Friday, February 20, 2009

Tournament of Books Bracketology

A couple of weeks ago I posted on The Morning News's Tournament of Books, which had just released the field of competitors, though without seedings or matchups. Now the bracket and judge assignments are out.

I've now done a bit of research into past results (2005-2008), trying to glimpse any patterns or tendencies in previous tournaments' outcomes. It's a small data set (just 64 books, 66 total matchups), and there are a number of variables which are really difficult to account for (mostly involving which judge is assigned to which book), but if ESPN can eke months worth of coverage out of the highly inexact science (a.k.a informed bullshit) that is "Bracketology," I suppose I have enough to fill out a post or two.

I tried five different types of variables to see if any patterns emerged: seeding, the publisher of the book, the awards it won, the page count of the book, and gender of the author. I also noted the books which were paperback originals and the books which were short story collections, and the tendencies of judges who have participated in more than one Tournament. I'll get to the awards, page counts, and judges' histories soon; for now, enjoy the following breakdowns.

Seeding turned out as one might expect:
  • Only #1 or #2 seeds have won the Tournament (two of each)
  • #1 seeds win their first round match-up 75% of the time (twelve out of sixteen), and 75% of those first round winners also won their second round. Two of the four #1 seeds knocked out in the first round were revived in the Zombie round.
  • However, only about 19% of #1 seeds advance all the way to the Championship bracket without a loss, which is actually less than one a year, so the #1 seed is not an automatic path to victory by any means.
  • The #4 seed is not a very good place to be: of the four books which were #4 seeds and won their first round matchup, only one won in the second round, and that one (Half of a Yellow Sun, probably a seeding mistake anyway) lost in the semifinals. (Remainder last year threw everyone a curveball and was revived in—and won—the Zombie Round).
  • #2 and #3 seeds have split first round victories, though it seems #2 seeds have significantly better luck in the second round, winning about 63% of the matchups they feature in, though that number's a little deceiving. Their winning percentage is split between three wins against #4 seeds—out of three matchups—and two wins against 1 seeds—out of five matchups.
  • #3 seeds do not fare well past the first round, winning only once against a #1 seed in the second round, and only 13% of their second-round matchups overall.
The publisher of the book actually turned out to be a good, though very general, predictor. (For uniformity's sake, I tried to use only the publishers for the hardcover edition. Also, I recognize that some companies are subsidiaries of others, but for simplicity's sake, I'm just going to keep them separate.) Here's how it turned out:
  • Knopf leads the way with 10 books in the field (or about 16% of all the books selected), Random House, Farrar, Straus Giroux, and Houghton Mifflin come in next with 7, 5 and 4 respectively, altogether representing more than two-fifths of the books in the field. No other publisher has sent more than 2 books to the Tournament. I guess these would be your analogues for the "power conferences" which routinely send 6 or 8 teams to the NCAA tournament.
  • Three of the four winners have come from these "conferences," half of the runners-up, half of the Zombies, a full half of all first round winners, and also half of the second round winners. In other words, they've been dominant collectively.
  • However, they have not all performed evenly. Knopf is great at getting its books into the second round (winning 6 of 10 first-round matchups), but bad at advancing past there. Of their books, only The Road made it past Round Two.
  • Random House is much better at getting their books deep into the Tournament: not only have they produced two champions, but of the four books that won their first round matchup, three also won their second round. And one of their books which was knocked out in the first round came back as a Zombie and made it to the Championship (Absurdistan, 2007).
  • Houghton Mifflin has had terrific success in the first round—all four of their books have won there. Two of those won the next round as well, and one ended up a runner-up.
  • FSG has had fairly mediocre results: of its five books, two have made it past the first round, but those two also won their second round matchups.
  • The only really successful "minor conferences" have been Riverhead and Back Bay. There have been a number of publishers that have had individual books do well, but Riverhead and Back Bay have had some success with each book they've sent to the Tournament. Last year, Riverhead's two entrants made it past the first round, and Oscar Wao won the whole shebang. Back Bay has had two entrants, and both have won their first and second and semifinal matchups, although both tragically lost to Zombies, the only Zombie victories in the (short) history of the Tournament.
  • In a glass half full kind of way, although "minor conferences" haven't had the kind of success the big publishers have, it's still nice to see almost 60% of the entrants have been from small or independent presses.
Gender is an interesting variable, although again, not really a case of overturning what would have been the expectation:
  • Of the 64 books to have participated (not counting this year), 38 have been by men, and 26 by women (a 60/40 split). Interestingly, the past two years have been increasingly less balanced: 2005 featured seven women authors, 2006 had eight, but 2007 dropped to six, and 2008 had just five. 2009 again has five.
  • But more than just quantity, men slightly overperform their numbers in the first round: in twenty-six total matchups of men vs. women, men win about 62% of the time. (There has never been a first-round woman/woman matchup).
  • In the later rounds, the women who survive the first round hold their own, splitting the remaining men/women matchups.
  • There has been only one championship match featuring a woman, though Ali Smith did win that, for The Accidental in 2006.
  • Poor first round performance might be an issue of seeding: only four women have held #1 seeds (against twelve men). Toni Morrison's A Mercy makes it five this year. [Edit 2/22: On further reflection, seeding doesn't completely account for the poor performance. Two of the women who have been seeded #1 lost in the first round—Ann Patchett last year and Zadie Smith in '06—making women three times as likely to lose as a #1 seed in the first round.]
Check back later for further analysis. Sportscenter coming up next.

[Part Two—analysis of the effects of page count—is here.]

1 comment:

Condalmo said...

Nice work. See you at the Tournament.