Fortune favors the bold.
-Roman playwright Terence, Phormio, 161 BC
Many of the most memorable moments in college football are risky ones. Think about Statue Left, called at the end of the Boise State-Oklahoma Fiesta Bowl, or Matt Flynn lofting a touchdown pass to Demetrius Byrd with a second left on the clock as LSU edged Auburn in 2007.
Anecdotal evidence is great, but how much does being bold pay off in football?
Surprise Onside Kicks
How you define "bold" is a big part of answering that quesiton. Consider the onside kick.
It's generally a play that only happens late in a game when one team scores but is still facing a defecit. In such situations, onside kicks are successful less than 20 percent of the time. However surprise onside kicks, as defined in that link by Advanced NFL Stats proprietor Brian Burke, are successful over 60 percent of the time.
Doing a surprise onside kick effectively steals a possession from the other team when you do it. Given that (depending on pace) most teams only get somewhere in the 10-12 possessions per game range, that's a big deal. Is it really that bold to execute a strategy that gives your team an extra possession when it succeeds 60% of the time? Put that way, it doesn't sound quite so bold after all. Of course if you do enough suprise onsick kicks, they cease to be as surprising and your success rate will fall.
For now, I'll consider strategies outside the conventional wisdom of the game as "bold." In that sense, a surprise onside kick is a bold play. As long as you pick your spots and use them sparingly, they're absolutely a case where being bold pays off.
Going for it on fourth down
First, a prologue on fourth down conversions. Here are how three basic stats on fourth down conversions correlate with winning percentage. The first two rows come from the NCAA stats archive and span from 2005-10. The last row comes from CFB Stats and covers 2006-10.
|4th Down Att./Game
|4th Down Conv./Game
|4th Down Conv. Pct.
|SEC, All Games
|SEC Conf. Play Only
The correlations for all of I-A aren't as strong as those for just the SEC are, but the trends are the same across each dataset nonetheless.
The number of fourth down conversion attempts per game has a clear negative correlation with winning percentage. That's not surprising given that teams don't tend to go for it on fourth down unless they're behind. Teams that spend a lot of time behind are usually bad, and that extra time being down leads to more fourth down conversion attempts. The conversions per game is still negative but a lot closer to zero, as good teams attempt fewer fourth down conversions but tend to be good at making them. How do we know that? Because conversion percentage is correlated quite positively with winning percentage. The better the team, unsurprisingly, the better chance it has to succeed on fourth down.
Back in 1893, having a good punting game was considered to have a demoralizing effect on the opponent because good punters were rare and field position was precious. The reverence for punting has faded, but its place in the common wisdom of the game has not.
In recent years, there have been plenty of people examining when teams should go for it on fourth down. Economist David Romer probably started the current trend of looking at that issue, but it has been continued by a variety of people including ESPN's Gregg Easterbrook, Burke, MGoBlog's Mathlete, and the authors of Scorecasting. Unsurprisingly, they all reached more or less the same conclusion: teams should go for it on fourth down far more often than they currently do. To date, the only person to take those studies and make a strategy out of them is the oft-profiled Kevin Pulaski, a high school coach in Arkansas.
When it comes down to it, the equation on what to do on fourth down is complex. Its variables include the quality of your offense, the quality of the defense you're facing, and sometimes the skill of your field goal kicker too. You can't make a chart that applies to all teams in all situations. It's at the very least something for teams with good offenses to look into (see the table above), especially when knowing you're not punting in a given region of the field opens up more options with the first three downs of a set.
The topic of what underdogs should do has been explored in great detail at Chris Brown's Smart Football site (see here, here, and the links contained here). Both he and those he links to generally agree on one point: underdogs should pursue high variance strategies.
The overarching principle at work here is regression to the mean. Most events have typical outcomes, described by things like the mean and median. In small numbers of trials, the outcomes can have a mean and median that varies wildly from that natural mean and medium. Over a large number of trials, that difference will smooth out and that set of trials' mean and medium will end up very close to the natural values.
In football terms, an underdog is generally going to want to keep the number of possessions in a game as low as possible. Adding more possessions increases the number of trials, so to speak, making it more likely that the outcome of the game matches what you'd expect. That's not everything an underdog team can do though, which is where increasing variance comes into play.
"Increasing variance" is a fancy way of taking more risk. Being bolder, in other words. It increases your chances of losing badly, but it also increases your chances of winning. As an underdog, you're more likely to lose than win anyway, so what's it matter if it's by 14 or 28?
Springing a surprise onside kick would be an example of being bold, though as noted above, it might not be adding much more risk. So is going for it on fourth down, especially if the other team leaves points on the board by refusing to ever do it. Running trick plays falls in here too, as they tend to be risky but also have large potential payoffs.
Passing has a higher variance in outcome than running does, so underdogs should probably toss it around a bit. Consider 2007 Stanford, a 4-8 team that wasn't very good. It somehow managed to beat 11-2 USC in Los Angeles despite starting Tavita Pritchard, a mildly talented true sophomore quarterback in his first amount of meaningful experience. Jim Harbaugh could have gone with conventional wisdom and tried to protect him with a ton of running. Instead, he called 33 pass plays (passes plus sacks) versus 36 runs, an agressive ratio given his quarterback's status and the ferocity of USC's defense. Pritchard only completed 11 tosses, but they went for 149 yards with a TD and only one INT. It worked. Of course, Stanford threw it a lot more the week before and got pasted 41-3 by an 11-3 Arizona State team. That's the other side of increased variance.
There is plenty of room in football for more boldness. Whether it's merely bold because it defies convention, like surprise onside kicks, or because it truly is risky, like an underdog throwing more than common sense would suggest, you can make reasoned cases for coaches to live closer to the edge. However until coaches start getting fired for punting on fourth down too much, I don't anticipate seeing a lot more boldness across the board.