Monday, May 18, 2015


New England Patriots Fumble More Often When Playing for Other Teams

28Jan15 Update:  The data below does factor in all fumbles (not just fumbles lost) but also includes punt return and kick return fumbles.  After adjusting for the return fumbles, the Patriots still fumble more often (the point of the article, and indeed, its title) but the rate is lower.  For example, players who left the Patriots and went to other teams fumbled 23% more on those other teams, including 38% more for the players with 300+ touches, when adjusting for the return data.  More commentary on this issue can be found HERE.
By Warren Sharp
Last Thursday’s groundbreaking article on the impossibly real fumble prevention statistics by the New England Patriots was certainly a perfect lesson on the power of social media and sending a critical story “viral”.  I tried to take a very unassuming, impartial look into a highly controversial topic, based not on opinion, but on statistics.  What caught the attention of many was that it was uniquely different from the other “theory based” pieces on this topic, and contained undeniable concrete evidence that whatever was happening in New England since 2007 was more than just ridiculously abnormal.  It was absolutely not a random fluctuation in the data and was extremely unlikely to be a mere coincidence.
After watching primetime TV and some of the Sunday morning national news shows, in addition to the Bill Belichick press conference from Saturday, it is clear that many in the media still believe this story is about a single-game incident.  And because that particular game was not close on the score board, the issue of ball inflation is not a concern.  I cannot deny the fact that I too was skeptical of this issue to begin with, and initially believed it to be immaterial – just another irrelevant story being hyped to create ratings.  Which is why I decided to investigate myself, and let the data show any abnormalities, should they exist.
The problem with dismissing this as an immaterial, single-game incident is that it ignores reality.  It ignores the abundance of data which shows a massive anonomly exists in New England which potentially could be tied back to the crux of this investigation:  proper ball inflation (or lack thereof).    Belichick spent an inordinate amount of time discussing a “simulation” they performed of their game day football preparation operation.  Regardless of what they specifically do to the footballs, it certainly would be remarkable if their techniques are so vastly different from the other 31 teams that only the Patriots footballs see a drop of 2 psi while all other teams remain in the legal range.  But as the data appears to indicate, since 2007, the footballs that the Patriots use on offense (or something else the Patriots do offensively) are completely dissimilar to the other 31 teams.  Not just by a slight margin, but so massively that it is an injustice to possibly cite the persistent, obvious variance over 8 years could be attributed to “climatic conditions”, “atmospheric conditions”, a “rubbing process”, or an “equilibrium getting reached”.
There were many follow-up thoughts and questions raised by the masses who gravitated to the piece across not just the United States, but from around the world.  In a way, it was a great example of how international the NFL has become.  Hundreds of thousands people are interested in this story and concerned about the findings I presented.  Certainly, the NFL is justified in spending ample time to research as much as they can to determine the full extent of the situation.  But more than encouraged by the sheer volume (which crashed multiple servers along the way), the reaction to the initial study was an affirmation of the power of data analysis and statistics.   The appreciation for crunching vast amounts of data to obtain a very simple, specific and precise conclusion which is unadulterated by a loud, distracting voice was clearly evidenced by how willing, enthusiastically and expeditiously the story was shared.  The feedback was tremendous, but of all the questions raised, the one most frequently asked (by literally hundreds of people) was: “What if you take a look at individual players when playing for the Patriots and when playing on other teams?  How does the data change, and what would it tell us?”  It was a terrific suggestion.
So I ran the analysis.  I’ll immediately share the results below, before diving into exactly how we arrive at these conclusions:
  1. Patriots players fumbled SIGNIFICANTLY more often when playing on other NFL teams than when playing for the Patriots:
    • Individual players who played on New England during the 2007-14 span and on other teams fumbled 46% less often ON the Patriots as compared to on their other teams (98 touches/fumble on NE, 67 on other teams).
  2. The most utilized of the Patriots players fumbled even more frequently when paying for other NFL teams:
    • The players who played the MOST often for the Patriots during this span fumbled the ball TWICE as frequently on other teams as they did on the Patriots (107 touches/fumble on NE, 53 on other teams).
  3. Learning ball possession skills in New England did NOT transfer to other NFL teams after players left:
    • Individual players who played on the Patriots fumbled 88% more often after LEAVING the Patriots as they did when playing on the Patriots (105 touches/fumble on NE, 56 after NE on other teams).
  4. In fact, the opposite was true – players were MORE secure carrying the football before even playing for the Patriots than they were after leaving the Patriots:
    • Individual players who played on the Patriots fumbled 25% less frequently before joining New England as they did after playing for New England and then leaving (70 touches/fumble before NE, 56 after NE).
Methodology.  I attacked this study it as I always try to do:  logically, impartially and always looking for “something more” even after finding data which offers a strong conclusion.  I looked at all player statistics for the Patriots between 2007-2014.  I started in 2007 because it appears from both my initial “fumble” analysis as well as my “weather” analysis that things suddenly changed in 2007 when looking at long term Patriots data. The 2007 season is when they skyrocket into the realm of other-worldly, where they stand alone from any team in NFL history.
Quick sidebar: Without even knowing what happened in 2007, I can tell from the data something changed for New England which did not change for the other 31 NFL teams. But the stars apparently are aligning on a NFL rule change which Tom Brady (and Peyton Manning) lobbied in favor of, and the NFL agreed to change policies. Brady wanted the NFL to let EVERY team provide its OWN footballs to use on offense. Prior to that year, the HOME team provided ALL the footballs, meaning the home quarterback selected the footballs the ROAD quarterback would play with on offense.
Brady’s quote at the time, when pushing for the change was: “The thing is, every quarterback likes it a little bit different. Some like them blown up a little bit more, some like them a little more thin, some like them a little more new, some like them really broken in.”
Regardless of exactly what started to happen in New England in 2007, the Patriots offense fumbled significantly less often, which is why I looked at their individual player data beginning in 2007.
Comparing the individual Patriots offensive players statistics against statistics for those same exact players when they played on other teams certainly is compelling, as the table below illustrates:
(click to enlarge)
As a whole, these players fumbled once every 98 plays when donning a Patriots uniform, but once every 67 plays when playing for any other NFL team. That is a 46% improvement when they played for New England from 2007 thru 2014.
Clearly there are many players at the bottom of the list (which is sorted by total touches when playing for New England) who carried the ball very infrequently (and never fumbled). To eliminate some of the noise created by these less utilized players, I decided to take a subset of data which looked ONLY at players with over 300 touches as a Patriot from 2007-2014 who also played on other teams before or after. They are the 5 players at the top of the list.
As the chart above indicates, these players fumbled once per 107 touches in New England, but once very 53 touches when playing on another team. That is an improvement of OVER 100%!
Obviously, that data is shocking.  These same individuals fumbled twice as often when playing for other teams.  When trying to explain the Patriots refusal to fumble, we no longer have the argument:  “maybe the Patriots draft/acquire players who don’t fumble.”  We are now at the point where we can say these individuals, for some reason, fumbled TWICE AS OFTEN on other teams as they did in New England.
The next logical question would be:  “What about players who LEFT the Patriots and THEN played for other teams?  Maybe these players learned a ball protection skill or a special secret trick while IN NEW ENGLAND to reduce their fumbles.  Surely they would take this with them elsewhere and their numbers AFTER leaving the Patriots would be good.”
So I performed that analysis.  I looked ONLY at players who left New England at some point after 2007, and I grabbed only their stats when playing for other teams AFTER playing in New England.  The caution here is we are slowly decreasing the data size, and thus adding variance into the numbers.  The chart is below:
(click to enlarge)
As you can see, these players fumbled once every 105 touches in New England, which is similar to the numbers we saw above, and is not surprising.  But when playing for other teams, they fumbled once every 56 touches!  That’s an 88% increase.  And it comes AFTER they recorded incredible numbers for the Patriots.
When trying to explain why the Patriots refuse to fumble, we no longer have the argument:  “there is a secret manner in which to hold the football or special coaching techniques which get entrusted to the Patriots during this 2007-14 time frame.”  These very same individuals, after leaving New England, see their fumbles/touch increase at an alarming rate.
Obviously Brandon Tate stands out in this analysis.  But the only players of these 10 to fumble LESS frequently after playing for the Patriots were Wes Welker and Ben Watson.  All of the other 8 players on this list fumbled MORE FREQUENTLY after playing for the Patriots.
Those of you paying close attention also likely noticed one final, remarkable takeaway from this analysis:
The entire collection of total players in this study actually fumbled LESS frequently BEFORE joining the Patriots as compared to when they left the Patriots.  That flies in the face of the thought that “fumblers” joined the Patriots, were straightened out, and then left as reformed ball controllers.  The first two lines on this next chart are pulled directly from the top two charts, and the third line is the difference:
(click to enlarge)
It is clear by these numbers, that these 18 players who qualified for this analysis fumbled once every 70 touches before joining the Patriots and once every 56 touches after leaving the Patriots. 
When trying to explain why the Patriots refuse to fumble, we no longer have the argument:  “the Patriots took fumble prone players, taught them better techniques while in New England to improve their ball possession, and they left better than when they were brought in.”  These players were actually 25% BETTER at securing the football before they ever even worked with the Patriots than they were after playing in New England and then joining another team!
There clearly are a number of takeaways to sum up this portion of analysis, and to restate them from the top of this article:
  1. Patriots players fumbled SIGNIFICANTLY more often when playing on other NFL teams than when playing for the Patriots:
    • Individual players who played on New England during the 2007-14 span and on other teams fumbled 46% less often ON the Patriots as compared to on their other teams (98 touches/fumble on NE, 67 on other teams).
  2. The most utilized of the Patriots players fumbled even more frequently when paying for other NFL teams:
    • The players who played the MOST often for the Patriots during this span fumbled the ball TWICE as frequently on other teams as they did on the Patriots (107 touches/fumble on NE, 53 on other teams).
  3. Learning ball possession skills in New England did NOT transfer to other NFL teams after players left:
    • Individual players who played on the Patriots fumbled 88% more often after LEAVING the Patriots as they did when playing on the Patriots (105 touches/fumble on NE, 56 after NE on other teams).
  4. In fact, the opposite was true – players were MORE secure carrying the football before even playing for the Patriots than they were after leaving the Patriots:
    • Individual players who played on the Patriots fumbled 25% less frequently before joining New England as they did after playing for New England and then leaving (70 touches/fumble before NE, 56 after NE).
Quick sidebar #2:  Another argument which is frequently made in defense of the Patriots abnormal fumble rate:  “Bill Belichick emphasizes possessing the football and cuts players who fumble.”  First, let me say its absurd to assume that other NFL coaches are “fine” with a player who fumbles frequently, and only Bill Belichick stresses reducing turnovers.  But even if we believe only Belichick cares, and he abnormally and quickly hooks any player guilty of this enormous transgression, let’s look at a few players on his team now and historically:
  • In 2013, WR Julian Edelman fumbled 6 times.  No non-QB for Bill Belichick ever fumbled more in one season.  Yet Edelman is the key WR still playing for New England.  In fact, Edelman followed that up with 5 fumbles this season (and twice more in the playoffs).  In total, 5 of these fumbles came on offense (not returning kicks).  But we don’t hear any discussion of benching Edelman for the Super Bowl.  In fact, Edelman’s fumble rate was one fumble every 18 touches in 2013 and one every 20 this year!  If you look at the numbers above, you will see how terrible those numbers are for a Patriot.  No receiver with 50+ catches under Belichick has ever fumbled more frequently than Edelman did this season.
  • RB Stevan Ridley is the poster boy of the “fumbler” for the Patriots, who fumbled 8 times between 2012 and 2013.  But his fumble rate was still only one fumble per 61 touches.  In 2013, he improved his fumble rate from once every 47 touches to once every 74 touches, a significant improvement of 57%.  Compare Ridley to one of Belichick’s most beloved players, Kevin Faulk, and you will see Faulk’s fumble rate was worse:
  • In 2005 RB Kevin Faulk fumbled a total of 3 times despite having only 80 offensive touches (he only returned 4 total kicks that year).  That rate of one fumble per 27 touches is obviously more than twice as bad as Ridley’s two-year average and almost 3 times worse than Ridley’s 2013 season.  Yet Faulk played his entire career for the Patriots, all the way thru until 2011, and is widely considered one of the Patriots most valuable offensive assets over that time.
  • In 2004, the Patriots acquired RB Corey Dillon from the Bengals.  Dillon recorded 5 fumbles on 360 touches, a rate of one fumble per 72 touches.  That rate was worse than Ridley’s 2013 season.  In fact, it was the worst fumble rate for any Patriot under Bill Belichick who had at least 200 rushes.  Yet Dillon finished his career 3 years later in 2006, still with the Patriots.
  • In 2013, RB LeGarrette Blount fumbled 3 times on 155 touches for the Patriots.  That was a rate of 1 fumble every 52 touches.  Very similar to the 2013 fumble rate of Ridley (once every 47 touches).  Yet Belichick brought Blount back to the Patriots in the middle of this season after he was cut from the Pittsburgh Steelers, and is now the starting RB for the Patriots in next week’s Super Bowl.  That is despite Blount having the 3rd worst fumble rate of any Patriots RB under Belichick with at least 100 carries in a season.
  • In 2006, TE Ben Watson had 49 touches and 3 fumbles, a terrible rate of 16 touches per fumble.  It was the worst fumble rate of any receiver under Belichick with over 10 receptions.  Yet Belichick kept Watson in New England for the next 3 seasons, where he played in 42 of the 48 regular season games, starting in well over half the games.
  • Despite fumbling once every 24 touches (an extremely bad rate) in 2008, WR Randy Moss caught 20% more passes for the Patriots in 2009.
Certainly Bill Belichick places emphasis on ball possession, as does every NFL head coach.  And more goes into decisions on who to keep, play and start than just fumbles.  Anecdotally it does seem like Belichick pulls players quickly who fumble often, but the fact is, if you look at the data, he treats other players who have similar (or worse) fumble rates differently.  Ultimately, its likely about the total package the player provides for Belichick rather than simply his fumble-aversion statistic alone.  This far less technical case study certainly appears to show that there is no set “limit” to the tolerance Bill Belichick has for fumbles from his players.  So I think we can dismiss the defense that “Bill Belichick yanks players who fumble often, and then cuts them.”  While he may do that to certain players, clearly it is not a consistent position he has taken historically across-the-board with all players, many of which are mentioned above.
Once again, we are left with highly compelling evidence which certainly refutes a number of arguments made in defense of the statistical anomaly that is the New England Patriots from 2007-2014.  While the data cannot tell us precisely WHAT the Patriots were doing to prevent fumbles, keep in mind, the data was the foremost clue that something was occurring, and told us precisely WHEN it began (all of which I introduced in my article last Thursday).  The additional data compiled from this latest research project is now beginning to shed light on what the Patriots were NOT doing to prevent fumbles.  To prevent fumbles for the New England Patriots:  they were not amassing a collection of highly fumble-adverse players.  They were not teaching a special ball-carrying technique which was entrusted to their players.  And they were not targeting fumble-prone players and reforming them in New England to permanently cure them of their bad fumbling habits.
_______________________________________________________________
Warren Sharp of sharpfootballanalysis.com is an industry pioneer at the forefront of incorporating advanced analytics and metrics into football analysis. A licensed Professional Engineer by trade, Warren applies the same critical thought process and problem solving techniques into his passion, football. After spending years constructing, testing and perfecting computer models written to understand the critical elements to win NFL football games, Warren’s quantitative analytics are used in private consulting work, and elements of which are publicly shared on sharpfootballanalysis.com. To contact Warren, please email sharp@sharpfootballanalysis.com or send a direct message on Twitter to @SharpFootball.

No comments:

Post a Comment