As analytics have become more widely used and more accepted in the NFL (and sports in general), conventional wisdom has come into question more. Old philosophies like running to set up the pass, offense sells tickets but defense wins games and sayings of that ilk have largely been either debunked or have forced spectators to at least reconsider what those things mean.

With that, you have to wonder what a win made is of.

Which statistics actually correlate the most with winning games?

This week, we took 14 different team statistics and compared them to win totals over the past five seasons to figure out where the correlation lies between certain stats and wins. Some results are obvious, while others may be shocking to some.

First, though, the methodology: These 14 statistics weren’t chosen for any particular reason other than the fact they’re constantly referenced, discussed or are at least widely and commonly known. There are others that could have reasonably been included. Additionally, the five-year window isn’t a huge sample size, but it’s enough to illustrate recent trends. We have a table below with the 14 stats chosen for this exercise with their correlation coefficients in regard to wins.

Two other important things to note are the inspiration and the meaning of this research. It was inspired by a 2011 post from Chase Stuart of Football Perspective that dealt with essentially the same topic but only used passing stats (and also a different range of dates). As for the meaning, this is not supposed to be a be-all, end-all way to determine the best ways to win football games. It’s also not predictive, so you can’t use this data as presented to predict that statistical leaders in certain categories will have repeat success in coming years. It just explains how stats have correlated with wins in the recent past.

Now, the data:

Statistic Correlation Coefficient
Scoring offense .86158
Turnover margin .82475
Scoring defense -.7582
Turnovers -.74743
Total yards .59017
Yards/pass attempt .53198
Yards/play .52661
Sacks .47723
Sacks allowed -.44931
Pass yards .39788
Total defense -.36873
Rush yards .27289
Penalties -.21301
Yards/carry .12084

This list is ordered from strongest correlation to weakest. The closer a number is to 1 or -1, the stronger the correlation, while a number closer to zero means it’s a weaker correlation.

The strongest correlation is the most obvious, that scoring more points has usually meant more wins. Some of the results are more surprising though, like the fact that in the past five years turnover margin has correlated more to winning games than the amount of points your defense allows has.

Here are some of the biggest takeaways from this information. Remember, this is not predictive, and correlation does not imply causation. There are just observances from the data presented.

Running has had minimal effect on winning

Of these 14 stats, the two based on the running game were among the three weakest correlations with wins. Surprisingly to some, potentially, the rate stat of yards per carry seems to have been even less meaningful than the raw counting stat of total yards on the ground. By comparison, passing the ball doesn’t strongly correlate to wins, however there’s clearly a stronger relationship than running.

This takeaway is the most significant for where the NFL is now with running back contracts being the talk of the preseason. Ezekiel Elliott ended his holdout on Sept. 4 when he signed the richest running back contract in league history with the Dallas Cowboys, while Melvin Gordon of the Los Angeles Chargers is still holding out. Elliott is the Cowboys’ best player and has proven to be among the best running backs in the league, if he isn’t the very best (which he probably is). He deserves a payday based on production, but the argument there is that it doesn’t make sense to focus so much on a running back when their main job (running the ball) has been minimally valuable recently.

For the record, there are other places where the relative insignificance of running the ball has been laid out much more clearly and effectively, notably this Twitter thread from Scott Barrett of Pro Football Focus, as well as a 2018 piece by J.J. Zachariason of Numberfire. With our data in mind, the point isn’t to discredit an individual player’s ability or disagree with them getting paid, but it’s to point out that there’s an argument against making the running game (or running back passing game, as those other sources point out) a focal point of your offense.

Efficiency through the air is important

Yards per carry may not have correlated to wins in our data, but yards per pass did. In fact, the relationship was stronger for yards per pass than yards per play (which is understandable given the weak relationship with running). Unsurprisingly, teams that move the ball in chunks of yardage through the air often win more. Over the past five years the top five teams in yards per attempt were Atlanta, Pittsburgh, New England, the Los Angeles Chargers and New Orleans; those five teams won better than 60 percent of their games during that time frame. All of those teams averaged over seven yards per attempt from 2014-18, with other teams in that category including the Kansas City Chiefs (third-most wins in the NFL over those five years) and the Seattle Seahawks (most wins in the NFC), among others.

Turnovers matter as much as anything

As was mentioned earlier, it’s somewhat shocking how strongly turnovers have correlated with wins since 2014, with turnover-based stats mattering nearly equal amounts as points-based ones. If you dive a little deeper, though, the numbers aren’t completely befuddling. In a 2014 writeup by Bill Connelly, then of SB Nation and now of ESPN, he wrote that in college football a turnover is worth approximately five points. Obviously, there are a ton of variables here, but the general principle remains that based on things such as losing/gaining possession and the change in field position, there is a tangible point value lost or gained when a turnover happens.

Needing to avoid turnovers, then, is probably the most important and correct coach cliché.

Let’s take a look at which teams have the best turnover margin since 2014. The Chiefs have the best margin and have the third most wins, the Patriots are second and have the most wins, and the Seahawks have the best turnover margin and win total in the NFC during our window. As for the worst? The Jets were a whopping -39 in the turnover battle, the worst in the league, while the Browns weren’t much better at -36 (and they had an embarrassing 18 wins over five years). The trend is obvious in terms of giveaway numbers, too. The Patriots had the fewest in the league, followed by the Seahawks, Chiefs, Packers and Vikings. Those five teams won over 64 percent of their games from 2014-18.

Don’t give up the ball and you’ll be in good shape.

Penalties don’t correlate, but they matter

Penalties accrued have the second-weakest relationship of any of the stats in question, but they’re a prime example of something to not take for granted based on a chart. Anyone who watches the sport knows how disastrous an untimely penalty can be. This information simply shows that the raw number of them has not correlated strongly with win totals lately. The Oakland Raiders lead all teams in penalties since 2014 with 626 of them, and they’re tied for the seventh fewest wins in the league. Second in total penalties is Seattle with 623 of them, but the Seahawks have the best record in the NFC in our timeframe.

Similar relationships between sacks and passing yards

As important as it could be to move the ball through the air in large increments, it’s also important to not give away yardage on pass plays. Sacks and sacks allowed both correlated more with wins recently than gross passing yards did, and they’re relatively similar to the yards per attempt coefficient, too. Interestingly, passing yards, attempts, sack yards and total sacks are the factors in net yards per attempt, a stat which Chase Stuart found in his aforementioned research as being as predictive as any quarterback stat he measured.

None of this is perfect science, and considering correlation, again, doesn’t imply causation, you could potentially reject any of the conclusions one might draw from this data. But it’s still worth pointing out what the most common trends with winning and losing teams have been in recent memory, especially given the hot-button topics in the NFL right now. The easiest conclusions to come away with are that you should throw far more often than run and that turnovers should be avoided at all costs. One of those is conventional wisdom, while one is one side of an ongoing, heated football argument, albeit a side gaining more popularity with fans and the media.