Originally Posted by cdcox
Nah, my method wouldn't be like Fantasy at all.
First, I wouldn't play any games until the end of the season so that all of the data that characterizes a players performance is known before we play any of our games. At the end of the season, we'd play a full schedule of our games. Each player's performance in a given week would be randomly generated based on the distribution of that players performances over the season. Based on the aggregate performance of all your players and the aggregate performance of your opponents players, the game out come will be determined. Also, there is a random outcome of each game.
If your OL is bad at run blocking, that's going to hurt the productivity of your running game. Similarly if you are playing against a team with a good run defense. that is going to hurt your running game too. The effects are compounded, so the matchups matter. Your whole team contributes to your victory, not just the skill players.
Furthermore, I think I can accurately simulate the effect of depth and team strategy. If you want to run 3, 4, or 5 WR during some of your plays, it will matter how good your depth is. Or if you are running a two TE set, you better have multiple good TEs. On the other hand, if your WR are Andre Johnson, Wes Welker, Austin Miles, Sidney Rice and Randy Moss, you are going to lose some of the productivity of those players relative to the real NFL because there are only so many balls to go around
So this is my vision. Like I say, I think the data is out there. But for this level of characterization you need a lot of data and you need to bring it together in an organized format to do the calculations. Then you have to develop and test your models, then bring it together. This is the kind of thing I enjoy doing,but it may take me a while (months) given my heavy work schedule.
Given the delay for me do develop my method, I think the idea of voting isn't bad at all. If/when I get my method pulled together we can run it then.
|