04/23/24
January 24, 2015
Stats Guru: Patriots Hold Onto Ball Far Better Than Any Other Team: Sports quantitative analyst Warren Sharp has found an interesting statistical anomaly involving the New England Patriots: The team has fumbled dramatically fewer times than any other NFL team the past five years, measured in term of fumbles lost per offensive play run: "There is no other team even close to being near to their rate of 187 offensive plays (passes+rushes+sacks) per fumble. The league average is 105 plays/fumble. Most teams are within 21 plays of that number." The Patriots five-year run of non-fumbling is the best five-year run of any NFL team in the last 25 years, raising the question of how the team is able to hold on to the ball so much better than anybody else.
The link isn't working for me (gives an error message about missing content) although I've seen this making the rounds, especially here in Seattle. Hey, maybe he took it down in embarrassment! :)
The article is innumeracy at its finest, but I'm too exhausted from #DeflateGate to keep running numbers. I've quickly copy/pasted a FB post I made a couple of days ago about this article. As always, I reserve the right to be wrong, gracefully. :)
-------------------
This has been making the rounds today. Seattle is quickly proving that its sports fans are not only fair-weather... they aren't exactly aces at math. ;) Even assuming these numbers are accurate:
FIRST, this "damning" graph is conveniently y-axis shifted so the trends *look* more exaggerated, in particular those fumble numbers. We should see the same thing without the deceptive framing. What are these supposed to be, national unemployment figures, ha ha? :(
SECOND, the far right of the graph shows 33 NE fumbles compared to Houston 40... over a span of five seasons. Wow, a whopping 1.4 fewer total fumbles per season. Surely, some kind of Belichickean dark sorcery is afoot here.
THIRD, since the 2010 season of this graph, the Patriots have 5 straight division wins averaging 12.6 wins per season (and never less than 12). The Texans in that same period have two division wins of 10 and 12 win seasons... but missed the playoffs the other years with win totals of 9, 6, and *2*. Again, it's really suspicious that New England has fumbled 7 whole times *less* than the Houston Texans in that same period.
FOURTH, this y-axis shifted graph is showing two values per team: fumbles per season, and plays per fumble. It doesn't, however, appear to distinguish between running plays and passing plays. How convenient. Unsurprisingly, a team led by a no-doubt Hall of Fame quarterback runs more passing plays than normal. Less running plays means less chances to fumble, means a greater ratio between overall offensive plays and fumbles.
FIFTH, the ratio of fumbles over this span between Houston and NE is .825 (33/40). If the Patriots otherwise executed the same number of plays over that time span- ha ha, that's surely the case- then inverting the ratio to get plays/fumble would turn 140 into... 170. Oh, but the perennial contender Patriots with their future HoF quarterback have a modestly higher ratio than that! Quelle surprise.
SIXTH, just so I'm not accused of picking the poor beleaguered Texans... if we compare Seattle to NE, we get an average of 10 wins per season (7, 7, 11, 13, 12) and find the Seahawks fumbled the ball on average just over 5 more times per season. Probably a lot of that difference in those 7 win seasons. Hm... I wonder what the fumbles per game (FPG) was in that span...
Year, NE, SEA
2010, 0.7 (1), 1.1 (7)
2011, 0.9 (4), 1.4 (9)
2012, 0.9 (5), 1.1 (8)
2013, 1.5 (24), 1.6 (26)
2014, 0.9 (2), 1.5 (22)
So NE is consistently elite but not exceptionally so when it comes to fumbling (see that 24th place rate in 2013). Oddly, Seattle started fumbling a lot more- near league worst- when they made their two SB appearances.
Maybe... maybe fumbling rate isn't such a great fucking stat to use to prove a point?
All this chart really tells us is:
a) How easy it is to mislead with statistics and rigged graphs
b) A consistently elite team will, over several years, make several fewer mistakes on average than other teams.
c) If we were to accept the original author's premise, the NFL really ought to be investigating the Philadelphia Eagles and Denver Broncos under suspicion of tampering with their footballs to the point of being comically *overinflated*. :)
posted by hincandenza at 08:00 PM on January 24, 2015
Unsurprisingly, a team led by a no-doubt Hall of Fame quarterback runs more passing plays than normal.
New England's pass attempts per game, ranking in league:
2014: 7th
2013: 8th
2012: 4th
2011: 3rd
2010: 18th
Do you really think that explains New England's five-year run of non-fumbling? Here's New Orleans' pass attempts per game, ranking in league:
2014: 1st
2013: 4th
2012: 2nd
2011: 1st
2010: 1st
All those passes, yet New Orleans fumbled once every 126 plays over that time.
posted by rcade at 09:33 PM on January 24, 2015
Which means, honestly, precisely nothing. Again, if the Texans fumbled one fewer time a year, they'd have NE's numbers. Do you dispute this ratio, or the idea that so small a change would negate the entire story?
This guy is a charlatan, trying to drum up business with a viral story that plays to people's pre-existing biases. The premise itself is threadbare: it's 100% statistical cherry picking, making up some random stat no one's ever heard of before- seriously, "plays/fumble"?- and then find a way to produce an outlier by using all sorts of trickery and conflation, such as I mentioned above.
There are so many conceptual mistakes in this whole process. If you suspected the Patriots of some ball-related mischief, you'd do year-by-year comparisons to find when they implemented this supposed deflation technique of lowering numbers. I mean, unless you somehow know for a fact that the "tampering" not only gives them an edge in fumbling but has also been going on all 5 years, averaging the stats would actually hide or diminish any recent outliers.
But then, we know that's not why he chose a 5-year average: he did it because most teams can't even boast the same coach and QB over the past 5 years, much less winning the division 5 years in a row. If you wanted to manufacture a controversy, you'd find some way to magnify even the slightest difference.
Heck, his updated "Fumbles" and "Fumbles Lost" chart shows the Atlanta Falcons having an even more dominating rate than the Patriots, and the Saints not far off. While he was savvy enough to put the y-axis at 80 out of 150- almost as if to exaggerate the graph- he somehow wasn't clever enough to split the numbers between home and way, given how concerned he was with dismissing those Falcons/Saints numbers as irrelevant to his "Patriot cheating" narrative since they played in a dome. Well, they played their only half the year of course, but still... odd, don't you think?
There are countless other explanations for a junk stat made up this week showing some kind of "trend". The current top comment on the article is from some guy, Glenn Brown, who points out that BenJarvus Green-Ellis had 510 carries in his 4 years with the Patriots... and ZERO fumbles. Just in 2010 and 2011, he had 410 carries for 0 fumbles. 205 plays a year without a single fumble- and with about 1,200 total plays a year passing and rushing for the Patriots, that means this single player running 200 times a year without fumbling would probably show up as an extreme statistical outlier in team fumbling rates.
By the way, while he's not in the league now, BJGE left NE in 2012, and spent the next two years in Cincinnati. On his modified Total Fumbles and Plays/Fumble chart spanning 2010-2014, guess which outdoor team is now fourth in highest Plays/Fumble rate? Go on, make a wild guess.
This whole thing is a joke, and I wish Howard_T or grum could jump in to debunk this further.
posted by hincandenza at 12:05 AM on January 25, 2015
Plays-per-fumble is hardly an unknown statistic. It is brought up in broadcasts where running backs are concerned.
I'm not a numbers geek. But looking at how often a team loses fumbles per offensive plays run seems like a relevant statistic to me. Turnovers have a strong effect on the outcome of games. If Belichick devised a way to increase ball security -- whether it was a "rubbing process" that results in deflation below 12.5 PSI after testing (his words), a coaching technique or a trait he can identify in players -- it could be a meaningful advantage.
You're making a lot about the graphs not having 0 as the base of Y axis, making them more visually dramatic and somewhat misleading. That's true, but I think you make too much of it. We're talking about a blog post, not a peer-reviewed academic study. Newspapers do what he did all the time in graphs -- I can remember my J profs yelling at us about doing it in the school paper. It's not always a calculated attempt to mislead. Sometimes it's just sloppy thinking in how a graph should be presented.
Personally, I've never seen it as much of a problem when the Y-axis range is right there on the graph and can be taken into account.
Which means, honestly, precisely nothing.
You use the Patriots passing a lot as a possible reason to explain this stat, but dismiss the Saints passing a lot as meaningless.
It seems to me you're trying to have it both ways. You're calling his statistical finding meaningless misleading junk while also coming up with reasons why the Patriots are so good at not losing fumbles.
posted by rcade at 08:27 AM on January 25, 2015
An interesting comment to a Slate piece about this blog post:
If you watched the press conference today the Patriots admitted what they did and what their practice is. They create friction on the ball resulting heat which temporarily raises the PSI of the ball. They do this until the ball is turned over to the refs. The balls are checked or tested when received by the Refs in an 80 degree room before the game. After that the balls are secured and then taken to the field in the cold outside air where the air inside the ball reduces with temperature and therefor cause the ball to lose pressure.They claim that don't break the rules technically because their method does not involve mechanically deflating the balls and they advise the refs to be sure the balls have pressure per the rules (which at that time they do). Their intention is to get balls that are under inflated and their practice in ball preparation makes it likely to occur.
posted by rcade at 08:34 AM on January 25, 2015
More fun from a Wall Street Journal piece about Sharp's post:
Additionally, according to Stats, LLC, the six players who have played extensively for the Patriots and other teams in this span all fumbled far less frequently wearing the New England uniform. Including recovered fumbles, Danny Amendola, BenJarvus Green-Ellis, Danny Woodhead, Wes Welker, Brandon LaFell and LeGarrette Blount have lost the ball eight times in 1,482 touches for the Patriots since 2010, or once every 185.3 times. For their other teams, they fumbled 22 times in 1,701 touches (once every 77.3).The WSJ presents a graph on this with a proper axis beginning at 0. It still looks plenty dramatic.
posted by rcade at 08:40 AM on January 25, 2015
A WSJ commenter points out that a better metric would be fumbles per rushing attempts and pass completions, since no one is fumbling an incompletion. He notes: "When you re-analyze regular season data from 2010 to 2014, with total fumbles (not fumbles lost) in the numerator and rushing attempts plus receptions (not total offensive plays) in the denominator, the lowest incidence of fumbles is with the Ravens (0.82 fumbles per 100 rushes plus receptions). The Pats are ranked #2 on this statistic, at 0.85. Next come the Saints (0.90), Falcons (0.99) and Packers (1.06)."
posted by rcade at 10:35 AM on January 25, 2015
Well darn those cheatin' Ravens. They're the ones that started all this trouble. (j/k)
posted by beaverboard at 11:40 AM on January 25, 2015
Bill Nye the Science Guy scoffed at Belichick's ball-rubbing theory.
posted by rcade at 01:35 PM on January 25, 2015
Bill Nye is a media personality, who lives in Seattle. He either chose- or was edited- to misrepresent Belichick's statement, by conflating BB's claim that the PSI drop was due to temperature change (which we've discussed here) with his statement that part of the football handling process was to rub them up as well for the benefit of Brady's preference.
These are all legal behaviors in the NFL mind you, and I continue to think the story will eventually come out that the refs only did a spot/hand check before the game, because no one has ever cared or more likely known about the pressure drop; after all, a 13.5PSI ball in San Diego and a 12.5PSI ball in Green Bay could be 3PSI different on their respective fields... yet no one has ever noticed. That we'll have different and more specific rules going into 2015 is certain; that the Patriots violated any rules whatsoever is completely uncertain.
rcade: You use the Patriots passing a lot as a possible reason to explain this stat, but dismiss the Saints passing a lot as meaningless.Because those aren't mutually exclusive. His presentation, analysis, and thus his findings are suspect due to sloppy methodologies and assumptions; that's what I'm calling junk. The choice of axis, including passing plays, etc.- for a "professional" analyst, it's suspiciously sloppy to me.It seems to me you're trying to have it both ways. You're calling his statistical finding meaningless misleading junk while also coming up with reasons why the Patriots are so good at not losing fumbles.
I further suggested that part of the reason it's junk is because when you average over a period of time, you can exaggerate a consistently high performer when other performers have mixed results- often due to changing personnel or simple lack of organizational success from one year to the next.
I asked that someone do actual leg work of a more meaningful kind, and at least one person did in that previous comment you quoted: when looking at the fumbles / (rushing plays + completed passes ) even averaged over 5 years, you get the Patriots #2, behind the Ravens and about the same distance in front of the Saints. So why, as beaverboard jokes, are we not talking about the Ravens' cheating ways?
posted by hincandenza at 02:54 PM on January 25, 2015
Because the Ravens weren't caught with dodgy balls.
Bill Nye has been a science educator for 22 years. He's got a mechanical engineering degree, has multiple patents and has done work in aeronautics and the space program. I think he's qualified to talk about the inflation of footballs without being dismissed as a "media personality." Nye declared himself a Seahawks fan in the GMA piece, so it's not like he hid that.
Belichick himself talked about the "rubbing process" in relation to PSI, so I don't see how Nye is being unfair to him.
To me, Belichick's explanation is pretty silly when you consider that the Colts balls were in the same environment as the Patriots, were tested in the same place and exposed to the same elements, yet were not below required PSI.
... that the Patriots violated any rules whatsoever is completely uncertain.
I think the fact 11 of 12 balls were below required PSI, while none of the Colts balls were, is enough evidence of rulebreaking to levy a fine. It might have been smart for Goodell to just do that right away and leave the rest for a future discussion on how non-kicking balls are tested and handled in games.
The world didn't end when the Patriots' groundskeeper cut a path for their kicker in the Snowplow game. If Goodell just punished the Pats for improperly inflated balls without getting into intent, this circus might already be taking down the tents.
posted by rcade at 04:50 PM on January 25, 2015
Rubbing a football in order to raise its temperature by any significant amount would require an amount of pressure to be exerted upon the ball that would require machinery. Human muscle would be inadequate. Further, it would take some large amount of time to raise the temperature, and once you stop the mechanical process, the temperature of the ball begins to drop. I do not have any numbers or formulae to offer for how much energy is needed to raise a given amount of air by any given temperature. Considering the temperature of the air inside the ball, if it is above ambient, unless the container is insulated in some way, the air inside will soon return to ambient. If you have ever dealt with home insulation, you have heard of the R value. This is a measure of heat transfer across a surface (i.e. from inside to out). If someone cares to look up the R value of pigskin and a contained rubber bladder, let me know. Anyway, raising the temperature of contained air without raising the ambient temperature in which the ball is sitting would be rather difficult.
I believe Belichick did not mean to imply that rubbing the footballs had something to do with pressure. He was talking about the process by which the team tried to prepare the surface of the ball, and mixed it up with inflation. Anyone looking to find fault would tend not to parse his words too carefully.
Those who talk about the Colts' footballs not showing deflation below specification seem to assume that they were inspected along with those of the Patriots. I have not seen anything from the NFL that says they were. Could someone enlighten me on this?
It is all boiling down to the smartest kid in your 8th grade class and his good looking buddy that gets the girls for the 2 of them scoring another A without seeming to work for it. Let's face it, every one of us had one of those pains-in-the-ass in his class, and you wanted to kick the living crap out of him or her. If indeed Belichick knew exactly how the balls were cursorily inspected by the officials and knew that filling them to 12.5 in a warm room would result in their deflating below specification, then he was working to the letter of the rule, but bending its intent. Can the NFL or anyone else prove that this is the case? It's now the classic "he said, she said", and comes to a point of whom do you want to believe. For those who wish to believe the worst of the Patriots, here's some ammunition. Defensive coordinator Matt Patricia holds a degree in Aeronautical Engineering from Rensselear Polytechnic Institute, and worked as an engineer for two years after having a job as a graduate assistant at RPI.
Once again, the NFL has a crappy rule that was not being enforced with the vigor needed. It was circumvented with ease, and now the league has egg on its face. Roger Goodell really has his tit caught in the wringer on this one. If he says that the Patriots worked to the letter of the rule and outsmarted everyone, there will be pitchforks and torches at the gates of NFL Headquarters. If he says that the Patriots cheated, Robert Kraft, a close associate of Sumner Redstone, majority owner of CBS, might have something to say about TV contracts, sponsors, and the like. As bad as it might look for the NFL, it seems they are the ones who need to take the hit, change the rules, or at least the execution of the process, and admit how screwed up they really are.
posted by Howard_T at 06:56 PM on January 25, 2015
It's weird Goodell seems to want to punish. I'd love to see a reporter contact as many NFL owners as possible to see how much they care about this.
posted by rcade at 09:54 PM on January 25, 2015
Sharp has posted a followup. I haven't read the full piece, but it looks like he doesn't address criticisms. Instead he finds more ways to say the Pats look dodgy.
posted by rcade at 10:51 PM on January 25, 2015
This reminds me of the Serial podcast in that once you have the implication of guilt, even pretty normal things look dodgy. If you would have learned that the Pats were 2nd in fumbles before all this started you wouldn't have thought anything of it but now it's supposed to mean something. You can barely notice the difference in PSI but now it determines fumbles, passing, and who knows what else.
I was a little annoyed by this story to start with but now I'm hoping it lasts til Sunday. Honestly, it's more entertaining than the standard Super Bowl coverage. I will watch the opening snap of the Super Bowl and wonder, for the first time in my life, what PSI that ball is at. Then I will drink a beer and never think about it again.
posted by tron7 at 11:19 PM on January 25, 2015
Bill Nye is a media personality, who lives in Seattle
He's also a scientist whose comments are backed by physics. Baseless defiant arrogance just doesn't sell as well as science.
You can barely notice the difference in PSI
2 PSI is a significant difference. Take 2 footballs out on a cold day and do your own experiment. A properly inflated ball is much harder to handle than an under-inflated ball.
posted by cixelsyd at 10:09 AM on January 26, 2015
Jay Glazer is reporting the NFL is zeroing in on a locker room attendant who took balls somewhere between the officials room and the field and has video.
posted by dfleming at 04:43 PM on January 26, 2015
the NFL is zeroing in on a locker room attendant
Highly probable to be terminated immediately ... with guaranteed life long financial security as long as compliance to the terms of the Patriots separation agreement "HUSH.2015.NEP" are maintained.
posted by cixelsyd at 05:45 PM on January 26, 2015
For the Patriots to blame a change in temperature for 15% lower-pressures, requires balls to be inflated with 125-degree air. -- Neil deGrasse Tyson
posted by rcade at 06:43 PM on January 26, 2015
Tyson doesn't specify the measurement scale and may not have employed accurate discrepancy data in psi. His 15% number casually tossed into a tweet without references is happy hour finger food bullshit.
We've talked about the warm room/hot air proposition.
FWIW, 125 degree air is not hard to achieve. Anyone who has operated an underpowered, small displacement air compressor knows that those things cycle on and off almost continuously just trying to keep the air hose pressurized.
Air compressors can get damn hot. The air can too. For the true air sleuth, the telltale clue would be trace amounts (or more) of oil, which can be smelled for one thing as air is let out of a pressurized enclosure. Some oil tends to migrate into the air line when a compressor pump head gets a bit overheated. Basic small compressors don't have fancy filtering apparatus and let the oil into the air fairly readily.
Of course, the refs weren't letting air out of the Pats' balls. They were busy putting more air in.
Where are those game balls now? If we're going to get uber technical with the postulations, how has the physical evidence been handled from the first moment someone took action on this during the AFC title game? I don't get a sense that this thing has been handled in a tidy, ironclad manner. The overall feel of this is not that far removed from the Secret Service balling up and tossing Kennedy's clothes into the back of the Lincoln after the autopsy at Parkland.
posted by beaverboard at 07:32 PM on January 26, 2015
Tyson bungles science of Deflate-gate scandal
posted by beaverboard at 10:08 PM on January 26, 2015
Video shows a Pats locker room attendant taking all 24 game balls into a bathroom, but only for 90 seconds.
posted by rcade at 08:01 AM on January 27, 2015
When I mentioned the Patriots' low fumble rate rather casually in the deflation mega thread the day before Sharp's piece came out, I had no idea how startling the actual numbers were.
The NYT ran an article yesterday on how the Pats cover the spread in bad weather games at a much higher rate than they do in good weather games. Sharp may have done the analytic work on that story as well.