Hey,
I was bored and I didn't feel like actually forming opinions about all the teams for my rankings yet, so I decided to try to approximate rankings from the Madden 13 projections that NFL.com gives all the players. We all know the projections are terrible, so there is no need to point how unreliable my data source is. I know it's unreliable. Ideally, the inaccuracy of the projections would average out across the teams and affect each teams standing about equally, but that probably isn't the case.
There are certain explicit cases in which idiosyncrasies of the projections could drastically affect a teams ranking. One example is Daniel: The current projection for two of his players is 0, which could mess with his teams total on the first three of my rankings (although in the last ranking, subs are removed, so those players probably wouldn't have mattered anyway). Additionally, the projections obviously can't take into account cases like MJD continuing to hold out; I assume that the projections have him playing every game.
Also, I took all this data down tediously from the NFL.com player pages yesterday by filtering by team. If you have done any transactions since then, they aren't represented in the data. However, once again, the final ranking does not include your subs, and I doubt any of you have changed your upper ranked players. Similarly, if your projections have changed for some reason, that won't be represented in the data.
I started out with a simple ranking that sums all of your players projections and sorts based on that total. Essentially, it ranks us based on our average player projection. That ranking is not very useful for several reasons that I address in the later rankings. Here are the results:
That ranking included kickers, an unfilled position for some currently. I figured the best way to account for that was removing the kickers from the teams who have them and removing the lowest projected substitute from the teams that don't. The results are shown below:
Another position skewing the data severely was the second QB position. All QBs put up a lot of points, even alternate ones, so teams that had 2 QBs had a large advantage over those who don't (Taylor, Fred, and me, I believe). The other substitutes constitute depth and are much more likely to be used at varying points in the season so they probably should count for something. Ideally I would weigh substitutes less, but that was too much work for a system based on unreliable projections. So for this ranking, I left the subs in, except for either the 2nd QB or the lowest projected sub if there wasn't a second QB. This changed the rankings quite drastically, and looks to have normalized the three owners who don't have 2 QBs (I definitely liked how far it moved me):
The last adjustment I made is beneficial in some ways and detrimental in others. Because owners have chosen to stock different numbers of subs at each position, the data is skewed in different directions for different owners. Although the skew created by the other position subs is certainly less than the effect the QB subs had, it is still there. For example, if an owner has more TE subs than another (1 as opposed to 0), their total will be skewed downwards since TEs earn less points than WRs and RBs, on average. The simple solution to the sub's complication of the data is removing the subs from the totals, however doing so removes a lot of depth from the ranking. If two teams have equal starters, but one teams bench is much better than the other, then the team with the better bench has a better outlook for the season. I went ahead and tried removing all of the lowest projected players from the teams such that each position was filled by the player with highest projection. This did raise the people without depth beyond their starters and lower those who had depth in their subs. I should have increased fairness by comparing apples to apples, but I'm not sure if that was worth the cost. Here is my ranking of teams by their starters using NFL.com projections.
I would say that my 3rd and 4th rankings are the most significant. The 3rd seems closer to where I think the rankings should be (although I really don't think I should be at the top), but there are a lot of confounding variables that don't exist in the 4th. My main problem with my final ranking is that it deviated from my initial direction with these rankings. I wanted to take depth into account in my rankings, but unfortunately the variability of depth strategies makes it difficult to compare depth statistically.
Anyway, I hope you guys maybe got something from my ramblings. If you'd like to take a look at the spreadsheets, I have them
here. Feel free to leave comments of derision or encouragement.
