The predictive power of second-order wins

Last month, as I got knee-deep in my S&P+ ratings redesign, I started tinkering with what I called second-order wins.

There are exponents available for turning college football points into a Pythagorean win percentage, but I'm more interested in another concept: second-order wins. That basically takes the same idea but uses advanced stats of some sort to determine not simply what you did score and allow, but what you should have scored and allowed.

My new ratings are based on margins in categories related to my Five Factors: efficiency, explosiveness, field position, finishing drives, turnovers/luck. As I flesh the system out with previous years of data, I'm able to basically use these margins to determine both what was your most likely scoring margin in a given game and, based on the plays that took place, your likelihood of winning a given game.

To further explain the second part of that last sentence, it basically says "If you took all the plays in this game, tossed them up in the air, and had them land in a random order, you'd win this game XX% of the time." It is a single-game win likelihood concept, and with it, we can look at wins and losses not as zeroes and ones, but as percentages. And if you're winning a lot of "You'd have won this game 60 percent of the time" games, you're probably getting a little bit lucky. And as with everything else, that luck is likely to change over time.

Over the last couple of weeks, I've basically been watching the Australian Open and playing in Excel sheets. (What a dream life I lead.) I've pretty much gotten to where I want to be regarding the redesign, so you can expect an unveiling of sorts some time in the next couple of weeks.

In the meantime, though, I wanted to take a look at what 10 years of second-order wins can tell us.

The premise of the second order is easy enough to understand, right? It's basically one more way of looking at unsustainable good or bad fortune. It basically says that playing the way you did, against the opponents you played, you would usually end up with a record of X, and it compares it to your actual record, Y. If one is too far from the other, you're probably going to see some regression (or progression, as the case may be) to the mean.

This is logical, and with years of data in my back pocket now, I can say it's nearly incontrovertible, too.

Like, ironclad.

From 2005-13, 35 FBS teams ended up with an actual win total at least 1.8 games lower than their projected second-order win total. That means they underachieved pretty drastically compared to what the stats would expect. Of these 35, four saw their actual win percentage regress the next year, four stayed the same, and 27 improved. Of those 27, 15 improved by at least 20 percent.

Meanwhile, from 2005-13, 32 teams managed an actual win total at least 2.0 games higher than their projected second-order win total. This means they overachieved compared to what the stats expected. Of these 32, two saw their win total improve the next year, two stayed the same, and 28 regressed. Fourteen of those 28 saw their actual win percentage drop by at least 20 percent.

That's pretty definitive.

50 biggest underachievers according to second-order wins (2005-14)

From this viewpoint, the 2014 TCU squad was likely to have a pretty good season no matter what this year. Gary Patterson's offensive coordinator upgrade, combined with a lovely amount of returning experience, made it even better than "pretty good."

50 biggest overachievers according to second-order wins (2005-14)

Naturally, Bill Snyder is the biggest trend-wrecker in the business. That 2011 Kansas State team was strangely fantastic at getting lucky and/or making the one single play it absolutely had to make. And then the 2012 team won the Big 12.

But KSU did have Collin Klein back the year after its strange 2011 run. Florida State won't have Jameis Winston back in 2015. And while there's obviously almost nowhere to go but down from 13-1, this suggests the regression could be worth a few games. And I guess this is a warning to fans of Bowling Green, Arizona, and Utah, too.

Other teams that overachieved by more than 1.0 games: UCLA (1.8), Northern Illinois (1.7), Missouri (1.7), Rutgers (1.5), Ohio (1.3), Hawaii (1.3), Colorado State (1.2), Ohio State (1.2), Wyoming (1.2), North Carolina (1.1), USC (1.1), and Michigan State (1.1).

On the flipside ... congrats, Pitt fans! You got yourself a new, exciting head coach, and he's probably going to have a strong first season! Same to you, fans of Kent State (underachievers by 1.8 games), Colorado (1.6), San Diego State (1.6), UMass (1.4), Stanford (1.4), East Carolina (1.4), Miami (Ohio) (1.3), Washington State (1.2), Boston College (1.2), San Jose State (1.1), Miami (1.1), Arkansas (1.1), Central Michigan (1.1), Indiana (1.1), and Louisiana Tech (1.1).