On Monday, I delved a bit deeper into the idea of second-order wins.
[I]t 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.
The idea of yesterday's post was to look at how particularly fortunate or unfortunate teams are all but guaranteed to see their luck regress the next year. That's good news for a team like Pitt (which won 2.3 fewer games than its stats would have suggested in 2014) and bad news for teams like Florida State (which won 3.5 more games than it maybe should have), Bowling Green (2.7), and Arizona (2.0).
When I started looking at individual schools, however, I began to notice things. Under Tom O'Brien, for instance, N.C. State overachieved (by this definition) at least a little bit for six consecutive years. Under Frank Solich, Ohio has overachieved in five of the last six years. Under Steve Spurrier, South Carolina has overachieved in five of the last six years. Louisiana-Lafayette has overachieved in all four years of Mark Hudspeth's tenure.
Meanwhile, under both Paul Johnson and Ken Niumatalolo, Navy has overachieved in nine of the last 10 years.
On the flipside, Tulane underachieved in four of Bob Toledo's seasons and did so rather drastically three times. Washington underachieved significantly three times in four years under Tyrone Willingham. Arizona underachieved in four of the last five years under Mike Stoops. Houston Nutt teams at Arkansas and Ole Miss underachieved in six of seven years.
Might coaching have an effect on your ability to win (or lose) close games that go against the run of the stats?
Because we're dealing with such small samples, it's hard to reach any specific conclusion in this regard. So instead I'll just share some data.
Below, you'll find a list of coaches who have been head men for at least three seasons in the last 10 years. You'll also find the average difference between a team's wins and second-order wins. A negative number is bad -- it means your average win total is worse than the stats say it should be -- and a positive number is good.
Samples are small, and it would be foolish to make any serious judgments off of three seasons. But let's just say that narratives and stats sometimes agree.
Avg. Difference per season
(negative = bad)
|John L. Smith||3||-0.71|
If a random college football fan were asked to create a list of what we'll simply call "bad coaches," he or she would probably fairly or unfairly mention Ty Willingham, John L. Smith, Dave Wannstedt, Ted Roof, Houston Nutt. Washington State fans might mention Bill Doba. North Texas fans would definitely list Todd Dodge. They're all on one particular end of the chart.
Meanwhile, asking someone to list "good coaches" might produce a list with men like Bill Snyder, Gus Malzahn, etc. They can be found anchoring the other end. Again, I'm not sure this actually means anything, but it certainly might help to confirm what some might already think about given coaches.
You can potentially distill coaching into two things: building a team that produces great stats and figuring out how to maneuver in tight games when neither team has a statistical advantage. The former matters above all (Nick Saban and Steve Kragthorpe, after all, are nearly even on this list ... and on no other list in existence), but if nothing else, this list might help us to understand which teams/coaches are good or bad at the latter.
In that regard, the most interesting names here are the unexpected ones. Paul Chryst, recently hired away from Pitt by Wisconsin, underachieved by about a game per season at Pitt: -1.02 in 2012, +0.28 in 2013, and 2.29 in 2014. Indiana has been -0.7, -1.2, -1.0, and -1.1 under Kevin Wilson. And while Justin Fuente (Memphis) and Joey Jones (South Alabama) have worked miracles in terms of program building, they seem to have seen a few wins leak away through the years.
On the opposite side, Flexbone masters Ken Niumatalolo (+1.2 wins per year) and Paul Johnson (+0.6) are both pretty far up on the overachiever list. That's hard to ignore. And UL-Lafayette's Mark Hudspeth (+1.1), long considered an up-and-comer, ranks highly, too. So does Toledo's Matt Campbell. And close-game crazy wizard Les Miles.
Does this tell us something we didn't know (or at least we didn't have confirmation of) about an aspect of coaching ability? Is it a complete and total coincidence, a product of gray area and noise? I'm only a couple of steps down this road, but this was too fun not to share.
UPDATE: or at least, clarification.
There are 100 different ways to be an overachiever or underachiever. This is one. I'm all for Niumatalolo, Hudspeth, etc. getting love...— Bill Connelly (@SBN_BillC) January 28, 2015
...but I wanted to clarify that "overachiever" in this case has nothing to do with developing talent, making more with less, etc.— Bill Connelly (@SBN_BillC) January 28, 2015
This is specifically about overachieving within the statistical framework of a given game. And it could be all randomness and noise.— Bill Connelly (@SBN_BillC) January 28, 2015
I'm willing to believe certain coaching styles (Navy's, for example) may be able to turn close games into wins more, though.— Bill Connelly (@SBN_BillC) January 28, 2015
But it's pretty damn funny that Nutt teams faltered in this regard every year. And that Ferentz is worth an extra half-loss each year.— Bill Connelly (@SBN_BillC) January 28, 2015