website statistics

For whatever reason, the Patriots do have exceptional ball security, especially for an outdoors team. And I mean exceptional. 

I was intrigued by a link sent to me via Twitter at Sharp Football Analysis, a handicapping site. The article demonstrated that NE's ball security was an outlier to the tune of several standard deviations. The charts are convincing, and the implication is that NE benefitted from under-inflated balls is unmistakable. But I wasn't sure how much stock to put in the numbers for a couple reasons. One is that they were so extraordinary they seemed unlikely to be true. And second, the analysis used plays per fumble as its basis rather than fumbles per play. For low-rate events, like fumbles, outlier cases can appear exaggerated depending on which way you look at them.

I started by verifying the numbers. They may vary slightly because I'm including playoff games, but overall the numbers look correct. The article looked at both fumbles lost and at total fumbles. I'd submit that if we're talking about ball security, fumbles lost is just total fumbles + noise, so let's just focus on total fumbles. Here are the offensive plays per fumble for each team in each season since 2010:

NE ranks third over that period. Very good, but nothing out of the ordinary. You'd expect teams with good QBs and good offenses to have fewer strip-sacks. But the article linked above makes a good point: Many teams that play indoors are concentrated at the top of the list. Let's see how the table looks if we exclude dome teams.

Whoa. In this case NE is at the top of the list, and the next best team is a distant second. Notice how the second team (BLT) through the second to last team (PHI) have rates that are within 1 or 2 plays of each other. NE, however, is better than the next best team by 20 plays per fumble.

You might notice that 2013 was a bit of a down year for NE ball security. That's partly due to the week 12 game against DEN in insanely frigid conditions. NE had 6 fumbles and Denver had 5. Extremely cold temperatures are associated with high fumble rates--They're about 35% more frequent than usual in the coldest games. But if we threw out that game, NE would have a 60 play-per-fumble rate. Not the best that season, but still better than any other outdoor team's 5-year average.

To get a better sense of just how NE stands out apart from the crowd, here are the team averages in chart form.

As I mentioned, sometimes outliers can be exaggerated depending on which way you look at a rate. Should we look at fumbles per play or plays per fumble? Mathematically, both measures contain the same information. I prefer fumbles per play, because I conceive of each play having an independent probability of fumble, and fumbles per play is an unbiased estimator of that probability. So this is what it looks like from the other direction. Not much difference.

I'm not sticking my neck out here and saying this is evidence of anything. It's fair to say that Belichick emphasizes ball security emphatically, and is quick to bench players who drop the ball. Everyone will have their own opinion anyway. I'll just say, either way, it's worth looking at. If it's a result of an unfair advantage, that's interesting. If it's the result of good coaching, that's just as interesting.

One last note. Hey WAS! What the hell? 37 plays between fumbles?

Addendum: @yonghang suggested a comparison of NE and opponent fumble rates. This would hold for environment. In other words, climate, indoors, outdoors, etc. would not skew the results.

Addendum 2: @brian30tw pointed out that NE's big improvement in fumble rate occurred in '07, precisely when the NFL's rule allowing visiting teams to bring their own balls went into effect. Other teams didn't have such good fortune. 

Leave your comments

Post comment as a guest


Comments (88)

Load Previous Comments
  • Guest - Travis

    @Daelix, you misread the BAL to PHI spacing part incorrectly. He was saying that between each team BAL->CAR, CAR->SD, SD->GB, etc. there is a spacing of 1 or 2.

  • Guest - Juan R

    The next domed teams after Atlanta and New Orleans are Houston and Minnesota which are domed teams and are far below New England as well and essentially along with the "normal" range. The remaining domed teams are even lower. So please explain how you justify removing Atlanta and New Orleans? I respectfully am asking you if you are being intentionally misleading with the idea that most people who read this will not see that connection, or is there something that I missed that I should consider? I ask because as it stands it seems like an intellectually dishonest move.

  • Guest - Andy

    Some people are arguing that the low fumble rate can be attributed to Brady, who is very good at protecting the ball when sacked, and doesn’t get sacked that much, anyway. But at least eight qualified QBs had a lower fumble rate than Brady on pass plays (PA + sacks): Dalton, Orton, Bridgewater, Ryan, Smith, Manning, Flacco and Hoyer. In contrast, only one team had fewer fumbles on running plays, Minnesota with just one. NE and NYG were tied for second fewest with two. NE was one of ten teams with no fumbles by a TE, while it was in the middle of the pack with 3 fumbles by WR. So clearly NE’s low fumble rate is not just due to Brady.

    Since the home team gets to supply the balls, though, I think we need to look at stats of just home games. I don’t have those at the moment, but the offensive players lost a total of five fumbles this year, and only one at home, so that admittedly very small sample size suggests they fumbled overall much less at home.

  • Guest - Andy

    Sorry, should have said NE lost no fumbles at home, while they lost five on the road, at least by the offense. So they probably fumbled much less overall at home, where they got to supply the balls.

  • Guest - eag97a

    I think people should check out this article in 538 for a more balanced view on the subject;

  • Guest - Espo74

    Perhaps a subset to the study could be looking at players who have gone from Patriots to other teams (or vice versa) and their fumble ratio per touch. Blount is a decent study in this although all his stats are post the '07 rule modification.

  • Guest - Greg

    I question tossing the Dome Teams outright. Can you re-run the data, limiting the comparison to the 16+ games per season of the Pats vs. their opponent of the day. (eg. for 2014, start with Pats vs Miami = 3 fumbles each, ? total plays each. Then on to the week 2 Vikings game, etc.)

    This would give you less data, but hopefully make up for in in that it (theoretically) filters out much of the noise caused by weather and field conditions, etc.

  • Guest - Pat

    Just to make things more clear:

    New England fumbles, 2001-2006: 148 fumbles, 6195 plays, 59 by Tom Brady
    New England fumbles, 2007-2014 (no 2010): 120 fumbles, 7713 plays, 37 by Tom Brady

    So, from 2007-2014:

    Change in New England total fumbles/play: -0.8% fumbles/play
    Change in New England non-Brady fumbles/play: -0.3% fumbles/play

    The -0.3% fumbles/play drop from non-Brady players is for the most part consistent with the rest of the league. The drop is all Brady.

  • Guest - Vince

    I wish people would read the article before commenting. First, this is a review of someone else's study - Burke didn't create the parameters, but does offer alternative ones. Second, you can safely ignore the 538 article on this, as it is heavily biased, conveniently ignoring the primary point of Sharp's initial hypothesis, i.e. the timing of the dramatic change of NE's rates after the rule change in 2007. Third, Burke's article does address the dome issues, particularly in the article's addenda.

    The point of Sharp and Burke is that the change in rate from pre-2007 to post-2006 is extremely improbable statistically, without some "outside" causation.

  • Guest - Tom

    Minnesota did not play home games at a Dome in 2014.