The Fate of Teams up 3-1 in a Series

M

MentalDisabldLst

Guest
So, the Giants are up 3-1 over the Cardinals in the NLCS.  And while some of us would be happy for the Cards' season to end ASAP, some others prefer more baseball to less baseball, at least when the Sox aren't involved.  So I thought, how often does each outcome happen, once a team goes up 3-1?  How do these series go?  I dug a bit through the Postseason History on B-Ref, and a few things surprised me that I thought I'd share.
 
To set the table, we all know the fate of teams that go up 3-0 in a series, and get reminded of it frequently this time of year.  The series-leading teams are 27-5 in Game 4 (84%!), 2-3 in Game 5 (SSS!), 2-1 in Game 6 (1998 and 1999 NLCS, the latter of which had an absolutely epic 2nd half of the series), and of course, 0-1 in Game 7.  Or 1-0, depending on how you look at it.  The remarkable thing there is the record in game 4s; if your team is down and you manage to win game 4, you're in a rare category, very poorly-charted territory, and 1 of the 5 teams in that small club pulled off the whole enchilada.
 
But what of teams that go up 3-1 in a series (without having been up 3-0)?  It's a much more common occurrence.  So first off, there have been 38 best-of-7 MLB series to end 4-1, and of those, only two (1910 and 1937 WS) came in the form WWWLW.  So that's 36 wins in Game 5 for our teams of interest.
 
How many losses, though, in game 5?  Well, those might be found from among series that ended in either 6 or 7 games.  I searched for series where a team was leading 3-1 (but not 3-0), lost game 5, but won game 6.  I found 19 of those, spoilered below.
 
Win in 6 after being up 3-1 & losing game 5:
 
2013 NLCS Cards over Dodgers
2011 ALCS Rangers over Tigers
2010 ALCS Rangers over Yankees
2010 NLCS Giants over Phillies
2009 ALCS Yankees over Angels
2005 NLCS Astros over Cards
2000 ALCS Yankees over Mariners
1997 ALCS Indians over Orioles
1995 WS Braves over Indians
1993 WS Jays over Phillies (Joe Carter come-from-behind series-winning homer! 15-14 slugfest in game 4!)
1992 WS Jays over Braves
1990 NLCS Reds over Pirates
1977 WS Yankees over Dodgers
1959 WS Dodgers over White Sox
1948 WS Cleveland over Boston
1936 WS Yankees over NY Giants
1935 WS Tigers over Cubs
1918 WS Red Sox over Cubs
1911 WS Philly As over NY Giants
 
So that's game 6.  But how often do such series go 7 games?  Well, here's the list of teams that completed the comeback and won the series after being down 3-1.  Those who came back from being down 3-1, but who lost game 7, are in parentheses.
 
- 2012 NLCS, Giants over Cards (so the Cards doing it this year would be revenge, really)
- (we came close vs Tampa in 2008, though really, it was game 2 we should have won, not game 7) 
- 2007 ALCS, Sox over Indians
- *2004 ALCS, Sox over Yankees (* started as a 3-0 lead)
- 2003 NLCS, Marlins over Cubs
- 1996 NLCS, Braves over Cards (last 3 games were a combined 32-1 for Atlanta)
- (Pirates came close vs Atlanta in 1992 NLCS, lost game 7 in famously heartbreaking fashion)
- 1986 ALCS, Sox over Angels (holy crap, that Game 4 was nearly as dramatic as Game 5)
- 1985 WS, Royals over Cards
- 1985 ALCS, Royals over Jays (they did it TWICE in one postseason?!)
- 1979 WS, Pirates over Orioles
- 1968 WS, Tigers over Cards
- (Sox came close vs Cards in 1967 WS, of course, but Lonborg was never beating Gibson on 2 days' rest)
- 1958 WS, Yankees over Braves (revenge for the Braves' win in '57)
- 1925 WS, Pirates over Senators
- (NY Giants almost pulled it off vs Red Sox in 1912.  The series had 1 tie, and the clincher was surely the most exciting Game 8 in history)
 
That's the list: 4 teams previously leading closed it out, but 12 finished choking away the series to the comebackers.  And of the 4 failed comebacks, the Red Sox were on the losing end of 2 of them.  So that covers all the situations, and I think we can put them together to make some conclusions:
  • Teams that win game 4 to go up 3-1 in a series are 36-34 in Game 5, basically a 50-50 proposition.  If there are home/road splits in those numbers, or any other subsets of interest, I don't know of them.  38-37 if you count the series that initially went WWWL.
  • Teams winning game 4 to go up 3-1, but losing game 5, are 19-15 in Game 6 (54%).  21-16 (57%) if you count the series that started 3-0.  Again, not really distinguishable from a coin flip.  But:
  • Teams that are down 3-1, but who then win Games 5 and 6 to force a Game 7, are 11-4 (73%) in those Game 7s. 12-4 (75%) if you count the 2004 Red Sox.
That last one is an eye-grabbing number!  Yes, small sample size and all, but the degree of difference between that and the outcomes of most other situations is striking.  Could there be psychology at play, with the comeback team feeling confident and the team who coughed up the series lead feeling nervous and vulnerable?  TV talking heads certainly hope so.  I've forgotten too much about Student's T-test to figure out whether a 12-4 record is statistically distinguishable from being 50-50, so I'd welcome a hand there.
 
But the point is this: teams up 3-1 are just as likely to win game 5 and game 6 as they are any other game.  But if they lose both, history says they usually lose Game 7 as well.
 
M

MentalDisabldLst

Guest
Not quite, I flipped the order of the listing results for Game 7.  If we count series that started 3-0, teams leading 3-1 won the series (38 + 21 + 4) = 63 times, and lost 12, meaning they won 63/75 = 84%.  If we don't count those, then teams leading 3-1 while winning game 4 won the series 36 + 19 + 4 = 59 times, and lost 11.  59 / 70 = 84%.  So you're right on the percentages, just not the numbers.  And that's a little lower than you'd expect just based on 3 coin flips (87.5%), primarily due to the performance in Game 7s, but it's not clear to me whether that differs meaningfully (i.e., in a statistically significant sense) from truly being a series of coin flips.
 
... and add another one in the win column based on last night, I guess.