≡ Menu

Thoughts on the value of a fumble vs. an interception

In the late ’80s, The Hidden Game of Football determined that an interception was worth -45 yards and a lost fumble was worth -50 yards. Why was a fumble five yards worse than a pick? That’s because Carroll, Thorn, and Palmer found that, on average, the team gained possession via the turnover was five yards closer to their opponent’s end zone when that turnover was a fumble.

Makes sense, but is that still true? Courtesy of Mike Kania of Pro-Football-Reference, here are some data on turnovers since 1999:

  • Ignoring interceptions returned for touchdowns, the team recording the interception loses about 4.41 yards of field position, on average, on each interception. So let’s assume the Patriots are playing the Jets, the Patriots have the ball at their own 40, and New England throws an interception. On average, the Jets will (ignoring pick sixes) have 1st and 10 at the Patriots 44.4-yard line on the next play.
  • If, instead, the Jets gained possession via a fumble, New York would, on average, start on the Patriots 39.2-yard line. That’s because following a fumble by an offense that is not returned for a touchdown, the line of scrimmage moves about 0.8 yards closer to the offense’s end zone.
  • In other words, teams gain about 5.2 yards of field position when recovering a fumble rather than an interception. That’s kind of remarkable, considering it matches the results found from researchers in the ’80s. However…
  • We still have to consider turnovers that are returned for touchdowns.  Roughly 10.7% of interceptions were returned for touchdowns during this period, compared to only 7.9% of recovered fumbles. Remember, interceptions are now much more likely to be returned for touchdowns than they were in the mid-’80s.

Thirty years ago, the penalty was 45 yards for an interception and 50 yards for a lost fumble.  We haven’t shown today whether those numbers in the abstract were correct, but the five yard relative difference still seems supported by current data, with one notable exception.  But as more interceptions are returned for touchdowns1, interceptions are becoming about as bad for offenses as lost fumbles.

  1. I’ll note that fumbles are also being returned for touchdowns at higher rates — that’s probably worth its own post — but it is not increasing at the same rate. []
  • Dan

    For fumbles, what if you looked at the spot where the ball was fumbled instead of the original line of scrimmage? (Or a regression including both.) A sack-fumble is pretty different from a fumble by a WR 20 yards downfield.

    You could also break things down by the position of the player who fumbled. I suspect there are significant differences between fumbles by WRs, RBs, and QBs, and once you’ve estimated the average cost for each position you could apply that to box score data without having to dig into the play-by-play.

  • To make the example more realistic, it really should have been Jets turning the ball over to the Pats. Also important to note is the 100% TD return rate on ass-induced fumbles.

  • Anonymous

    I’d think that there’s a lot more variation in where a team gets the ball on an interception than there is with a fumble. An interception on a 50 yard downfield pass with no return is a lot different than a pick six. However, most fumbles are recovered around the line of scrimmage with little return.