IMO, this article is very much an incomplete piece of work . Yes, getting sacks is good and yes, allowing sacks is not, but taking the difference between the two and using that as a predictor of success is a stretch - as the article itself shows. E.g. Winless Cleveland is much higher ranked at sack differential than 4-1 Arizona. When the data you're using to support your hypothesis actually undermines the hypothesis you have to wonder if you've really thought it through.