clock menu more-arrow no yes mobile

Filed under:

Why Do We Bench Running Backs for Fumbling When We’re Happy to Let Quarterbacks Throw the Ball Away?

Comparing the risk-reward ratios of different offensive skill positions 

Kansas City Chiefs and Washington Football Team Photo by John McDonnell/The Washington Post via Getty Images

There are a few basic truths that anyone who knows anything about football understands. One of them is that running the ball is the most secure way to move the ball down the field and control time of possession. Passing is more efficient at gaining yards, but carries a much higher risk of turning the ball over.

In the five years since Kirk Cousins left town, the Washington football franchise has struggled to establish a respectable passing offense. The few times when the offense has looked halfway decent during this period, it has been because the team has relied on a run-oriented, ball-control offense. This has led a contingent on Hogs Haven to suggest that that should become the team’s identity. I can’t help but think that they are rationalizing being fans of a team that has struggled to field a complete offense, but they might have a point.

Recently, some Commanders’ fans got a shock when their favorite running back, Antonio Gibson, was sidelined after fumbling the ball in the second series of the preseason opener against the Panthers. It wasn’t the first time Gibson dropped the ball. He led all running backs in fumbling, with six drops in 2021. I would imagine that Ron Rivera’s crew had become concerned that his recently developed lack of ball security was getting to the point that it was undermining the chief advantage of rushing the football.

One fan who had an issue with their reasoning was long-time Hogs Haven commenter, LASkin, who wrote an excellent debut article pointing out that fumbling by running backs is so infrequent and variable that it is difficult to be sure if a one season increase in fumbles is a trend, or just random variation. That’s a really good point, and I’m not going to argue with it.

This article was inspired by an aside in LASkin’s piece, in which he pointed out the very different treatment that running backs get for fumbling at most a few times per season, compared to quarterbacks, who drop the ball and throw interceptions much more frequently.

Presumably, quarterbacks get away with being turnover machines because they are much more productive than other positions. Or maybe it’s just because TV audiences like to see a lot of passes. To figure out if quarterbacks are really worth the extra turnover risk – and they turn it over much more than anyone else – I used the Risk-Reward Ratio concept, which I introduced in my recent article on Gibson’s ball control issues.

Washington Football Team v Carolina Panthers Photo by Grant Halverson/Getty Images

Which Offensive Positions Are the Biggest Ball Security Risks?

To start this off, let’s have a look at how the different offensive skill positions compare in terms of ball security. I just wanted to look at major contributors, so the sample started with all the running backs, wide receivers and tight ends who logged more than 400 yards from scrimmage. That might be a little unfair to the tight ends; maybe not, because it means we are just looking at the most productive tight ends.

I included all the fullbacks who had at least one yard from scrimmage last season because they don’t get the same opportunities as the more glamorous positions.

And last of all, I included each team’s primary starting QB. For teams that did QB by committee, I used the one with the most starts.

The first graph plots average turnover risk for each position. This is calculated as fumbles per touch for RB, WR, TE and FB. Quarterbacks have another way to give the ball away, so the turnover measure for them is fumbles + interceptions. Some readers might object that I am comparing apples to oranges, since an interception is a turnover, while a fumble is just an opportunity for a turnover. Ideally, I would have used interceptable balls to maintain equivalence with fumbles, or, less ideally, fumbles lost, to maintain equivalence to interceptions. But I don’t have easy access to those statistics, so I made do with what I had.

This graph shows that quarterbacks have more than 3.5 times greater ball security risk than the next closest skill position, wide receivers. They achieve this by throwing interceptions, as well as by fumbling much more than any other position.

In case anyone is worried about the apples to oranges comparison of fumbles to interceptions, I should point out that it results in underestimating the turnover risk of QBs relative to other positions. If I had access to statistics on interceptable balls, then the turnover risk for QBs would be even higher than what is shown here, while the other positions would be unaffected. Bottom line, quarterbacks are even greater ball security risks than these data indicate.

Remember that Antonio Gibson was “benched” after fumbling the ball six times in 310 touches in 2021, which gave him a fumble risk of 0.019 fumbles/touch. The average starting QB had 7.6 fumbles in 2021, and a fumble risk of 0.014 fumbles/touch, just slightly below Gibson. But they also threw an average of 11 interceptions. Only two of Gibson’s fumbles were lost, but all of those interceptions were turnovers.

Gibson had a pretty high fumble risk for a running back at 0.019 fumbles per touch. It was more than twice the position average of 0.0080. Running backs were slightly less fumble prone than wide receivers (0.0086), but were much looser with the ball than tight ends (0.005).

It should come as little surprise that the kings of ball security were the fullbacks. Fullbacks are big, burly men with strong arms and no-nonsense running styles. They don’t wave the ball around when they run like halfbacks and scatbacks. The NFL’s 10 active fullbacks handled the ball 128 times in 2021 without a single fumble. Giving the ball to a fullback is like putting your money in Fort Knox. Admittedly, the sample size is smaller than the other positions. Still, zero fumbles is zero fumbles.

Are Quarterbacks Worth the Risk?

What we’ve seen so far suggests that quarterbacks have a much greater risk of turning the ball over than any other skill position. They must make up for that with their outsized offensive production, or else teams would just run the ball all the time. Let’s see if that’s true.

In my article on Antonio Gibson’s fumbling issues, I introduced a simple metric to quantify the balance of turnover risk against offensive production. The Risk-Reward Ratio is the ratio of turnover risk to a measure of production, such as total yardage or touchdowns. It can also be thought of as turnover risk, adjusted for production.

A player with a high Risk-Reward Ratio turns over the ball too much relative to their on-field production and should be avoided. A player with a low Risk-Reward Ratio produces enough to make up for any turnovers they may commit. Quite a few running backs didn’t fumble at all in 2021, earning them a Risk-Reward Ratio of 0.

To compare other skill positions with quarterbacks, I needed to add passing game variables to the Risk-Reward Ratio calculations. The revised calculations are as follows:

Risk = (Fumbles + Interceptions)/Touch

Risk Reward Ratioyardage = Risk/(Total Yards from Scrimmage + Passing Yards)

Risk Reward Ratioscoring = Risk/(Rushing TDs + Receiving TDs + Passing TDs)

Now, let’s see how the different skill positions stack up in terms of Risk-Reward Ratios. We’ll start with turnover risk adjusted for total yardage production:

This is a type of statistical graph called a box and whisker plot, which compares the distributions of Risk-Reward Ratios (RRR) for every single player in the sample at each position. Individual player RRR values are shown as dots. The box illustrates the middle two quartiles of the distribution, the line through the box is the median, the X is the mean, and the whiskers are kind of complicated. If the distributions don’t have outliers, they show the minimum and maximum values, but if there are outliers (all of these distributions have them), they illustrate 1.5 times the interquartile range. Just think of the whiskers as the total spread of the main parts of the distributions.

Just as expected, adjusting turnover risk by total yardage production has dropped the QBs’ relative risk down into the range of the other positions. It doesn’t make them less risky than other positions, as you might have expected given how much production they account for. It’s actually still a bit higher than wide receivers. But if you think about it, you can’t really get the ball to a wide receiver very easily without going through a quarterback; Tight ends too for that matter. The really meaningful comparison here is between QBs on the one hand and the positions that primarily run the ball on the other, FB and RB.

Overall, running backs still have a small advantage over quarterbacks in terms of production-adjusted ball security risk. However, there is so much overlap between the ball-security risks of the RBs and QBs that the question comes down to who you have playing each position. Three running backs (Dontrell Hilliard, Nyheim Hines, Ronald Jones) had higher adjusted ball security risks than the riskiest QB (Justin Fields). Fullbacks remain risk free, because they don’t fumble.

That could have significant implications for offensive game planning. But before we get too excited, let’s see if the same pattern holds up when we adjust turnover risk by scoring production:

Correcting turnover risk by scoring also drops the QBs’ relative risk into the range of the other offensive skill positions. Once again, it doesn’t show them as being better than the other positions, which might have been expected, given how much more productive they are. The distribution of Risk-Reward Ratios for QBs and RBs are overlapping, meaning that there are plenty of QBs with lower adjusted turnover risk than a lot of RBs and a fair number of RBs with lower adjusted risk than most of the QBs.

But remember, using interceptions rather than interceptable balls as the passing game risk variable tends to underestimate their total turnover risk relative to other positions. The true relative turnover risk for QBs is actually somewhat higher than this data shows.

Once again fullbacks remain the safest bet on offense. In 2021, fullbacks handled the ball 128 times, producing 6 touchdowns, without a single fumble.

Summary of Findings:

1. Quarterbacks put the ball at risk of turnover far more than any other offensive skill position.

2. However, quarterbacks are so much more productive than other positions that their adjusted turnover risk (Risk-Reward Ratios) is comparable to that of running backs.

3. The distributions of Risk-Reward Ratios for QBs and RBs are overlapping. This means that some teams’ running backs have lower adjusted turnover risks than their quarterbacks and other teams’ quarterbacks will have lower adjusted turnover risks than their running backs.

4. Fullbacks are more secure ball handlers than the other “skill” positions. The data beg the question, how much skill is there really in dropping the ball or throwing it to the other team?

Washington Football Team v San Francisco 49ers Photo by Christian Petersen/Getty Images

What Does it Mean for NFL Offenses in General, and the WFT in Particular?

NFL offenses are driven by the passing game. If your team has an average quarterback or better, then it makes sense to lean heavily on the passing attack. But what do you do if your quarterback is significantly worse than average, as has been the case in Washington for most of the past three decades?

That is where point #3, above, becomes highly relevant. If your quarterback throws a high ratio of interceptions to touchdowns, or has low total production relative to turnovers, it might make sense to turn to the running game. But that’s only going to help if you are one of the teams whose running back is a significantly lower turnover risk relative to production than the quarterback.

In 2021, Washington QB Taylor Heinicke ranked 25th out of 32 starting QBs in TD/INT ratio, and had the 10th highest turnover Risk-Reward Ratio adjusted for total yardage, and 11th highest Risk-Reward Ratio adjusted for touchdown production. Remember, high Risk-Reward Ratios are bad. Those metrics place him in the bottom third of starting quarterbacks. Clearly Washington had a need to rely on the running game, which is probably why lead back Antonio Gibson had the fourth most touches among all running backs in the NFL.

This begs the question, was Washington one of the teams whose running back was a better Risk-Reward proposition than their quarterback? The answer, unfortunately, was no. Due to his newly developed fumbling habit in 2021, Antonio Gibson had a higher yardage-adjusted turnover risk than Heinicke (Risk-Reward Ratios: Gibson 0.0000145 vs Heinicke 0.0000106) and the two were practically tied on touchdown-adjusted turnover risk (Risk-Reward Ratios: Gibson 0.00193 vs Heinicke 0.00189), with the slight advantage going to Heinicke.

In other words, Washington needed to rely on the running game to make up for subpar QB play, but unfortunately, Antonio Gibson was not a secure enough ball handler to make that work out.

While I sincerely doubt that Ron Rivera’s team were driven by analytics, they seem to have arrived at the same conclusion as I did. The team needed an upgrade at QB, so after the season ended they paid a hefty price to acquire Carson Wentz from the Colts, after trying a few other options unsuccessfully. Then they drafted a running back who was known for good ball security and moved Gibson to a flex role to make better use of his talents.

The one thing they didn’t do, which I am sure will come back to haunt them, is to incorporate a fullback into the offense. As I have shown here, when you absolutely need to move the ball a yard or two, and you can’t afford a turnover, a fullback is your best bet. As an added bonus, they are also useful in the backfield to protect your investment at quarterback when your team faces players like Micah Parsons twice a year.

Acknowledgement: Thanks to James Dorsett for editing.


The WFT made 550 pass attempts to 477 rush attempts (1.15 ratio) in 2021. The Eagles with Wentz had a 1.38 pass to rush ratio. What is the most likely scenario for the Commanders in 2022?

This poll is closed

  • 12%
    570 pass/540 rush (1.06 ratio)
    (18 votes)
  • 38%
    600 pass/500 rush (1.2 ratio)
    (56 votes)
  • 31%
    635 pass/470 rush (1.35 ratio)
    (45 votes)
  • 17%
    650 pass/430 rush (1.5 ratio)
    (25 votes)
144 votes total Vote Now