On #TeamBadLuck, and what Florida's misfortune is — and isn't

The saying "three is a trend" — another hoary cliché that everyone knows without knowing the provenance — is always fun to use; if we're going to decide something is a consensus opinion, three people having that opinion, or furthering it, seems like a good arbitrary threshold.

If that saying is true, "Florida is unlucky" went from fringe belief to trend this week.

Adam Silverstein wrote about it on Monday at Only Gators, more to challenge the assumption by presenting facts about historically woeful shooting and scoring and the structual issue of recruiting mismanagement than to accept it. Richard Johnson took his swing in The Alligator on Tuesday, writing what's essentially the standard "Here is a topic, here are quotes, here are some short sentences" column, and mentioning the statistic about Florida being dead last in luck on KenPom. (No offense meant!) Chris Harry filed his report on Florida's luck with a truly awful headline and good quotes from Ken Pomeroy — purveyor, not creator, of the "luck" metric, he, I, and others have used to dub 2014-15 Florida #TeamBadLuck.

Here is Pomeroy's pretty damn lucid explanation of "luck":

"The idea of measuring luck is based in that no team has total control of what happens at the end of the game," explained Pomeroy, who has a civil engineering degree from Virginia Tech and a masters in atmospheric science from Wyoming, to GatorZone.com earlier this week. "If you have 10 close games, you're not going to win all of those games. In fact, even the greatest teams wouldn't be expected to win more than six out of 10 games that are one-possession games. So those that do win more than you would expect, that's where you can assume the result came with some good fortune. Losing more than you would expect? That would be the result of some bad fortune."

Here is how Harry unnecessarily confuses the point:

Florida’s "luck" figure is a minus-.162, a measurement that is determined by a complicated mathematical model that takes in a multitude of statistics -- such as offensive and defensive efficiency, turnovers per 100 possessions, offensive rebounding, free-throw attempts, etc. -- and spits out a baseline number for an expected winning percentage. It's called the "Pythagorean winning percentage" and for the eggheads out there is derived from (AdjO^10.25)/(AdjO^10.25+AdjD^10.25).

Harry's next paragraph is just "Got that?" — because math is hard, or something — and his explanation is complex for no good reason: Pythagorean winning percentage, which I explained at length in previewing Florida's game against Pythagorean kings Kentucky, is a decades-old means of measuring how good a team is expected to be (against a theoretical average team) based on two simple variables; Pomeroy's formula for college basketball uses his adjusted offensive and defensive efficiency.

Explaining that said formula uses "turnovers per 100 possessions, offensive rebounding, free-throw attempts, etc." confuses the matter further, as it seems like Harry not understanding that those things all get baked into the adjusted offensive efficiency and adjusted defensive efficiency that are Pomeroy's only two variables when calculating his ratings.

Pomeroy computes those component efficiencies with a formula that does some smoothing to count possessions, and they are the basis for his ratings, which are given in Pythagorean winning percentage, mostly because that just makes sense: Using the most holistic stats is better than not using them. You could calculate Pythagorean winning percentage with whatever stats you want, like steals per game, or missed threes per half, but they probably wouldn't be nearly as predictive; Bill James, creator of Pythagorean expectations (and/or winning percentages, and/or ratings), just used runs scored and runs allowed for baseball, and that turned out to make a lot of sense.

Adjusted offensive and defensive efficiency are probably the best stats we have (and may ever have) for measuring how teams play on both ends in basketball, and they help to standardize a sport with 351 teams generating tens of thousands of data points to incorporate.

Florida, despite being 12-13 on the season, is actually not a "bad" team by both of those metrics.

Florida is 96th in adjusted offensive efficiency, which is simultaneously by far the worst mark of Billy Donovan's career in the KenPom era (Florida has finished no lower than 28th in offensive efficiency since 2001-02, the first year covered by the database) and still close to the top quartile in Division I. And the Gators are, somehow, 18th in adjusted defensive efficiency, putting them right on the cusp of being within the top five percent of teams in D-1.

Why? Because Florida has played a ton of close games, has only rarely been blown out, and has blown out several teams. And because efficiency is, in essence, a measure of how comprehensively a team is better than another, rather than just that a team is better than another.

It's hard to recall this now, a month and three days after Florida's last win by more than two points, but half of the Gators' wins this season came by more than 20 points, and four of those came by at least 25. Even after starting 7-4, Florida was No. 14 in KenPom, buoyed by those blowouts, before it trekked to Florida State, and took the most hysterical of its close losses on the season. Though the Gators haven't been back to the KenPom top 20 since that night, they also haven't fallen below No. 42.

As of now, their raw adjusted efficiency numbers (which, you know, matter, unlike ordinal rankings) are good enough to situate Florida at No. 33 in Pythagorean winning percentage (or, colloquially, "in KenPom") entering Wednesday.

How, then, could the No. 33 team "in KenPom" also be 12-13? Shouldn't a good team, uh, win games?

That's where a lot of people lose the thread to begin with. But this is where "luck" comes in — and where very few folks grasp what exactly "luck" is in this calculation.

That statistic, the raw one, is -0.162, which ranks 351st, dead last nationally, among Division I teams. But its units are the same as "KenPom," or Pythagorean winning percentage. It's not a measure of how "unlucky" Florida is; it's a measure of how much Florida's underperformed what an expectation says should be its record.

By its Pythagorean winning percentage of 0.8302, Florida should, in theory, be winning 83.0 percent of its games against average Division I teams, which would translate to 20.75 expected wins over a 25-game schedule.

However, that expectation doesn't take into account opponent strength (Florida's Pythagorean strength of schedule is fifth nationally, per KenPom) or game location, the two major variables for a healthy team's performance, nor does it take into account injuries, except as they have already affected a team's efficiency. Sand off a few wins to account for that, as KenPom does, and Florida's expectations drop significantly, to about 16 wins over its 25-game schedule.

And yet: Florida has 12.

This is underperformance as bad "luck," in the sense that Florida has come up against a razor's edge and bled more often than not, when theory suggests that razor should nick exactly as often as it doesn't. Efficiency numbers used as analytic metrics suggest that Florida should have more wins than it does. If you can grasp that, that Florida's performance on a per-possession basis this season is more like a 16-win team than a 12-win team, you have basically grasped all you need to know here.

In case you haven't, though, or if you want more context on Florida's luck, or you want better reasoning for this being an explanation rather than an excuse, read on.

Consider this: Most of what a basketball team does is within its control. It influences the quality of shot an opponent gets, its own shooting percentages, the frequency with which it blocks shots, or forces turnovers, or takes threes instead of twos. Even the actually subjective stuff, like how often a team commits or draws fouls, can be influenced by a team's play: Maybe someone like Casey Prather is particularly adroit at drawing fouls, or a Scottie Wilbekin is great at defending cleanly.

But teams can't control everything. Sometimes a bad shot, be it a low-percentage long two or a contested three-pointer or a twisting fallaway lay-up, goes in. Sometimes the other team makes all or almost all of its free throws, converting on plays that a team literally cannot defend. Sometimes a player ends up in foul trouble, or hurt, at a crucial moment in a game. Sometimes a missed shot turns into two points in a tie game because of a weird carom off the rim: That's good luck for one team and horrific luck for the other.

If teams were perfect and did exactly as we expected them to do based on projections, then there would never be games that go against what we imagined would happen. Those games have dotted Florida's schedule this year, and most of them have resulted in close games: Florida has lost six games by one or two points, winning just two by the same margin, and has lost an additional four more games by seven or fewer points, adding just three more wins in single-digit games.

The Gators are a woeful 5-10 in games decided by single digits this season, after being a staggering 15-2 in those games in 2013-14 (while also, er, being a lot better at basketball), but the rate of the single-digit heartbreakers, at 15 such contests in 25 games compared to 17 over 39 games, is alarming and wrenching. And there were just five games in those 17 "single-digit" games in 2013-14 decided by one or two points (Florida went a moderately lucky 4-1 in them), while Florida has already lost six of those games this year, and played eight.

The best teams, you see, don't usually have to win a bunch of close games, because they don't play them. They blow out their competition on a nightly basis and run up fantastic efficiency margins. This was the blueprint for Florida over much of its last two years of SEC dominance, in which the Gators finished No. 2 and No. 3 in KenPom while beating SEC teams by more than two tenths of a point per possession in each campaign.

But Florida has also made being very, very good at basketball and not so good at close games a habit.

Florida managed to be "unlucky" in 2012-13 (when Florida finished at No. 2 in KenPom and No. 339 in Luck), thanks to beguiling losses in close games. Fans with memories longer than fishbowl residents may recall Florida going 0-7 in games decided by single digits and 29-1 in all other games that season. (Pomeroy, you may also recall, had the proper snark about the reaction to that team.)

That peculiar pain gave those Gators the worst season of luck a team that finished in the KenPom top five a team has ever had, and the second-worst luck of a KenPom top-10 team behind last year's Tennessee squad, which lost 12 single-digit games, and won just four. (Just one team beat Tennessee by double digits in 2013-14, though that team did win by 26.)

And No. 3 on the list of "unluckiest" KenPom top-10 teams ever? 2001-02 Florida, 6-7 in single-digit affairs.

Florida also made KenPom's top 10 with negative luck in 2004-05 and 2005-06, and narrowly missed the top 10 in 2011-12 with bad luck, finishing at No. 11 in KenPom and No. 269 in Luck.

The 2008-09 Florida team that barely missed the NCAA Tournament? It was No. 327 in Luck. The 2007-08 Florida team that met the same fate? No. 274.

It's possible, maybe probable, that Florida's especially prone to "bad luck" because of how it plays in close games. Donovan's teams have always been shooter-dependent, and my list of game-winning drives or post shots by Florida players in the last decade — rather than true jump shots — would be quite short. Shots from greater distances are always less likely to go in and more variable, obviously, and Florida's lack of ability to even create looks inside isn't entirely unique to this year's team.

Florida's also been relatively bad at the free throw line for much of Donovan's career, and thus incapable of creating "bad luck" for opponents there, and has largely turned its defense into one that forces perimeter shots — which turns bombs by gunners like Miami's Angel Rodriguez and Georgetown's D'Vauntes Smith-Rivera into prayers that sometimes get answered.

But though post play has been a weakness over time, it's certainly never been this dramatic a weakness before. And while this year's Gators' have been a bit better at defending the three than a few other Donovan squads in the KenPom era, 2014-15 Florida is currently on pace to be the first Gators outfit to make less than two of every three its free throws over that same span.

Essentially, this is the perfect set of circumstances for a really good, really unlucky team. And Florida is pretty likely to have historically bad luck: Just four teams in the KenPom era have finished a season with a worse Luck rating than Florida's current one, and all of them — 2003-04 Florida Atlantic, 2004-05 Tulsa, 2012-13 Seattle, and 2013-14 Charleston Southern — hailed from outside power conferences.

So, yeah: This Florida team, for right now, is the "unluckiest" power conference team of the KenPom era. And yeah, this may be the culmination of years and years of susceptibility to bad luck turning into an excruciating object lesson in how painful bad luck can be.

The good news? Florida plays Vanderbilt, No. 348 in Luck, tonight. Someone's luck will turn for the better!