Predictions: Cowboys Will Improve Point Differential, Triple INTs
At NBC, I posted two predictions for the Cowboys in 2013. The first is that the team will dramatically improve their point differential.
I’ve already explained why the Cowboys will almost assuredly improve upon their -24 point differential from 2012, namely that they’ll secure more takeaways. With just seven interceptions last year, the ‘Boys ranked last in the league. That will change under Monte Kiffin, and it will have a double-barreled effect, limiting the opposition’s points and increasing the probability of points for the offense.
One of the cool things that we can do with the point differential is estimate the Cowboys’ projected recorded with great accuracy. Year in and year out, NFL teams’ actual records closely resemble their points scored and allowed. Using what’s known as the Pythagorean Expectation, we can 1) figure out how lucky a team was in the past and 2) predict their future record by accurately estimating points for and against.
The formula itself is kind of messy (you have to use an exponent of 2.37), but the results are incredibly accurate. Last season, the Cowboys had a Pythagorean Expectation of 7.4 wins, meaning they got lucky to win eight games given how they played. However, take a look at how their 2013 Pythagorean Expectation will change as their point differential does the same:
425 points scored, 400 points allowed: 8.6 wins
425 points scored, 375 points allowed: 9.2 wins
450 points scored, 350 points allowed: 10.3 wins
The second prediction is that the Cowboys will triple their interception total from 2012.
In projecting the Cowboys to improve their point differential in a big way in 2013, one of my primary assumptions is that they’ll have more interceptions. The Cowboys hauled in just seven picks in 2012—the fewest in the NFL—and forced a turnover of any kind at the third-lowest rate in the league. But there’s reason for hope. Below, I listed the top four reasons the Cowboys’ interceptions will triple—to 21—in 2013.
Interceptions are Fluky
Although interceptions obviously aren’t completely random, they’re more random than you think. There’s only a very weak correlation for interceptions per game thrown by quarterbacks from one year to the next. So using a quarterback’s picks in Year N to predict those in Year N+1 has nearly no value, and that’s for the person throwing the ball! No wonder defensive interception rates fluctuate so much from year to year.