As any of you who know me are quite aware, I have been reciting a mantra for several years now, one adopted from Fan in Thick and Thin, one of the BTB members whose statistical acumen and resultant analysis I most respect. Essentially, it goes like this: close games are a subject to luck; in large sample sizes, all teams' winning percentages in such games, regardless of overall record, are roughly .500. Consequently, what distinguishes good and bad teams is the paucity of such contests in which they are involved; the best teams win more games comfortably, while the worst are blown out more frequently. By extension - and this is the main point for today - teams that play a lot of close games should expect their overall records to hover around the .500 mark.
Skeptical, you say? Let's untangle this a bit. A man named Michael Mauboussin, an investment strategist who thinks deeply about all kind of ideas that are tangentially related to the investment world, has written books exploring everything from psychological biases and how we think to the science of complex systems. His newest book is The Success Equation: Untangling Skill and Luck in Business, Sports, and Investing. As the title suggests, he's really interested in understanding and unraveling the relationship between skill and luck in all its complexity; indeed, the book pursues a set of ideas first introduced in an earlier publication, Think Twice: Harnessing the Power of Counterintuition.
In the more recent book, Mauboussin develops a spectrum with luck (roulette) and skill (chess) at opposite ends, and located the various major sports between these poles. The most skill-based sport, he argues, is NBA basketball; the most random is NHL Hockey. The second "luckiest" sport is NFL football. Why? Mauboussin offers three reasons: the number of players and how often they are on the field (basketball players log far more minutes that hockey or football players, thus skill plays a larger part); sample size (there are so few games in a football season, and so many fewer possessions than in, say, basketball, that there are fewer opportunities to bleach out luck and randomness); the way the game is scored (a team can be very successful and have nothing on the scoreboard to register that success - or play very poorly but get a couple of lucky bounces and have a lead).
Because of these factors, goofy plays - tipped passes, fumbles in heavy traffic, long touchdown passes wherein a defensive back slips after having solid coverage - have more value. Unlike sports with a lot of games (baseball) or many more possessions (basketball) or fewer players on the field, court. or rink, the weird plays in football factor more heavily in the final outcome. With this in mind, what distinguished a good team is not that it has "heart" and wins close games, but that it's good enough to blow out a fair amount of opponents, thus limiting the number of games it can lose due to a bad bounce, questionable penalty or blocked field goal at the buzzer.
To substantiate this, I'd like to begin by looking at the 90s Cowboys, specifically at how many close games Dallas played in their "dynasty" years:
The Cowboys played between five and seven "close games" per season during that spectacular four-year run. For all that team's general dominance, they were a mere 12-11 - and were never better than 4-2 - in such contests. Overall, they had one of the most impressive four-year stretches in league history - emerging victorious in 81 percent of their games - but only 52 % of all close games.
The 90's Cowboys were one of the best teams the league has ever seen, yet for all their excellence, they didn't fare any better in close game than myriad other (and significantly weaker) squads. In 1994, all four losses were by margins of a touchdown or less; one of these was a week 15 home loss to Cleveland in which the Cowboys trailed 19-14 with mere seconds remaining on the clock. The always-reliable Jay Novacek caught a Troy Aikman slant at the two and headed into the end zone...only to slip on the wet Texas Stadium turf as time ran out. The difference between winning and losing on that particular Saturday afternoon was the fact that it has rained overnight.
On one hand, the above table suggests that good teams tend to outscore their opponents by large margins. But the Novacek example perhaps allows us to look at this another way: they were good because they managed to avoid close games. In an NFL landscape in which a close game can turn on a flukey, chance moment - a fumble, a controversial officials' ruling, a deflected pass, a slip on the turf - the best way to be a winning team is to avoid being in a situation wherein luck can determine the outcome. Every season between 1992-95, Dallas enjoyed five to seven comfortable wins, games in which they outplayed the vagaries of the game so that a bad bounce or dropped pass couldn't hurt them.
I revisit this history so that we can use it as a basis of comparison to Cowboys teams of more recent vintage. Below, I've engaged in the same exercise for each year since 2005, the first season that the team's current core first played together:
A couple of striking numbers that jump out. The first of these is how dominant the 2006 Cowboys could be; they won six games by two or more touchdowns (but, strangely, also lost four games by such wide margins).
The more important number, however, is the Cowboys' record in close games over this period: 30-32. To my mind, this reinforces the notion that, because football is the most "random" of professional sports in terms of chance occurrences that contribute to winning or losing, teams will tend to hover around .500 in close games, regardless of overall winning percentage. Sure, there are year-to-year exceptions like 2010, when Dallas was an unfortunate 3-8 in games decided by a touchdown or less, but that is an aberration, a reversion to the mean which "corrects" a luckier year (2007) in which they went 4-1 in such contests.
The other number that I find worth noting is the staggering number of tight games the team has played under the Garrett administration. As noted above, the 1992-95 Cowboys played between five and seven such games per season. Notice how this accords to the 2006-08 Cowboys, who played four, five, and six such games. In 2009, this number ballooned to eight, then, in subsequent years, to eleven, nine and, thus far in 2012, eight. From 2006-08, 37.5% of Cowboys games were decided by a touchdown or less; since 2010, 65.1% of their games have been "close."
More troubling is that Dallas' number of comfortable wins has suffered a precipitous decline. As much as the Cowboys struggled last season, they still managed a comfortable victory over Seattle and blowout wins versus the Rams and Bills; the only win of more than 7 points this season was marked by a once-in-a-generation scoring flurry wherein the Cowboys tallied 21 fourth-quarter points. Other than that? A series of ulcer-inducing too-close contests which, like the home games against New York and Cleveland, can go either way until literally the final second - and can be decided by the width of a receiver's finger or a football that just scoots inside a the right upright.
The takeaway here is this: the Cowboys don't need to become "more clutch" or develop more character or "learn to win" close games. The best teams in the history of the league won roughly the same percentage of tight contests as the 2011-12 Cowboys. What they do have to do is to get better, in (apologies in advance) "all three phases." The current iteration of the offense isn't explosive enough, especially in the passing game, to establish quick leads; the defense doesn't generate the turnovers that lead to cheap points and short fields - an area in which the special teams could certainly make some welcome contributions.
I'll repeat what I wrote above: the best way to be a winning team is to avoid being in a situation wherein luck can determine the outcome. In far too many games in the past three seasons this team has been involved in contests wherein a single flukey moment can decide the outcome. And, I'm not sure they're good enough to change this any time soon.