I had a history professor in college, I don't remember her name or even the class she taught, but she would routinely stop her lecture in order to make sure we understood that no one knew how to say things quite like the ancient Greeks. In a constant attempts to get at the meaning of something, she was routinely frustrated with the fact that the English language simply wasn't capable of getting at the truth.
And so I can only imagine if she somehow got herself into computing Points Per Possession for college basketball teams and indulged in the related analysis.
When you take the offensive and defensive PPP stats, you can get a net number that allows for a projected record (Pythagorean Wins, which are also commonly used in baseball).
When you stray from that number, the term "luck" is used to explain the difference. As John Gasaway of Basketball Prospectus explains:
Luck is defined by me as: the difference between a team's actual winning pct. (in-conference, the way I do it) and the winning pct. that would be predicted by that team's per-possession scoring margin. For instance if a team scores one point per possession and gives up one point, they would be predicted to go 9-9 in the Big Ten. If they go 11-7 they're "lucky." If they're 7-11 they're "unlucky." This post by a Georgetown blogger covers a lot of the same ground that I do and has way better graphics.
And here is where things take a horrible and tragic turn.
Lucky, as we know it, is usually interpreted as a bad thing. It often means you didn't earn it at best and you don't deserve is at worst. So why is this variance called luck? "The English language has failed us this time," Gasaway goes on to say in his email, and he's absolutely right.
Why does any of this matter? Because the stats, and then the word, then get used the wrong way, for the wrong purposes. Big Ten Geeks suggests after the Iowa loss:
The loss by Penn State killed their chances at a #2 or #3 seed in the Big Ten Tournament. Maybe that's for the best, given their efficiency margin. It just doesn't seem right to hand out such a high seed to a team that's actually been outscored by its opponents in conference play.
Now I'm operating under the assumption that this is partly said in jest; BTG did a fantastic job covering the basketball season, often with keen observations, and so I'm going to give them the benefit of the doubt and dismiss the idea that they don't think Penn State would have earned a 2 seed with an Iowa win because their overall point total.
But to pick out of the comments for a second, something I hate to do but need to this time in order to illustrate poorly applied logic:
I don't see why everyone is assuming Penn State is in. I agree that 8 teams will be hard to get in and I really think PSU might be the ones left out. Not only am I pretty sure that they are the worst of the 8, but their rating on Pomeroy and RPI confirm my thoughts. I would be happy with 7 teams in the dance. As an Ohio State fan, I'm kind of hoping Penn State loses to Indiana, which would make me less nervous about the Buckeyes getting in. Their non conference schedule was a joke and they happened to catch MSU and Purdue on off nights. They also have a ton of close wins and are high up in the "luck" factor on Pomeroy. Great job (as usual) of pointing out that Penn State was outscored in conference play. That statistic is extremely telling.
This couldn't be more backwards. You play to win the games, as somebody famously said, not to look pretty doing it. This recent trend in both college football and basketball to selectively pull out stats to enhance or discount the actual result on the field is, well, to use an old person's word: disturbing. Coaches can no longer be coaches, they have to work as politicians after each game and sell their accomplishments to ill-informed voters and elite committees. It's sad, really.
I love more telling stats because they allow for enhanced in-game strategy as well as a better understanding of "what happened", but I don't understand the concept that they should be included in the standings, that they are 'telling' from a legitimacy point of view, or that ranking low in highly correlated but still flawed numbers means your wins are legitimate.
Gasaway gave me a heads up of an email sent to him that he published in his latest column:
I had a comment concerning the column on "luck."
As far as I understand it, "luck" is the difference between the pythagorean winning percentage and the team's actual winning percentage.
In the business that I work in (economics), I would call that "error," not "luck." Calling it luck presumes the pythagorean percentage is the exact specification for winning.
This is what I've been saying all along: the confusion here arises from the English language, not from the numbers. "Luck" connotes meanings which I do not hold with regard to these teams. Therefore I propose to start calling this phenomenon a team's degree of DeChellis, in honor of the Penn State coach who's had an extremely high degree of DeChellis for two seasons running now.
Though, truth be told, the Nittany Lions actually lost a close game (!) Saturday at Iowa. It looked very strange. I didn't know that was possible.
So not all hope is lost, there are people out there who get it. It's just a matter of filtering out the noise, I guess. Either that or saying forget the whole thing and going back to talking grittiness with Joe Morgan.