Which college football coaches consistently produce overachievers or underachievers?

In each FSH college football stat profile, you'll see something called "win expectancy" in each team's "Schedule & Results" section. The definition from my last advanced stats glossary:

It is intended to say "Given your success rates, big plays, field position components, turnovers, etc., you could have expected to win this game X% of the time." It has nothing to do with pre-game projections or opponent adjustments.

For preview purposes, I note when teams strayed pretty far from their postgame win expectations, one way or the other. Some always do.

Five biggest 2017 overachievers, per postgame win expectancy:

1. Army +2.8 (7.2 expected wins, 10 actual wins)
2. Troy +2.2 (8.8 expected wins, 11 actual wins)
3. Akron +2.2 (4.8 expected wins, seven actual wins)
4. Northwestern +2.1 (7.9 expected wins, 10 actual wins)
5. Kansas State +1.9 (6.1 expected wins, eight actual wins)

Five biggest 2017 underachievers, per postgame win expectancy:

1. Miami (Ohio) -2.8 (7.8 expected wins, five actual wins)
2. Baylor -2.1 (3.1 expected wins, one actual win)
3. Ole Miss -1.8 (7.8 expected wins, six actual wins)
4. Arkansas State -1.8 (8.8 expected wins, seven actual wins)
5. Bowling Green -1.5 (3.5 expected wins, two actual wins)

From year to year, this is a sign of randomness. The teams on the extreme ends are all but guaranteed to regress (or progress) toward the mean the next year. Of the 20 biggest underachievers between 2005-16, 16 improved their win percentage, two saw their records stay the same, and only two regressed. Similarly, of the 20 biggest overachievers, 18 saw their win percentage decline the next year and only two improved.

• 2016’s biggest overachievers were Idaho and WVU, which went a combined 19-7 despite only 14.5 expected wins; in 2017, they went 11-14.
• The biggest underachievers, meanwhile, were Notre Dame and Michigan State, which went a combined 7-17 despite 12.2 expected wins. In 2017, they improved to 20-6.

So yeah, at the extreme ends, postgame win expectancy is a good gauge of the randomness involved in this funky sport of ours.

Over time, however, it can tell us a little bit about certain coaches.

As I noted a few years ago, the coaches at the top and bottom of the below list tend to be the ones fans say are particularly good or bad coaches.

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.

With this year’s Northwestern preview on deck, I thought it was a pretty good time to update this list for 2018, as for the fourth time under Pat Fitzgerald, the Wildcats overachieved by at least 1.9 games in terms of postgame win probability.

Over the long-term, only Kansas State’s Bill Snyder and Navy’s Ken Niumatalolo have been overachieved at a higher level than Fitzgerald, who averages 0.9 wins per year above expectation; he has fallen below his expected win total only three times in 12 seasons, and one time was in his first year as a head coach. Over the last four years, his Wildcats have overachieved at a clip of 1.4 extra wins per season.

Is that level sustainable? Probably not, considering that’s about 0.3 more wins per year than even Snyder, the coaching wizard himself, has managed. But Fitzgerald has been a steady overachiever, to the point where we cannot call it simple randomness.

Below is a look at everyone who has been a head coach at least three years since 2005 and what their average difference is per year — actual wins vs. win expectation.

Some quick impressions:

• Your most impressive debuts (i.e. guys who just finished their third seasons and are making their first appearance on this list: Houston’s/Texas’ Tom Herman (plus-1.0 wins per year), CMU’s John Bonamego (plus-0.7), and Troy’s Neal Brown (plus-0.6). Herman has already upgraded from UH to Texas, and Brown will likely make a P5 move of his own pretty soon.
• Your, uh, least impressive debuts: Colorado State’s Mike Bobo (minus-0.7) and Kansas’ David Beaty (minus-0.6). Quite the contrast there. Bobo has been to three bowls in three years, but his three 7-6 finishes have felt increasingly disappointing, while Beaty took a nearly impossible job and, at 3-33, has made the absolute least of it.
• Last year, I began to joke that with his increasingly conservative ways, Urban Meyer has slowly morphed into Woody Hayes at Ohio State. But as frustrating as it is to watch a potentially fun and exciting team throw it into ball-control mode in the second quarter, it’s hard to argue with the results: Meyer overachieved by an average of 0.4 wins per year at Florida but has increased that to 1.0 wins per year in Columbus. Aesthetics are for losers, apparently.
• Dave Christensen’s Wyoming tenure remains one of the strangest on record. Now Arizona State’s OL coach, he went 7-6, 3-9, 8-5, 4-8, and 5-7, alternating between bad (but achieving at where the numbers would expect) and overachieving the entire time.
• The triple option remains a hell of a way to overachieve. Navy’s Niumatalolo (plus-1.1 wins per year) and Georgia Tech’s Paul Johnson (plus-0.5) are both in the 89th percentile or higher, and after some initial underachievement (minus-0.8 per year through three years), Army’s Jeff Monken just delivered a game management masterpiece in 2017, too.
• This is more theoretical than statistically significant, there’s something to be said for being a “system” guy and winning close games. Niu and Johnson know exactly how they’re going to maneuver in key situations, and it seems to reflect with a win total that is almost always higher than the stats would suggest.

I’ve written this before, but here’s your annual reminder:

This is only one aspect of good coaching. Creating a good team and winning easy games is a larger aspect.

Alabama’s Nick Saban is the greatest coach of his generation but is only plus-0.3 wins per year here, the same as Charlotte’s Brad Lambert. Meanwhile, Matt Rhule briefly turned Temple into an AAC powerhouse but is an extreme underachiever here (minus-1.12 wins per year).

Still, I love this list because of how it seems to home in on what we consider good coaching.

The common college football fan or writer would quickly tell you how good a coach Bill Snyder is, for instance, and this illustrates how. He doesn’t sign top-10 classes, and he hasn’t contended for a national title in a while, but Snyder, Niumatalolo, Fitzgerald, and others have consistently milked the most possible gain out of their win opportunities.