There are plenty of kinds of luck in football: injury luck, fumble luck, weather, turf conditions, a new coach inheriting a supremely talented quarterback who probably shouldn't have signed with the old regime because he was a mismatch for it. You know, those kinds of luck.
One way to put a number on luck is to use Pythagorean expectation. That method comes up with an approximate number of games a team should have won based on its points scored and allowed. The difference between the expected wins and actual wins is luck.
This is not a perfect measure of luck, of course. Bad injury luck won't show up, for instance. As your team declines in quality as guys go down, its point differential will get worse and therefore not manifest as bad luck in the metric. Plus a team that runs up the score in blowouts will have artificially low luck, while a team that pulls the starters early will have artificially high luck.
It's not a bad way to look at luck though, and I'm going to use it to address one of the most persistent questions of the offseason: just how lucky is Gus Malzahn?
I have Pythagorean expectation data for I-A football since 1998. I made a master list of every luck score that every team has had in that span, a total of 1883 team-seasons. I compiled luck scores for a variety of coaches who have been head coaches for at least five seasons. I found the median rank for those luck scores and then looked at where that would stack up among all of the seasons on the list.
That's hard to follow, so let's do an example with Dan Mullen. Among the 1883 team luck scores, the luck scores for his five seasons in Starkville rank 1348, 1049, 1674, 1061, and 1475. The median there is 1348, which puts his median luck at the 28.4 percentile. Mullen has had some bum luck in his seasons there. Guys who have been coaching for a while tend to have their median luck be right in the middle due to mean reversion. The percentile for Mark Richt's median luck is 53.2%. Bob Stoops's is almost dead in the center at 50.9%.
Here is how 30 notable coaches stack up:
|Coach||Median Luck Rank Percentile|
Ouch, Bill Snyder. Also, what does it say about Charlie Weis that he has had generally pretty good luck and still got fired from one job and is on the hot seat in another?
Anyway, no matter how you slice it, Malzahn's sample size is small. He's been coaching on the I-A level for just eight seasons. His first one at Arkansas famously had Houston Nutt overruling him constantly, and he was a co-offensive coordinator in his two years at Tulsa. His era of having truly significant influence is just the five years from 2009 and on.
Malzahn's first year at Auburn wasn't anything notable, with a luck rank of 850 (54.9%). From there, it gets crazy:
- 2010: Rank 3, 99.8%
- 2011: Rank 20, 98.9%
- 2012: Rank 337, 82.1%
- 2013: Rank 143, 92.4%
His median luck is his 2013 season: rank of 143, or 92.4%
Yes, two of the top 20 luckest seasons from 1998-2013 featured Malzahn as offensive coordinator, and three of his five seasons of influence are in the top decile of luck. The tipped pass against Georgia and the Kick Six are just two games out of dozens; even setting them aside still gives Malzahn an unsustainably high career luck score.
If there is a close analogue to Malzahn, it's Mark Dantonio of all people. He was defensive coordinator for Jim Tressel at Ohio State for three years prior to becoming a head coach. His first year as Tressel's DC was nothing unusual (rank 1097, 41.7%), but the next two were very positive: in 2002 it was a rank of 80 (95.8%), and in 2003 it was a rank of 53 (97.2%). Dantonio took the Cincinnati head coaching job after the '03 season, and Tressel's luck never ranked above 814 (56.8%) from then on.
Dantonio's second season in Cincy featured a rank of 79 (95.8%), matching Malzahn's three top decile scores in his first five years as a sole coordinator or head coach. Dantonio's luck has remained pretty positive overall, as evidenced by his median rank percentile being 74.4%, and as a head coach he's had three seasons in the top decile of luck with none in the bottom decile of luck.
As we see with Dantonio and Brian Kelly, who each have ten years of head coaching experience, it is possible for a coach's luck to remain quite positive over a longer period of time. Kirk Ferentz and poor Bill Snyder show the inverse is possible too.
What isn't possible, however, is for luck to remain consistently at one level. Dantonio has had three seasons between the 10% and 20% level; Kelly has had three at 35.8% or worse. Ferentz has two top decile seasons to go with his five (!) bottom decile seasons, and even Snyder managed the 13th most positive luck season among his rash of bad ones.
Malzahn has had a knack for both playing in and winning close games, something that will really inflate a Pythagorean expectation luck score. His mastery of both running the ball and the two minute drill certainly help with that.
What he hasn't done yet is be a part of a team that dominated from wire-to-wire. Both of the best teams he's been a part of—2010 and 2013 Auburn—started slowly and were dramatically better at the end of the year than they were at the beginning. Improving as the season goes along is a sign of good coaching for sure, but again, it will increase your luck score. The best teams typically top out in the low to mid 80s in percentile, and some don't break even: 2008 Florida (43.8%), 2011 LSU (48.0%), 2011 Alabama (31.2%), and 2012 Alabama (40.4%) are examples there. The general consensus on the best team of 1998-2013 centers on 2001 Miami (FL), and its percentile is a solid but unspectacular 62.3%.
Malzahn won't churn out top-decile teams forever. If he continues his run of success, he'll eventually get a really good team that doesn't have to play so many close games. He will also sometimes encounter bad luck, because bad luck strikes everyone.
He's been both lucky and good, with the kind of good he's been being the type that also makes a coach look lucky. Unless this truly is the only way to explain Gus Malzahn's exploits, he won't continue to appear outlandishly lucky for long.