by Sam Waters
Since this was written, the Chargers have played their Week 5 game. Jackie
Battle started, but Mathews received more touches. Mathews still ceded
goal line carries and late-game snaps to third stringer Ronnie Brown.
Prior to his Week 4 matchup with Kansas City, there were hints that San Diego Charger’s starting running back Ryan Mathews’ leash was getting shorter. “Extremely talented players like Ryan get multiple opportunities to improve. What happens to fumblers is, first, they play less… When you get the call to go back on the field and the fumbling continues, then you will be somebody else’s fumbler,” said Chargers general manager A.J. Smith. A few days later, Smith initiated step one of his guide to handling “fumblers”, benching Mathews for plodding “non-fumbler” Jackie Battle. After the game, head coach Norv Turner explained that Battle gave the team the best chance to win.
The notion that Battle actually improves his team’s chances of winning and that all “fumblers” need to be purged from rosters regardless of talent level seems questionable, at best. Is it really possible that a pedestrian runner with great ball security is more valuable than a dynamic back with a fumbling problem? In order to answer this question, first, we will look at what Mathews and Battle each contribute to their team’s likelihood of winning. Then, we will figure out how often an elite runner actually needs to fumble to break even with a replacement-level option.
As we go through the exercise of comparing Mathews and Battle, we will use a statistic called expected points to gauge their respective values. We can compare the amount of expected points from a Mathews-led backfield to those generated by a Battle-led backfield. (In each case we assume that the starter receives 75% of snaps and carries while the backup gets 25%.) I used each player’s 2011 rate statistics as approximations for their hypothetical future performance (yards/carry, targets/snap, yards/target, fumbles/touch). These numbers are by no means great projections for the true talent level of these players, but are still useful as ballpark estimates. The following tables show the approximate production that the Chargers might get with each player as the starter:
When they start Mathews in the above example, the Chargers nets 180 rushing yards and 314 receiving yards, while taking on 4.69 additional fumbles. This result is not surprising- playing Mathews more results in more yards and more fumbles. Now we have to compare the expected points yielded by these results.
Since 16 yards is worth about one expected point, Scenario One’s advantage of 494 total yards equates to 30.9 points. Meanwhile, a fumble is worth a loss of about 1.6  points on average, so Scenario One’s extra 4.69 fumbles cost only about 7.5 points. In fact, since we’re using a small sample of 2011 statistics, we are assuming that Mathews will continue to fumble at a high rate, while Battle will continue to NEVER fumble. It is more likely that both will regress closer to the mean over a larger sample, narrowing the fumbling gap even more between the two players and increasing Mathews’ point advantage.
Mathews’ advantages in rushing and receiving clearly outweigh his fumbling issues, to the tune of 23.4 additional expected points in this example. And while Pro Football Focus grades Battle as a better pass blocker than Mathews, the added threat that Mathews presents probably helps the passing game more than his blocking (which is still league average) hurts it. Let’s be conservative and say that Mathews’ presence improves Rivers’ efficiency by 0.15 yards per attempt. Over 500 attempts, this comes out to 100 yards, which is equivalent to a little over six points. Tack that on to our original total of 23.4 and we get about 30 expected points.
Mathews’ fumbling issues clearly are not as costly as the Chargers seem to believe. Thirty points is a lot to throw away over the course of the season. Since a full win translates to about 36 points (different sources had different numbers on this, but they are all in the mid-30’s), 30 points translates to about 0.8 wins. Remember, this is a somewhat crude estimate. I am not saying that Ryan Mathews is unequivocally worth exactly 0.8 wins more that Jackie Battle. There is a pretty large margin of error on this number, but the Chargers are definitively losing production by playing Battle over Mathews. Assuming that Mathews is a significantly better runner and receiver (no one seems to contest this point), Mathews simply gives his team a much better chance to win.
The large gap in value between Mathews and Battle made me wonder how often a stud running back actually needs to fumble to be replacement level. Let’s say we have a workhorse who is a great runner, receiver, and blocker- think Arian Foster or Maurice Jones-Drew. How often does he have to fumble until you bench him for the likes of a Chris Ogbannaya or Montario Hardesty (RIP, 2011 Browns). For this example, our hypothetical stud will produce 5.0 yards/carry, 200 yards receiving above replacement, and 0.3 additional yards/pass attempt in one season. Our replacement-level option will offer 3.5 yards/carry, 0 yards receiving above replacement, and 0 additional yards/pass attempt in one season. The starter will shoulder 350 of 400 carries and the quarterback will throw the ball 500 times.
If this situation holds, starting your stud nets 450 rushing yards, 200 receiving yards, and 150 indirect passing yards, for a total of 800 yards. This comes out to 50 expected points. Assuming a fumble costs 1.6 points, then your stud needs to fumble over 31 times (50/1.6=31.25) more than his backup in a single season to truly merit a benching. No running back has reached double digit fumbles in a season since Travis Henry in 2002 (Who barely even got there, with 11). Basically, no stud running back is ever going to fumble enough to justify being tossed away as “somebody else’s fumbler.”
Obviously, fumbling is bad. But it’s not as important as some teams make it out to be when making personnel decisions. Guys who fumble might seem risky, but the real risk is repeatedly trotting out an inferior back who costs you yards and points every time he takes the field. Norv, do your team a favor. Send Jackie Battle to the bench. Start your stud.
How did you calculate the value for points per fumble?
I approximated it using the average loss in expected points of a turnover (using a chart of XP at different down/field position/distance situations). It was an estimate though (I only had the values at certain distances)- could probably be plus or minus 0.2 XP/Fumble.
That value also varies depending on if a team is more likely to run in certain game situations w/ either a higher or lower than average expected cost of a turnover. (Eg. Fumbling at the goal line is much worse than fumbling at midfield)