If you are tired of deflate gate, go ahead and click the back button now .

A little discussion on Stats

First, I want to explain a bit about my novice approach to some numbers. A while back, I was the lead for a website overhaul designed to improve online sales of the company I worked for at the time. We knew prior to the change that about 1.2% of visitors made a purchase when they visited the website. We launched the new website at 5am. During the first hour, we had 3 sales--good for us for 5am in the morning. The CEO was hopeful. The 6-o'clock hour we had 2 sales, which was less than our average for that hour. The CEO panicked and told us to roll it back. Being the good employee that I was, I refused. The following hour we had 11 sales -- the most ever for that hour. By the end of the day, we had over 100 sales from about 5000 visits, beating our best ever previous day by over 30.  Eventually, we found that the new site converted at over 2%, so we nearly doubled the number of sales per visit to our website.

The moral of the story? Like any statistician will tell you, looking at results like 3/100, 2/200, 11/400, etc doesn't tell you much. The numbers are too small still to have much confidence in them. This is why saying the Seahawks are the kings of PED cheating with 5/1100 compared to someone else who has a result of 4/1000 or even 0/1050 doesn't tell you anything yet. When all 32 teams are evenly distributed between 0 - 5 or 6 over 1000+ tests, we don't really know a thing about whether or not any team is doing anything different than any other team. Maybe the Seahawks are "cheating" more than everyone else. Could be. But we wouldn't know that from those numbers. Likewise, anyone who points to the second half of the NE/IND game from last week to say that the non-deflated balls prove nothing was really wrong is missing the boat as well (where the Pats probably ran 30-50 plays).

With all that said, I've only taken a couple of stats classes a loooong time ago.  I'm sure there are many here who could get much more technical and also explain things better in layman's terms.


Asterisk

In the comments that follow, I am not attempting to say the numbers "prove" that the Patriots have been deflating footballs since 2007, etc, etc. There are, however, some interesting numbers. Other teams ought to try and figure them out.  If the Patriots are cheating, it should be stopped.  If they've figured something else out, it should be mimicked.

Suppose there is a robbery at the corner market.  A car is seen speeding through the neighborhood, and someone sees the license plate.  It could be that they've spotted the getaway vehicle.  It could also be that they've just seen a husband rushing his wife who is in labor to the hospital (or any one of a number of other things).  

Similar kind of thing with these numbers.  They are interesting.  They may or may not be related to under-inflated footballs.  Here is your fair warning, though: I will point out how the numbers appear to make a case against some other theories while continuing to leave the possibility of under-inflated footballs open.

If you are a Pats fan, I hope you can see that I've attempted to not offend while still looking at the stats and some possible explanations for them.  You may find errors in my thinking or in the numbers.  I'd welcome any such discoveries along with any additional thoughtful commentary.


Fumble Numbers

My curiosity with this topic started with this article that analyzed a few numbers about the Patriots ball security over the last few years:

http://www.sharpfootballanalysis.com/blog/2015/the-new-england-patriots-prevention-of-fumbles-is-nearly-impossible

However, there were some flaws with it, and it raised more questions than it answered.  After perusing other forums, etc, I've compiled several bits of information.  Most of these numbers are based on pass attempts, rushes, receptions, and kick/punt returns.  To my knowledge, a few corner-case plays such as when a member of the defense intercepts a pass and subsequently fumbles the ball are not included.

1. Since 2007, the Patriots have fumbled 1 in 94 of such opportunities [122 / 11,513].  Prior to 2007 (counting only players who also played on the Patriots for at least one season after 2007), they fumbled 1 in 53 [110/5,858].  Tremendous improvement.  According to a stats program I use, this offers a 99% confidence that the results are statistically significant.

2. Tom Brady, himself, fumbled 1 in 111 after 2007 [37/4093] while fumbling 1 in 52 prior to 2007 [59/3061] .  Again, amazing improvement. (99% confidence that it is statistically significant)

Realizing we won't "prove" any theory here, still, what could possibly account for this?

Theory A: The Patriots improved their offensive line to protect Brady better, and Brady improved his awareness to cut down on fumbles.  Additionally, the post 2007 numbers above include many players who weren't on the team prior to 2007.  Maybe the new players were brought in specifically because they are better at protecting the ball.  If this were the case, the performance of other players that were on the team both before 2007 and after 2007 should remain mostly constant.

3. 8 WRs/RBs/TEs were on the team both before and after 2006/2007.  Prior to 2007, they fumbled 1 in 57 [48/2757].  After 2007, they fumbled 1 in 131 [9/1178].  These numbers are a little small for my liking with a larger margin of error, but they still point to a likely significant improvement, and they wouldn't have been impacted by the OLine or Brady's individual improvement. (Software says statistically significant)

Theory B: The Patriots taught a new technique to their team for better taking care of the football.  If this were the case, one would expect players to continue their improved ball security after leaving for other teams.

4. For WRs/RBs/TEs who made both meaningful contributions to the Patriots after 2007 as well as a team following the Patriots, those players fumbled 1 in 107 with the Patriots [21/2239], and 1 in 80 after leaving [21/1680].  The difference appears meaningful and would indicate that these players somehow lost the technique to a degree after leaving (if they were taught a special technique).  These numbers do not provide the same level of confidence as some of the other numbers above.  A larger sample size is needed, though it definitely is trending in the direction of there being a real difference once players leave the Patriots.

Theory C: The Patriots have brought in players since 2007 that are better at ball security.

5. For WRs/RBs/TEs who have joined since 2007 and played meaningful amounts both with their prior team and with the Patriots, those players fumbled 1 in 78 times prior to joining NE [90/7009].  Those same players have improved to 1 in 85 with the Patriots [28/2392].  A larger sample size is definitely needed to make any kind of conclusions here.

SUMMARY

First, some of these numbers don't have the volume I'd like.  A statistician can get more fine-grained calculating the differences if they'd like.  Don't take the above numbers as exact percentages/ratios/etc.

Second, I make no warranty that the numbers are exactly correct, though I have reason to believe that are likely quite accurate.

Now, Tom Brady and his 8 WRs/RBs/TEs that were on the team both before and after 2007 cut their fumbles nearly in half in 2007.  They somehow just flipped a switch.  These numbers are statistically significant.  This would be a big deal to me if I was the coach of another team--trying to figure out how they accomplished this. The thing that seems suspicious here is that, similar to free-throw shooting, it is hard to get players to change their stripes when it comes to fumbling--especially an entire team.

"Technique" or "renewed emphasis" as an explanation seems less likely to me since the numbers are trending towards players regressing after leaving the Patriots and since it would likely be observable and something that would pass on to other teams.  It also doesn't seem likely to me because the existing players showed such HUGE improvement almost overnight from 2006 to 2007.  Yes, Belicheck emphasizes ball security.  But, so have many other coaches.  Holmgren famously traded away Ahman Green to Green Bay for a late-round draft pick solely because of fumbles.  Green went on to be a pro-bowl RB in Green Bay.  I find it doubtful that new or renewed emphasis in NE would have had such an impact with existing players and yet couldn't be duplicated elsewhere.

The Patriots in recent years may be attempting to specifically bring in players that are better at ball security.  I applaud NE if they are strategically attempting to do this.  Fumbling 1 in 80ish touches as the new players have done prior to coming to NE is quite good from what I've seen.

What about under-inflated footballs being a key reason for the Patriots improved ball security?  I see nothing in these numbers that would make a case against that argument.  The two most eye-catching observations are the overnight change in 2007 with existing players, and how NE has performed compared to the rest of the league in ball security since 2007.  In *my opinion*, it seems like there is a very reasonable chance that under-inflated (or otherwise manipulated) footballs is a contributing factor.  If it is, it is likely one of at least a couple of contributing factors to the Patriots overall improvement since 2007.

Even if the manipulated balls are a key reason, it is possible that it was unintentional or technically within the letter of the law--similar to their WR as an OL trick play from a couple of weeks ago.  In their press conferences, however, that aren't owning up to anything being "technically" correct about their ball preparation.

The Patriots have the right priorities.  The question is, what are their methods?

----------------------------------------------------------------------------

I know you're asking it: Do I have too much time on my hands?

#1, Yes.  #2, I like numbers and facts, even if I'm not very good with them.  #3, I've been sick and pretty much stuck with nothing else I can do other than watch TV or use my laptop this weekend.

If you've read this far.....I warned you in the subject line that it would be long.  I made no claims that it would be worth your while.