Is Pat Fitzgerald a brilliant in-game coach, or just getting lucky? (Part 2)

Welcome back. In part 1 of this ongoing series, I did a bunch of boring math and showed that Pat Fitzgerald's remarkable success in close games could simply be the result of luck.

Now that we've gotten the math out of the way, let's take a closer look at the individual games during Pat Fitzgerald's career and see if we can find any trends that might explain his 70% winning percentage in close games. The prevailing theory in the comments of part 1 was that Northwestern has played down to their competition against weak opponents, and that all the close wins actually reflected negatively on Fitz, since he should have been winning more easily against the worst teams on the schedule.

To see if there's any validity to that theory, we'll break Northwestern's games into three categories (favored by 5 or more, 5 or more point underdogs, and the middle ground of +4.5 to -4.5 spreads) and see how Fitzgerald has fared. Then we'll further analyze the definition of "close game" to see if meaningless late touchdowns are inflating Fitzgerald's close game record.

Note: Most of the point spread data is taken from the Northwestern page on covers.com. For the few games where no point spread was given, I used the Sagarin ratings.

NU's record when favored by 5 or more points under Pat Fitzgerald

Close games: 7-3

Not close: 9-0

Overall: 16-3

Clearly, there's something to the theory that Northwestern plays down to inferior competition; over half the games where NU was a sizable favorite have been decided by 8 points or less, and 4 of the 9 blowouts were against FCS opponents. That's the bad news. But the good news is that NU usually wins those games, winning 70% of the close games and 100% of the blowouts.

There have definitely been a few games that were embarrassingly close, most notably in 2009 against Eastern Michigan when NU needed a last-second field goal to beat a horrendous EMU team that would finish 0-12. Games like that help Fitz's close game record a little bit, but he's been involved in 30 close games in 5 years and only 10 were as a sizable favorite, so clearly that's far from the only factor.Also, Fitz's 16-3 overall record in these games is quite good, as upsets are part of college football and you can't expect him to go undefeated in games he should win. A rate of about one upset every other year* is par for the course.

* The 3 upset losses were Duke in 2007, Indiana in 2008, and Purdue in 2010. Oddly, NU is just 1-2 against Big Ten teams when 5+ point favorites. The one win? Minnesota in 2007 (in double overtime).

NU's record as 4.5 point favorites to 4.5 point underdogs under Pat Fitzgerald

Close games: 6-1

Not close: 3-4

Overall: 9-5

What's most surprising here is how few NU games fit in this category. 30 of NU's 63 games under Fitz have been decided by 8 points or less, yet just 14 of the 63 had close point spreads.

With so few games to choose from, it's hard to figure out much in the way of patterns. 6 out of 7 close wins is great, but there are obvious sample size issues. Half those wins came last season, against Vanderbilt, Indiana, and Minnesota. The one close loss was in 2009 against Syracuse, after a late Mike Kafka interception set up the Orange for a game-winning field goal.

Embarrassingly, one of those blowout losses came in 2006 to FCS New Hampshire, by a 34-17 score. That New Hampshire team was pretty good by FCS standards; Sagarin's predictor says NU was just a 1 point favorite in the game and he ranks that UNH team as better than all but 3 MAC teams. Still, it's rather humbling when an FCS team beats you and it clearly wasn't a fluke.

NU's record as 5+ point underdogs under Pat Fitzgerald

Close games: 8-5

Not close: 1*-16

Overall: 9-21

Here is where Fitz really earns his stripes in close games. We discussed in part 1 how you'd expect the favored team to be over .500 in close games, and that proved true for games where Northwestern was favored, but even as an underdog Fitz still manages to win more than he loses. Also, 3 of the 5 losses came in bowl games, meaning that against Big Ten opponents, Fitz is an impressive 8-2.

There's no denying NU has benefited from some good luck in those wins, be it injuries to key players on the opposing team (2008 Iowa, 2009 Iowa) or some unforced turnovers at opportune times (2008 Iowa, 2009 Purdue), but Fitz has to get a lot of the credit as well. He's managed to win 30% of the games he shouldn't, while he's only lost 15% (3 out of 19) of the games he should have won.

* The one win? 21-7 over Iowa in 2006. NU was 20.5 point underdogs that game, the biggest upset of the Fitz era** per point spread and biggest since the 1995 team won at Notre Dame as 27 point underdogs.

** The #2 upset under Fitz was the 2007 win at Michigan State. 3 through 5? 2009 Iowa, 2010 Iowa, and 2008 Iowa respectively. Excuse me while I go cackle maniacally.

The bad news here is that Fitz has a ton of experience in games he shouldn't win: in almost half of his games as NU head coach his team has been 5 or more point underdogs. That's not a model for sustained success, especially considering the weak non-conference schedule NU has played doesn't account for any of those 30 games. 3 of the 30 are the 3 bowl games, meaning that the other 27 all came during Big Ten play.

Think about that for a second: Northwestern has been 5 or more point underdogs in 27 out of 40 Big Ten games under Pat Fitzgerald, and they've been 5 or more point favorites just 3 times (bizarrely, he wins 33% of the time in each case: 9-18 as a 5+ point underdog, 1-2 as a 5+ point favorite.) That makes his 18-22 career record in Big Ten games all the more impressive (or lucky, depending on your level of skepticism).

While a lot of interesting facts came out of this analysis, there hasn't been much light shed on how Fitz has won so many close games. It doesn't matter the caliber of the opponent, as he's well over .500 against better teams, equal teams, and worse teams. But as I looked at each game individually, I realized that the definition being used for "close game" is somewhat flawed. Just because the final margin of victory was 8 or less, that doesn't mean the game was close; we've all seen games where a team was comfortably ahead the whole way before surrendering a meaningless touchdown in garbage time. Conversely, we've all seen games that were close the whole way before a late score in the final minutes made the final margin deceptively wide.

So I looked at all the NU games over the last 5 years and came up with 12 games that I think need to be re-categorized.

Deceptively Lopsided Games

Nevada 2006: NU trails by 3 with just over 2 minutes left, and gets the ball back with a chance to tie or take the lead. Instead, Mike Kafka throws an interception for a touchdown, and NU loses by 10.

Illinois 2006: NU leads by 1 at halftime, and by 8 through nearly the entire second half. A Joel Howells field goal with 2 minutes left gives NU an 11 point win

Michigan 2007: NU leads by 9 at halftime and by 2 after 3 quarters, but Michigan gets 2 touchdowns in the 4th quarter, the last with 4 minutes left, to win by 12.

Eastern Michigan 2007: In front of about 7 people at Ford Field, NU struggles (as usual) with a weak EMU team, leading by 6 late in the 4th quarter, before a 73 yard bomb from C.J. Bacher to Kim Thompson makes the final margin 13.

Purdue 2007: NU leads 17-14 after 3, but Purdue storms back with 3 4th quarter touchdowns for a deceptive 18 point win. Hilariously, the NUsports.com box score doesn't have the 4th quarter stats.

Iowa 2007: In Fitz's only career loss to Iowa (also the only time he wasn't a sizable underdog vs Iowa), NU leads 17-14 after 3, but Iowa takes the lead with 8 minutes left, then intercepts Bacher and gets another TD with just under 4 minutes left. The final margin is a deceptive 11.

2009 Minnesota:  NU leads 24-21 after 3, but Minnesota takes the lead in the 4th quarter on a touchdown, then forces a Kafka turnover in the final minutes and scores another touchdown with 1:43 left to win by 11.

2010 Penn State, NU leads 21-0 early, then falls behind 35-21 in the 4th quarter. Mike Trumpy drops a TD pass on 4th and goal late in the 4th quarter that would have brought this game into one score territory.

Deceptively Close Games

Illinois 2009: NU leads 21-3 early in the 4th, but Illinois gets two late touchdowns to get within 5. They did get a chance to win with a minute left but a Sherrick McManis INT sealed the game for NU.

Central Michigan 2010: NU leads by 17 early in the 4th quarter, but the Chippewas chip away (sorry) and get 2 late touchdowns, the last with 1:11 left, to make the final margin 5.

Indiana 2010: NU leads by 10 late, but a long IU TD with 44 seconds left makes the final margin 3.

Texas Tech 2010: Tech leads the entire game, and by 14 late in the 4th, before a Jordan Mabin INT return gets NU within 7. NU got a chance to tie in the final minute but ran out of time.

So if you add in the 8 deceptively lopsided games (in which NU went 2-6) and take out the 4 deceptively close games (in which NU went 3-1), Fitz's close game record drops from an incredible 21-9 to a much less impressive 20-14. That's still a good record, but it's far from spectacular and much more easily explained by good luck.

As for why NU has let so many teams put them away in the final minutes, it's hard to say. A common theme is NU blowing a 4th quarter lead and being unable to respond, but that's only happened 4 times in 5 years. It's also a bit surprising that NU hasn't gotten any late touchdowns vs prevent defenses to get within one score, since their spread offense is perfectly designed for such situations. Pass defense with a big lead in the second half is a problem though; we saw it last year against Central Michigan and Indiana, and this was never more of a problem than it was against Michigan State in 2006.

So what has all this analysis told us about Fitz's talents in close games? Not a whole lot. All we really learned was that Fitz's 21-9 career record in one-score games was deceiving, and that if you want his record in games that were actually close, it should be 20-14. That discrepancy can be explained nearly entirely by pure luck, but he's still well over the .500 record we'd expected, so we can't write it all off as luck. Part 3, which appears next week, will look at Fitz's coaching tendencies to finally answer the question posed in the title: is he a brilliant in-game coach, or just getting lucky?