Shortly after the San Francisco Dons defeated the BYU Cougars on Thursday night, Matt Olson of Omni Rankings tweeted what many women's basketball fans were already thinking: Jennifer Azzi's Dons might've pulled off the upset of the year.
Helen of Women's Hoops Blog reiterated the point on Friday, also suggesting that USF's feat wasn't quite as "sexy" as the Alabama Crimson Tide's 77-75 upset win over the Kentucky Wildcats. Then there was the St. John's Red Storm's 57-56 win over the UConn Huskies, which was an upset in part because of the Red Storm interrupted a 99-game home win streak for the Huskies.
But this brings up a fun dilemma for basketball geeks, inspired by the very existence of Omni Rankings' upset rankings: What constitutes the "Upset of the Year? And how on earth would we go about determining that with all the games that are played during any given season?
Obviously, there's some subjectivity in figuring out which game would be the "Upset of the Year": as a Michigan Wolverines football fan, for example, every loss is an upset because they are the best, champions of the West, heroes, leaders, and The Victors (as the fight song says, though not necessarily in that order).
In seriousness though, the point is that a lot of how we define an upset depends on our expectations. Every year there's a "seeding upset" in the NCAA tournament in which a higher seed loses to a lower seed that many close observers of women's basketball would hardly consider an upset. Not just 9th seeds upsetting 8th seeds or 10th seeds upsetting 7th seeds, but also cases like the 11th-seeded Gonzaga Bulldogs beating the 6th-seeded Iowa Hawkeyes in Spokane last year. Many astute fans expected Gonzaga to win, if for no other reason because they were at home, but also because they were on a roll and being led by a point guard that would end up being a WNBA starter a few months later.
For now, I'll use Gonzaga's tournament run to further illustrate to the point: there are times when records, rankings, or seeds would suggest that an upset occurred, when in reality a closer look at the matchup indicates that the so-called underdog should have been expected to win the game. Perhaps one or both teams was over/under-rated by the average fan or perhaps the "underdog" possessed a strength that just so happened to exploit a fatal flaw for the favorite.
That can be hard to determine, particularly with so few women's basketball games being televised and advanced statistics - that can help make these dynamics transparent - not made as widely available as in the men's game. But that's essentially what Omni Rankings attempts to do with their upset rankings: account for all the dynamics that contribute to winning in order to determine which games are truly upsets or the most unpredictable wins by the underdog.
Yet after seeing the Omni Rankings tweet, Shannon and I went back and forth a bit about how they determine how they might rank the biggest upsets. Part of the "problem" in figuring that out based on their rankings is that they have two lists: macro and micro upsets. But I'd argue that some additional information is still needed.
In short, there's no one thing that frames expectations that shape an upset, but the following might be some major factors:
- Omni Rankings macro & micro upsets: The labels make a lot more sense after knowing what they're based on. "Macro ratings" are based on the big picture things: strength of schedule and margin of victory or the things that might go into determining RPI. "Micro ratings" are based on more specific performance variables: I would usually use Dean Oliver's Four Factors for that, but Omni Rankings uses 24 offensive and defensive variables. So those two things answer two questions, to put it in simpler terms: 1) How does a team stack up relative to other teams? and 2) What are there specific strengths and weaknesses that might contribute to winning and losing?
What their upset wins rankings reflect is two separate things: teams that have overcome the biggest macro deficits and teams that have overcome the biggest micro deficits. So there's an easy answer to start narrowing the list: an upset that shows up at the top of both lists - meaning an outcome defied both macro and micro expectations - is a clear candidate for Upset of the Year. More importantly, Omni Rankings provides an easy way to look at whether a team over/underperformed to truly pull off an upset.
- National standing at the time: However, one thing that Omni Rankings doesn't easily offer is where the team ranked nationally at the time. WBB State does though, listing their "State ratings" for every Division I team at the bottom of every box score from before and after a game. That's just another way to look back at what expectations might have been at the time to help stave off the lure of presentism. On a few occasions, State ratings suggested that what many people considered an upset was actually just a matter of one team being overrated and another being underrated.
- National Significance: But even knowing rankings don't tell the full story. For example, let's say Team B upsets Team A and Team D upsets Team C with rankings before and after as follows (taken from actual examples):
Upset 1 Before After Upset 2 Before After Team A 9 13 Team C 96 124 Team B 194 175 Team D 293 268
Although both of these are rather significant upsets with a similar ranking disparity between the two teams, Upset 1 was much more significant nationally - while Team A is a top 10 team in the nation, Team C looks WNIT bound.
- Stakes/implications: Let's take that a step further: in addition to national significance, there have been some upsets that have major implications. For example, an upset that knocks a team out of first place in a conference that is likely to send only one team to the tournament is probably more significant than a game that ends up being a less consequential bump in the road.
- Time of season: This turns out to be a big one, particularly given this year's set of upsets. There were a number of early-season (non-conference) upsets that looked huge at the time statistically. But for more than one of them, it turned out that early-season performance was just misleading for one or both teams. The sample size is just so small early on that it's almost just as likely that our expectations were just wrong moreso than an inferior team beating a better team.
- Location: It probably goes without saying that a road upset is bigger than a home upset.
Top 5 Upsets of the 2011-12 Season
Games formatted as follows: Underdog (change in State rating) at/vs favorite (change in State rating)
This is actually Upset #1 above and, yes, it was pretty surprising, although Scott Venci of the Green Bay Press-Gazette correctly warned that Detroit might have been the team most likely to knock off Green Bay. And since Green Bay was without their starting forward Sarah Eichler, this one can almost be excused.
Nevertheless, it was the highest rated macro and micro performance for Detroit while being the lowest rated macro and micro performance for Green Bay. And it is still Green Bay's only loss of the season.
As the UTPA release stated, there were about 100 RPI spots between these two teams at the time and about a 200-spot gap in the State ratings. It lacks the significance or implications of some other games, but when you start narrowing things down, there were few bigger upsets (without injuries) on paper.
This game is Upset 2 in the table above, with Team D being Clemson and Team C being UNC. In other words, although it's a rather big upset UNC wasn't quite as strong as some of the other teams who have been upset this season. But this was the biggest micro upset according to Omni Rankings and was the highest rated performance for Clemson and the lowest rated performance for UNC in both the macro and micro ratings, meaning that both teams defied expectations on that day.
The thing is that, despite the arguments of some national commentators, you could quite easily argue that UNC was not a tournament team even at that time. So in the end, the game could end up having little national significance.
Similar to the game above, this was the highest rated performance of the year for St. John's in both the macro and micro ratings but only the lowest rated game in the macro ratings for UConn. And put the Top 25 rankings aside - at first glance, the State ratings don't reflect as big an upset as either of the two above. People might have to come to terms with the fact that the pollsters simply underrate St. John's, for whatever reason.
However, the fact that the Huskies haven't lost a game at home in so long has to figure into the equation here. And on top of that, this also puts UConn in a precarious position in the Big East standings: as John Altavilla of the Hartford Courant wrote, if the Huskies lose to the Notre Dame Fighting Irish and St. John's wins out, Geno Auriemma's team will end up seeded third in the Big East tournament. Is that reason for panic? No. But it does mean that this loss will end up being pretty significant.
When all is said and done, yes, this still ended up being the Upset of the Year. USF entered the game with 2 wins second from last in the West Coast Conference. USF has won 9 games in coach Jennifer Azzi's tenure with only a few games left in this season. It was by the far the best defensive performance USF has had all season.
For their part, BYU was coming off a 30-point win in Provo against Gonzaga. They were ranked #23 in the nation at the time. The differential in State rating at the time between these two teams: 304. Nevertheless, BYU had their worst shooting performance of the year (31.6%). On top of all that, this was a game BYU needed to keep pace with the Bulldogs atop the West Coast Conference standings.
It was the highest rated macro and micro performance for USF and the lowest rated in both for BYU.
BYU has since fallen out of the top 25 as a result of this loss, although they are still receiving votes. Even more importantly, if they don't win the WCC's automatic bid, this is the loss that would make it nearly impossible for them to make the NCAA tournament as an at-large bid because their only significant win is against Gonzaga.
Other upsets might have gotten more attention from national commentators, but this is almost indisputably the biggest of this season so far.
All the statistics suggest that there's not quite as much separating these two teams as there was the others above; that's punctuated by the fact that neither team's State rating really changed a whole lot after this game.
This was a huge upset but the reason it isn't higher is because the Terrapins were missing Alyssa Thomas, a starter and the ACC's second-leading scorer. So although this is at the top of Omni Rankings' list, you have to adjust for that. Still, that's a pretty huge loss for Maryland at home as they are currently separated by about 300 spots in the State ratings.
It's almost surprising to see this game on the list looking back on it now - St. Mary's also beat BYU before beating Gonzaga and this is not the dominant Gonzaga team of past years in light of a 30-point loss to BYU themselves. But it's not insignificant either as that loss could hurt the Bulldogs if they don't get the WCC's automatic bid.