"TB" stands for team building, not for Tuberculosis or Tampa Bay or something else. It's things I've been looking into for awhile and how I intend on spending my offseason. It's a series of long winded rants by myself on the moneyball-esque basics of attempting to put together an NFL roster that can perennially compete for playoff spots and ends up with a nice window in which to potentially win a Super Bowl. I call these basics because the premise that NFL teams should follow will always be simple. Basic and simple not to be confused with easy to execute. You've got to play the odds to gain some kind of advantage over some of the 31 other teams and that's what this series of posts is all about because it's so difficult to build a winning team.
Despite my constant pro pass offense and pass defense rhetoric and some nicely put together recent fanposts by other Rumblers, I think fans don't understand just how much more important passing and stopping the pass is. This is one instance in which the assumptions of average fans and wisdom of conventional thinking is proven totally incorrect by statistics and it's time to explain myself.
First thing is that everybody should know that this is not a series of arguments or points to defend what I think. I'm a very math based thinker and believer in statistics and I don't use them to prove points as much as I believe what I believe because of statistics. You see a lot of people argue against a stat saying that numbers can often be used to prove both sides of the argument and while I've certainly been guilty of cherry-picking stats, I have these team building beliefs because of what I see in the numbers.
The math: Karl Pearson developed this formula to show correlation over 100 years ago. Don't let the math scare you because all you need to know is that, in it's brilliance, it produces a number between -1 and 1. If the solution comes back as 0, then there is no correlation. The further that number moves away from zero, the higher the correlation there is. A positive correlation (between zero and one) means that as one number moves in a direction (ie: the size of NFL players) the 2nd variable also moves in the same direction (ie: the strength of those players). A negative correlation is when they move in opposite directions (ie: the size of NFL players and how fast they are).
Anything between -.3 and .3 is considered to be a really weak correlation. Anything over .6 or -.6 is considered to be a strong correlation. Anything in between is notable, especially if it's up in the mid 4s into the 5s.
Conclusion: I went back five years and used the data from every team over that timespan. It's basically 1,300 regular season football games. I'm a big believer in using per play numbers as opposed to counting stats like total yardage. I also wanted to get the best pass vs. run numbers possible, so instead of only doing four different correlations (run D, pass D, run O, pass O), a combined each team's run and pass defense numbers. I took each team's yards per carry number and subtracted the yards they allowed per carry and did the same per pass.
A team's difference in passing yards per play and passing yards allowed had a very strong correlation to wins over the past five years to the tune of .778. It's a downright huge number. Teams that threw the ball well and defended the pass well over the pass five years were largely very successful.
A team's difference in rushing yards and run yards allowed per play was a weak .191. While many fans would have guessed that over the past five years, passing was more important than running, but I doubt many people would suggest that it was four times more important. Correlation doesn't prove causation, but looking at the past five years, it's impossible to dispute the gap in success that passing and running teams had.
Some Numbers: Here are a bunch of other stats and how they correlated with wins over the past five seasons.
.8243 - QB rating differential
.7780 - Yards per pass differential
.7200 - QB rating
.6494 - Yards per pass
(-.5454) - Yards per pass allowed
(-.5363) - QB rating allowed
(-.2316) - yards per carry allowed
.1911 - yards per carry differential
.0290 - yards per carry
How about that last number. Over the past five seasons, there has been no correlation whatsoever with running the ball efficiently and winning. Which brings me to the inevitable discussion of how reliable these correlation numbers really are. The first thing to understand is that the math part of it is sound. This is is an extremely accurate representation of how the last five years have gone. The problem is that the correlation doesn't necessarily imply that the first variable caused the second (which in this case is stat X leading to wins).
Things that throw off or explain correlation: The premise is that stat X correlates with wins, so therefore, that stat causes wins. In a lot of more logic based problems than sports stats, correlations can be tricky. But it doesn't take a degree in math or logic or philosophy to see that having a good QB rating does lead to wins, so many of the standard issues with correlation can be thrown out. There are still many possibilities of things that effect these correlation numbers that cause them to be at least somewhat inaccurate.
Does stat X only lead to wins or does winning and losing also play a factor in stat X?
This problem is an interesting way to look at the super low run offense numbers. How much does winning and running the ball in the 2nd half trying to kill the clock negatively impact a team's yards per carry numbers? Does being a bad football team actually help their yards per carry numbers if they're getting blown out a lot?
Running QBs is another thing that can throw off the rushing stats to some degree. Cam Newton and the Panthers had the highest yards per carry of any team this past season. They also had a high yards per pass attempt and scored a lot of points though. I don't think there have been enough running QBs to do much other than slightly skew the stats.
Any other reason that people can think of that the correlation would be so high for passing and so low for rushing other than those things just mattering that much more or that much less?
Next TB post is offense vs. defense
Since I did a mock pretty recently, I changed this one up by throwing in a preplanned trade down to 25 for picks numbered 57 and 89. To make things interesting for myself, I decided I'd trade back up at some point and see how that turned out. I do these mocks when I'm bored all the time where I find a mock draft I've never read (this is the latest update from NFLdraftsite.com) and do my mock off their mock without looking ahead to see who will be and won't be available with the next pick. They often turn out terribly, but I don't post those ones. Here is one I thought went very well:
1 - Chase Minnifield, CB, Virginia - Early impact, I like his long term upside. He's got all the tools to be a Pro Bowl type player, but had a bit of an up and down year this past season.
2 - Vinny Curry, DE, Marshall - Good combo of pass rush talent and motor. DEs were going fast and I was happy to jump on Curry here.
2 (trade up, both 3rd rounders) - Dwayne Allen, TE, Clemson - Smooth athlete for somebody his size and the speed to run away from LBs. He'll wow from time to time by making difficult catches look routine. Strong enough to play in-line, even though he's not much of a blocker at the moment.
2 - Ronnell Lewis, OLB, Oklahoma - Hopefully Buffalo realizes that the best way to utilize their new SAM LB spot is with a pass rusher. Lewis may not be much of a real linebacker, but he'll fly to the football with reckless abandon. I'll happily trade some bad angles for a few sacks and a guy who tries to run through people.
4 - Audie Cole, LB, NC St. - True linebacker who may have the versatility to backup all three spots. In this scenario, he'd battle for time at the SAM spot with Lewis.
4 - Brandon Mosely, OT, Auburn - Tough, scrappy player who finishes his blocks like you'd expect from an Auburn lineman. Good length at 6'6'', just needs to polish up his game.
5 - Janzen Jackson, FS/CB, McNeese St. - Superstar CB recruit who Tennessee moved to safety where he started as a true freshman. He was eventually kicked off the team for reasons that were never made entirely clear, but were most likely substance abuse related. A pure BPA pick here that would be well worth the risk. Could back up Byrd and provide depth at the nickel and dime corner spots. Tons of special teams upside too. I think a team is going to jump on him in the 2nd round based on talent alone.
6 - Tyrone Crawford, DE, Boise St. - Was planning on taking another pass rusher at some point and got lucky that Crawford was still available. Surprising amount of potential as a pass rusher to be available this late.
6 - Lance Lewis, WR, East Carolina - Big guy who plays physical. Smooth athlete who may be able to separate some in the NFL. Kind of a depth flier here who should be a good special teamer as well. Wouldn't be a surprise if he, ya know, was a surprise in the NFL.
7 - Kelvin Beachem, OT/OG/OC, SMU - More of a first dibs on a priority FA situation than anything else here. I've read some decent scouting reports on Beachem and his athleticism. He's a college OT who will have to move inside because he's 6'3'' and a little short armed. I thought he was a value pick (if such a thing exists in the 7th round) who Buffalo could stash on the PS and maybe cross-train as a guard and even a center. A project worth the draft pick here.