I found an interesting blog called Chaos Horizon where proprietor Brandon Kempner predicts the winners through data-mining past award patterns. According to Chaos Horizon, Leckie’s Ancillary Sword would have won in the novel category in any normal year. Here were the predictions:

Ann Leckie, Ancillary Sword: 25.7% chance to win
Cixin Liu, The Three-Body Problem: 22.4% chance to win
Katherine Addison, The Goblin Emperor: 21.1% chance to win
Jim Butcher, Skin Game: 18.1% chance to win
Kevin J. Anderson, The Dark Between the Stars: 12.7% chance to win

Kempner also provides the indicators of a likely win:

Indicator #1: Nominee has previously been nominated for a Hugo award. (73.3%)
Indicator #2: Nominee has previously been nominated for a Nebula award. (73.3%)
Indicator #3: Novel won a same year Nebula award. (87.5%)
Indicator #4: Nominated novel is science fiction. (53.3%)
Indicator #5: The nominated novel wins one of the main Locus Awards categories. (53.3%)
Indicator #6: Nominee places in the Goodreads Choice Awards (100%)
Indicator #7: Nominated for at least one other major award (80%)
Indicator #8: Nominee highly regarded by critics, as judged by Critics Meta-List. (86.7%)

Although Cixin Liu was pretty much unknown in the US, Kempner credits translator Ken Liu for his high rating for The Three Body Problem. It is interesting that science fiction wins more often than fantasy. Because of the trend toward sentimental works, I had predicted Sarah Monette’s The Goblin Emperor would win this year.