With most teams having played four games to date, a quarter of the NFL regular season is in the books. This then is a good moment to open up the playbooks and box scores and look at what has been happening across the league on a macro scale.
Over the next couple of days we'll dig a little deeper into different aspects of the NFL box scores. Today we'll start by looking at some of the stats that have correlated highly with winning in the first four weeks of this season. And if you think you're in for the Same Old, Same Old, you may be in for a big surprise after the jump.
But before we jump into the numbers, a few words on correlation and causation. One of the most common errors we commit when we look at stats is that we confuse correlation and causation. In principle, both are easy to keep apart, but just because two things occur together does not mean that one caused the other, even if it seems to make sense.
One of the stats we'll look at later is time of possession (TOP). If you are a proponent of a run-first, pass-second philosophy, you're probably convinced that the higher your own TOP, the higher your likelihood of winning the game. You would therefore be assuming a causation between TOP and winning, when TOP is actually little more than a by-product of many other factors that result in a win. You could even argue that there is a reverse causation between winning and TOP, in that teams with a lead to protect will resort to the more time-consuming run game more often than not.
Similarly, if you're a proponent of the pass-first school of thought, you probably think that having Sir Pass-A-Lot as your QB is a good thing, and you tend to assume that a lot of yards will lead to a lot of wins. Again, this would be assuming a causation between passing yards and winning, when the passing yards may be nothing more than the lack of a decent ground game.
The data below is all about correlation, and not necessarily about causation. What I did is I looked at all the box scores for the 62 games played in the first quarter of the 2010 season so far and looked at which stats had a higher correlation with wins. All fairly straightforward. Here's where the league stands after four weeks:
The Ho-Humm Stats
These are stats that have little to no correlation with game outcomes so far this season. You may be slightly more likely to win if you're better than your opponent in one of these stats, but with the small sample size, the winning percentages for these stats could very well be random noise.
1. Passing yards. Of the 62 teams that have won their games, only 28 (45%) passed for more yards than their opponent in that game. This is a somewhat surprising result, and certainly is different from the trend we've seen in recent years towards more passing success. But it is what it is. When you pass more than your opponent, you're just as likely to win as you are to lose. In 2010. In the first four weeks.
2. First downs. 32 of the 62 winners this season (52%) recorded more first downs than their opponent.
3. Sacks. 33 of 62 winners (53%) allowed fewer sacks than their opponent in the game - and there I was preaching the gospel of the sack as a drive killer. In the end, there are probably not enough sacks in an average game to influence the game outcome consistently.
4. Penalty yards. 35 of the 62 (56%) winning teams accrued fewer penalty yards than their opponents. Now, if you're going to get 18 penalties for 152 yards (Packers in week three loss vs Bears) that might influence the game outcome, but overall no big correlation here. Move along please.
5. Yards per pass attempt. 35 of 62 for 56%. I used a slightly modified YPA calculation [(Pass Yards – Sack Yards)/(Pass Attempts + Sacks allowed] for this. It seems that this particular stat, which has been a pretty good indicator of success in previous years, has lost some of its punch this year. Is this just a momentary aberration or could this be a more significant change. My vote is for the former.
The "Winning Stats"
1. Time of possession. 40 out of 62 (65%). Now we're getting somewhere in terms of correlation, but as I mentioned earlier beware of cause and effect on this stat.
2. Rushing yards. 43 out of 62 (69%). On the face of it, teams that run more appear to win more, but this may be the clearest case of where cause and effect are often confused. I may look at this in more detail in a future post, but suffice it to say that in general, winning teams generate more rushing yards mostly because their RBs are running out the clock at the end of wins, not because they are dominating early in games. You run when you win, not win when you run. For more on this specific correlation and causation, read Football Outsiders classic reflection on genuflection, "On bended Knee".
3. TO ratio. 43 out of 62 (69%). Now we're getting somewhere. "Win the turnover battle and you win the game", how often have you heard that one? And because a number of games ended with a tie in the turnover battle, the opposite is even truer (as with many of these stats): Lose the turnover battle and you lose the game. 43 of the 51 teams (84%) with a negative turnover ratio in the game lost that same game.
4. ANPY/A. 44 out of 62 (71%). I have extolled the virtues of ANPY/A for quite some time now, and am a little disappointed that this stat did not come out on top. But that a passing efficiency measure should rank close to the top on this list does not shock me at all.
5. Passer Rating. In 46 of the 62 games (74%) the QB with the better passer rating in the game walked away with the win. Good old QB Rating beating out some of these newfangled gadget stats, that's got to please some of our more, uhm ... veteran members.
6. Field Position. 47 of 62 (76%). If you are as obsessed with drive charts and the kicking and punting game as I am, this should come as no surprise to you. For the first four weeks, winning the field position battle (measured as the sum of the starting field position for each drive in yards) is the stat most highly correlated to winning in the league that I could find - outside of points scored of course.
Field position was also one of Bill Parcells pet themes. Dave transcribed a Parcells press conference in 2006 where Parcells said the following: "McBriar is a field position weapon, he's netting 40 yards a kick. He picks up hidden yards that can lead to points". According to Parcells, 100 yards of "hidden yardage" or field position are worth seven points (others may use the same formula, he's just the one I associate this with).
Let's look at the "hidden" yardage for the Cowboys game this season: +82 versus the Redskins in a game that should not have been lost. -172 against Chicago in a game that looks closer than it actually was and -10 against Houston. If you were wondering why I put up all these posts about our kicking and punting game and keep clamoring for a better return game, now you know why.
Surprised by the results? I'll look into some of these stats in more detail over the coming days, and I'll probably repeat this exercise every four weeks to see where the game is heading.