|
El Duderino -> RE: Around the NFL (News) - 2013 Season (1/27/2015 2:16:27 PM)
|
quote:
ORIGINAL: djskillz
What about the fact that the "study" completely took out Dome teams' fumble rates? They just don't count?
The biggest problem is that the numbers are presented upside down. You need to look at fumbles per play, not plays for fumble. The reason is that there is no effective maximum number of plays you can have per fumble, but there is an effective maximum number of fumbles you can have per play. (Yes, it is technically possible to fumble more than once per play, but that is such a rare occurrence that it really doesn't affect the numbers.) Consequently, the variance from mean gets larger in a non-linear manner.
For example, in a 1000 play season, 25 fumbles means you fumbled once every 40 plays. 20 fumbles means you fumbled every 50 plays, a difference of 10. However, 15 fumbles means one every 67 plays, a difference of 17, and 10 fumbles means one every 100 plays, a difference of 33 - twice as much as the previous gap.
When you look at fumble rate (fumbles per play), the numbers are comparable regardless of where a team is on the curve - A difference of 0.1% in fumbles per play translates to basically 1 more fumble per season whether the team is really good, like New England, or really bad, like Saint Louis.
So, by that measure, New England is still on the high end, but not what could be classified as an outlier. They do indeed have the lowest fumble rate over the last 8 years, but they haven't led the league in fumble rate for a single season since 2010. In fact, Minnesota had a lower fumble rate than the Patriots did this season, at 1.1% vs 1.2%. The biggest outlier season looks to me to be New Orleans in 2011, when they had a rate of 0.5% - more than 0.4% lower than New England's best season of 0.9% back in 2010. Looking at rolling averages, if you shorten the horizon to the last four years, both New Orleans and Atlanta have lower fumble rates (1.2% and 1.3% vs 1.4%), while Green Bay and Baltimore, both outdoors teams, were in the same ballpark at 1.7%.
What jumps out at me is that whatever the impact of a lost PSI might be, it's less than the impact of pure chance, and a hell of a lot less than the impact of having a pass-heavy offense with a QB who likes to get the ball out quickly - something common to the teams that do well by this metric.
Note: I only ran the numbers for seven teams: Baltimore, Green Bay, New England, New Orleans, Minnesota, Saint Louis, and Atlanta. And I can't seem to find numbers for special teams fumbles, which I would remove from my analysis if possible due to the fact that different balls are used on special teams plays. Make what you will of those limitations.
|
|
|
|