For the projections, I wanted to create a projection system similar to Marcels in which Tom Tango created for baseball players.

As Tom describes it, "... uses as little intelligence as possible. So, that's the allusion to the monkey."

I went with the same approach with just a tad more detail in the weights and regression. The basic procedure was to take a three-year weighted sample, regress the data as needed and then add an aging factor. I did not project playing time in which I'll examine in a future article.

Overall, the final formulas came out as expected, heavy regressed. This aspect is always the hardest for owners to digest. With just 16 regular-season games, a big game or two can really boost a player's value and they will, on average, see their production drop the next season after a breakout season. Finally, for these projections, I used data from 1990 to 2016 from ArmChair Analysis.

Additionally, I found a few attributes which had no year-to-year correlation, like fumbles. For these stats, I just used the 2016 league average value. For all the regressions amounts, I regressed to just the 2016 values. I may change this value