Series the very next day against the New York Yankees and eventually falling short. The parallels had to be striking half a century ago, but you don't hear much about it today. Now you have. I've done my part.
We've been looking a lot at the general cases, but what about the specific ones? How do teams fare in the exact circumstances that Wild Card knockout game survivors are going to face this year?
Barring any tiebreaker games (and given the tight scheduling, Bud Selig has to have his fingers crossed hoping for none of those), Wild Cards will get a day off before their one-game playoff. After that, one Division Series in each league begins the next day, the other going off the day after. The Wild Card survivors will be playing in the later series: more maneuvering room in case of those confounded ties, one assumes. That means the Wild Card will have one extra game and one day off, facing a team with three days off.
The combination isn't too common, but both times it's happened have come within the last few years. In 2008, the White Sox won a 1-0 division tiebreaker against Minnesota to face the Rays, and in 2011 the Tigers edged out a 3-2 series win against the Yankees before facing the Rangers. Both times the long-route team lost.
(Playing the earlier series has even worse precedent. The only one added game/zero days off/two days off combination came in 2009 with the Twins. Their epic 12-inning tiebreaker against Detroit earned them a late flight to New York to face the Yankees. As most commentators feared, they didn't have enough left in the tank for the next round, getting swept.)
Is this predictive? Scarcely. A combined 0-2 series record, and a 3-7 mark in games, isn't worth a plugged nickel in the coinage of statistics. At most, it serves as an omen, and not remotely a good one for the Wild Cards of 2012. If I had to guess, though, I'd say at least one of the teams will break that short but dire precedent about a month from now. Long-route teams do have a history of outperforming expectations.
The other angle
Our yield from this vein has been limited to one small nugget—the two-day rest optimum—and one larger one—the long-road team beating the Log-5 numbers. Could the lost rest and the stretched-out rotation not mean anything, or even marginally help a team? I'd rather have more games in the sample, but those extra games haven't been played, so I'm stuck with what I have.
Beating the expectations means less than one might think, though, when the expectations themselves begin to shift. My calculations derive from over 65 years of results in a variety of differing circumstances, but the upcoming Wild Card knockout games will create a new set of circumstances. Prime among them is that the Wild Card winner will promptly have to face the team with the best winning percentage in its league.
From 1946 to 1993, there was no fixed relationship in winning percentages between long-road and short-road teams: either one could have the better record. This changed with the first Wild Cards in 1995, but not completely. A Wild Card team would face the best team in its league in the first round, but only if it wasn't in the same division: it then would face the second-best record. This meant there could be a much smaller gap in winning percentages between the Wild Card and the No. 2 team, or even a reversal, with the Wild Card being better.
This has happened amazingly often. From 1995 to 2011, 20 out of 34 Wild Cards have come from the division with the league's best team in winning percentage, and thus played the second seed instead. (Before anyone complains about the over-stacked American League East, it happened 10 times apiece in the AL and NL.) As for the Wild Card playing its first round against an inferior record, that's happened five times in the AL and three in the NL, with one series apiece between teams of equal records.
I tallied up the Log-5 numbers for every LDS in this time frame. The Wild Card teams had a combined expected winning percentage of .4752 against the first-round opponents they faced (though in real life they went .5111). Had they instead always faced the No. 1 seed in their leagues, the expectation would have been .4514 instead, almost twice as far away from .500.
The new Wild Card system has changed things. The knockout game survivor now faces the team with the best record in the league, division foe or not. The long-road Wild Card will necessarily face an uphill climb, as steeply uphill as possible.
I compared winning percentages of league leaders and teams in the two Wild Card spots during the 1995-2011 Wild Card era. No. 1 teams come out at .6192; clubs in the first Wild Card slot (the only one back then) went .5725, and what would have been the second Wild Cards went .5481. Applying Log-5 to the Wild Card playoff, the survivor would average out at .5609—but if I throw in a historically average 54 percent home-field edge to the first Wild Card for hosting the game, it would be more like .5619.
Games between the league's top seed and the average Wild Card survivor—.6192 versus .5619—would thus come out to a .4410 winning percentage for the long-road Wild Card, well below the projected .4817 for long-road teams in this survey and the .5012 that they actually produced. It also undershoots the .4514 I previously gave for this era, because of the chance that an inferior No. 2 Wild Card would get through to the LDS.
Also, the .4410 figure does not count home-field advantage for the top seed, which would only reduce the Wild Card's prospects. Granted, it also did so for a good swath of the 1947-2011 survey, however little that may have showed up in the results.
I began this article wondering what effect extra games and shorter rest would have on the Wild Cards reaching the Division Series. Instead, it may be the guarantee of facing the best winning percentage in their league that proves the greater drag on their chances.
Will this projection pan out in reality? The presumed disadvantage of fatigued teams hasn't borne out; the presumed disadvantage of Wild Cards in the Division Series has turned instead into an all-time 69-66 mark for Wild Cards in the LDS. Combine the two, and maybe they'll finally get dragged back down to expectations.
But we'll have to watch the games to know for sure.
References and Resources
Baseball-Reference provided the figures as thoroughly as usual.
Shane Tourtellotte is a long-time, occasionally-nominated science fiction writer, currently living in Asheville, North Carolina. He will tell you all about the baseball novel he’s shopping if you give him an inch.