Thanks for the answers. That is interesting data.MEuRaH wrote: ↑Mon Apr 10, 2017 11:43 amDamn, great questions.kamihamster wrote: ↑Mon Apr 10, 2017 10:45 am (1) In your analysis of each draft class, did you find any stand out classes?
(2) The '14 draft is talked about as one of the best ever, but did your analysis show that statistically? If so, how much better?
(3) Would you be able to adapt your chart to take into account bad, average, and good rookie crops?
If each "stud" is 7 points, "starter" is 3, and "role player" is 1, the median score would be a 33, and that score perfectly divides the upper from the lower classes. On average, each draft class has 2.5 studs, a little over 4 starters, and 6 role players.
(1 & 2) The 2014 class definitely stands out. Watkins, Evans, Cooks, Beckham, and Allen Robinson headline that class. Then you still have Jordan Matthews, KB, Hyde, Freeman, Moncrief, Derek Carr... the list can go on and on. It's a crazy class and it dominates any other class with a whopping score of 71. The next closest is the 2010 class with a 38.
(1) Another class stands out in the wrong way. The 2009 draft class was terrible with a score of 17. That class was HORRID. LeSean McCoy was the lone stud, with Michael Crabtree helping out, and those guys had an ADP of 5 & 6. After that.... not much to see. TY Hilton was the ONLY stud ever drafted, on average, in the 4th round.
(3) This is a tough one. How does one assess the strength of an unknown draft class vs the average? I suppose I could make a "strength" setting that would increase or decrease the likelihood of studs. I guess it would be easy enough to do, and it would be opinion-based on the user's preferences. The current setup is based on the expected value based on 10 years of data, but doesn't take into account the possibility of a strong (2014) or weak (2009) draft class.... and going into those classes, we all knew how strong and weak they were.
Good feedback. This is something I can work on later today.
Regarding question/answer 3. It would defiantly be opinion-based the same way stud/starter/role player values are opinion-based, but considering the variability of 71 to 17, I think taking this into account, even based off an opinion, can be helpful. While projecting is more an art than science, community opinions can change the market as we've seen with these '17 draft picks as it's considered a loaded one. Even if it's not a loaded class the perception is there which makes the picks more valuable. (Last year I could not get a 1st to same my life... i traded for a bunch of 2nds instead...) Maybe an adjustment of +/- 1 or 2 STDEV can be made to adjust value based off class strength? I'm no statistician, but love this kind of analysis. Thanks for taking the time to do it!