I wanted to follow up on my post from a few weeks back with regard to working on a running back rating system, based on this paper. The point of the paper is to apply, essentially, the NFL passer rating system to the running back position, utilizing a similar methodology to that which was used to create passer rating, attempting to go beyond simply using rushing yards to declare who the top running back in the league is.
First, the good. The methodology used was sound, as it used a large data pool (all seasons of 1000+ rushing yards going back to 1950), giving them a much larger and more extensive data set than that which was used to create passer rating (which was made, I believe, in 1970). They also incorporated receiving, which I think is helpful, since a running back's true contributions should encompass everything they're doing.
The bad: First, we must operate under the assumption that passer rating is an effective method of evaluating quarterbacks in order to assume that such a rating system should be applied to other positions to begin with. In that regard, I think that it is a decent starting point, and as with any other statistics, cannot be taken in isolation. It is a piece of the puzzle, but it is not the only piece of information we use to judge quarterbacks. Thus, a rusher rating system (I prefer running back rating system, since the rating takes into account receiving as well) is a piece of the puzzle for determining who the best backs are.
Another gripe I have is that receiving essentially receives the same weight as rushing. While I will agree that the receiving aspect is important, by equally weighting it, it punishes those backs who are not large contributors in that aspect, regardless of how good they are as runners. I think that aspect is another one that should be revisited and possibly revised.
Overall, I think it is a good starting point, and if nothing else, gives us another way to look at the players with the data that is readily available (box scores). No, box scores don't tell the whole story and doing something like grading every play, like Pro Football Focus and others attempt to do, is a better approach, but most of us do not have the time or inclination to do so.
Some additions I made were to figure up career ratings for each player, as well as to assign a "Rating+" figure. This is inspired by the ERA+ stat from baseball, in which a pitcher's ERA is adjusted for league and park factors, so that you can compare across eras. For instance, a 3.00 ERA at the height of the dead ball era is completely different from a 3.00 ERA at the height of the steroid era. Thus, by using ERA+, we can begin to compare pitchers from different eras, at least how good they were in relation to their contemporaries.
To do this, I used Pro Football Reference's methodology for their QB metrics, and factored up the player's rating and how many standard deviations above they were, and multiplied the number by 15. This is then added to 100 to give you the Rating+ number. By this metric, holy hell Marshall Faulk.
Additionally, I then calculated 3, 4, and 5 year peak ratings and rating+ figures for each of the players I looked at. This was inspired by the JAWS system or the Jaffe WAR Score, which uses a 7 year peak for baseball players. However, to account for shorter football careers (especially at running back), I shortened it to 3, 4, and 5, just to see where they stood at the peak of their powers.
All in all, it was a lot of fun to do, and it gave me a chance to dick around in Excel and do some fun stuff there. I've also come across a WR rating system that I think I am going to tinker with and do many of the same things when I have some time, as well as going and getting Pro Football Reference's Rating+ numbers for QBs and fucking around with those a bit. If I can figure out how to upload files, I'll upload the Excel file in the off chance anyone ever comes across this and wants to look it over.
