This is great. I love the way you don't hew to RAS, but rather you use position-specific metrics demonstrated to be predictive with respect to outcomes. Have you shared these metrics by positon? If not, can you do so? Do you have a database that compares athletes based on these position specific metrics? I have long thought RAS was much too broad a brush and that its broad brush was more relevant for some positions than others. I have read, for example, that bench press is an important metric for WR's, and that the short shuttle is predictve for OT's and the 3-come for CB's. Would be super helpful to see what your analysis has shown by position group. Cheers! Your posts are great.
Hey Nolaboy! I would love to do something like that! Before I do that, however, I'd like to make sure that my measurements are rock-solid on that point -- the models were built using an older definition of success and a slightly different era of football. And, honestly, I'm not well-versed in statistical modeling. It does sound like Kent, who does RAS, may create a position-specific model and I'd rather he do that and get the credit for it if that indeed comes to fruition because he has a better capacity to share it and make it usable.
If not, I can do something along those lines next off-season because I think there's something valuable to be gained here.
Sounds great Arif! I do think this would be a real step forward to your point. I loved your critiques of RAS - lack of position specificity tied to outcomes; the challenges with adjusting for height and weight; and the fact that it bunches players in the middle and under weights the outliers. I am guessing the lack of position specificity is biggest opportunity area for more useful / more predictive athleticism metrics. I am eager to see what Kent produces and understand he's best person to do that. It would be fun for me to speculate on what insights Kent will produce but that would bore your readers haha. In addition I would love to see someone like you or Cooper Klaus advance Cooper's work on a player/draft ranking incorporating Kent's forthcoming position-specific athletic metrics, PFF production metrics adjusted for level of competition (could be as simple as removing grades for games against teams ranked under # 50 in FBS); and surplus value using real time contract data. Forecasted/expected surplus value dollars could be the output measure...everyone understands and can easily relate to dollars as a metric. Just some food for thought. I speculate that taking your Consensus and adjusting for these things would produce some interesting findings with respect to NFL outcomes.
PS - One tweak for next year - I think the denominator for "hit rate" for those outside of the Top 300 ought to be the number of UDFA's in total signed by teams less the number of those signed who were inside the Top 300. So for example: if 20 players outside the Top 300 saw a snap count of 200+; 75 UDFA's inside the Top 300 were signed; and 225 UDFA's outside the Top 300 were signed by NFL teams, then the "hit rate" for those outside the Top 300 = 20/225 = 9%. I think your "hit rate" for those outside the Top 300 is slightly understated the way it is currently being calculated. However the point still remains that the hit rate for players inside of the Top 300 is far higher than those who fall outside of that. I am sure that if there were a Top 350, most of the productive NFL UDFA players would fall within that. Your posts are terrific!
Arif,
This is great. I love the way you don't hew to RAS, but rather you use position-specific metrics demonstrated to be predictive with respect to outcomes. Have you shared these metrics by positon? If not, can you do so? Do you have a database that compares athletes based on these position specific metrics? I have long thought RAS was much too broad a brush and that its broad brush was more relevant for some positions than others. I have read, for example, that bench press is an important metric for WR's, and that the short shuttle is predictve for OT's and the 3-come for CB's. Would be super helpful to see what your analysis has shown by position group. Cheers! Your posts are great.
Hey Nolaboy! I would love to do something like that! Before I do that, however, I'd like to make sure that my measurements are rock-solid on that point -- the models were built using an older definition of success and a slightly different era of football. And, honestly, I'm not well-versed in statistical modeling. It does sound like Kent, who does RAS, may create a position-specific model and I'd rather he do that and get the credit for it if that indeed comes to fruition because he has a better capacity to share it and make it usable.
If not, I can do something along those lines next off-season because I think there's something valuable to be gained here.
Sounds great Arif! I do think this would be a real step forward to your point. I loved your critiques of RAS - lack of position specificity tied to outcomes; the challenges with adjusting for height and weight; and the fact that it bunches players in the middle and under weights the outliers. I am guessing the lack of position specificity is biggest opportunity area for more useful / more predictive athleticism metrics. I am eager to see what Kent produces and understand he's best person to do that. It would be fun for me to speculate on what insights Kent will produce but that would bore your readers haha. In addition I would love to see someone like you or Cooper Klaus advance Cooper's work on a player/draft ranking incorporating Kent's forthcoming position-specific athletic metrics, PFF production metrics adjusted for level of competition (could be as simple as removing grades for games against teams ranked under # 50 in FBS); and surplus value using real time contract data. Forecasted/expected surplus value dollars could be the output measure...everyone understands and can easily relate to dollars as a metric. Just some food for thought. I speculate that taking your Consensus and adjusting for these things would produce some interesting findings with respect to NFL outcomes.
Cheers and thanks again for everything you do!
You mention Jaquan Brisker as being a notable name here, but he was a 2nd round pick in 2022
It's fixed. Wasn't in the data, not sure how I just threw him into the article
Thank the gods we have appeased the almighty comment section.
Who needs editors.....
PS - One tweak for next year - I think the denominator for "hit rate" for those outside of the Top 300 ought to be the number of UDFA's in total signed by teams less the number of those signed who were inside the Top 300. So for example: if 20 players outside the Top 300 saw a snap count of 200+; 75 UDFA's inside the Top 300 were signed; and 225 UDFA's outside the Top 300 were signed by NFL teams, then the "hit rate" for those outside the Top 300 = 20/225 = 9%. I think your "hit rate" for those outside the Top 300 is slightly understated the way it is currently being calculated. However the point still remains that the hit rate for players inside of the Top 300 is far higher than those who fall outside of that. I am sure that if there were a Top 350, most of the productive NFL UDFA players would fall within that. Your posts are terrific!