The article doesn't stand on its own. There are a host of issues that you have to be familiar with before things like "the value of a touchdown" makes sense..
That phrase is typically used in expected points models to mean the average value of a touchdown minus the value, on the field, of the average runback. Before 2011, that was about 26-27 yards. Typical values before 2011 were 6.3 to 6.4 points.
The PFR model appears to value the touchdown as 7 points. Further, with a model derived value of 6.97 points for first and 10 on the 1 yard line, they appear to be saying that the odds of a touchdown on the 1 exceeds 99%. That's not true. Their model has some issues.
Things like this make it worthwhile to suggest to those folks that they tell us how they came by their model.
Almost all the modern stats (EP, win probability) have to be derived by parsing play-by-play data into a spreadsheet or database, and then you analyze them for the kinds of results you want. Run success calculations require a set of play by play data. If you start doing run success calculations, and then folks start arguing about whether a running back is "done for" ..
It's going to be hard to read my site unless you spend a bit of time on the articles I wrote in August and September of last year. That was an amazingly productive period. A number of the results published then were original. A fair number of those articles were retweeted or linked to by Brian Burke.
And thank you all for the "thumbs up!"
D-
