Bob Magaw
Footballguy
having an agreed upon definition of what constitutes a hit or miss (within + or - 3-5-10?) would be a good (necessary) start to measuring accuracy...
though david brought up a good point... should a +3 value in top five be weighted more heaviliy in importance, compared to a +5 increase in the 30s-40s...
he brought up another good point, about the temporal aspect of the changing ADP landscape... if you make a call in JULY that hardesty is overvalued at RB40, and in AUGUST harrison goes on IR, hardesty's ADP shoots up to RB20, and he finished RB 12, was that a miss? presumably savvy readers would recognize a substantive material change in the situation and re-rank/draft accordingly...
on the no better than a coin flip take...
a few years ago, i subbed a few times for the weekly IDP upgrades/downgrades... on the upgrades, there were a few simple guidelines... restrict the handful of positional picks to outside a certain range (say top 20 - not much use in the captain obvious point that patrick willis is likely to do well that week...
), that would score above a specified range... it soon became apparent that the odds were often stacked against a high % of hits... in some weeks, out of possibly dozens of DL/LB/DBs, only a few might have satisfied the criteria... getting a couple right and a couple wrong could involve a degree of difficulty not obvious by the seemingly 50% hit rate... use DBs for an example... 32 teams (non-bye week) play 5-6 regularly... but even sticking with 4 starters = 128 possible to be chosen from... if only a few satisfy the criteria in any given week, even a 50% hit rate (on a much more lower percentage chance play) could be much better than appears at first glance...
the value/under-value play articles aren't structured the same, but just trying to make the point that 50% could mean different things depending on the context... if a value call not appreciably exceeding ADP (but not missing, either) is by definition a miss, this could lead to a situation where... out of lets say 15 eligible TEs to choose from (in the targeted ADP range we are tasked with), during the course of the season, seven do about what they were expected to (by the "hit" criteria, this is a "miss"*), five outperform and three underperform... by my reckoning, you have about a 1-3 chance of getting a value hit... about 1-5 chance of an undervalued hit... than to do better than a 50% score on the basis of one hit and one miss among your two TE value or undervalued picks (basically, two hits = 100%), those odds would be 1/3 X 1/3 OR 1/5 X 1/5? this would entail a degree of difficulty not obvious from the coin flip analogy?
* clearly if we are keeping score, there should be degrees of misses (jeff pasquino alluded to this, i think)... if a value play with TE 8 ADP finishes TE7... that is a far less problematic "miss" than a case where he finished TE30... making a case for adding push to hit and miss designations... maybe even a scale of DEGREES of hits and misses... football largely isn't binary (in some things... either you are on IR or not)... why should our analysis be? imo, our quantitative (and scouting) analysis needs to have a level of sophistication commensurate with the complexity of the subject matter under investigation...
** the +/- range for what constitutes a hit or miss would almost certainly have to be recalibrated by position... for position like WR, a big range (like 10) might make more sense... for a position like TE, that would seem too high... if there are only 15 POSSIBLE (see above), and some of the top ones are probably thrown out (not too many calls of the TE1 being a value play at his ADP )... there could be a case where most of the top 15 fall within their expected ADP... for the few that jumped up from TE15+ range... it is important to note we are barred from making these value picks... also, if after the first top ADP TEs (that are also getting tossed out for the most part), many of the other TEs are just getting reshuffled inside the top 15... than the range within which even hits are judged misses, would nearly overlap with the choices possible to us... this would lead inexorably to a situation where you would almost be wrong no matter what.
*** another problematic aspect to account for in this kind of exercise, is when words are mixed with numbers... even in the same class of getting a hit right... if some say that the player should crush or destroy their ADP, but they don't... does that actually count as a miss?
though david brought up a good point... should a +3 value in top five be weighted more heaviliy in importance, compared to a +5 increase in the 30s-40s...
he brought up another good point, about the temporal aspect of the changing ADP landscape... if you make a call in JULY that hardesty is overvalued at RB40, and in AUGUST harrison goes on IR, hardesty's ADP shoots up to RB20, and he finished RB 12, was that a miss? presumably savvy readers would recognize a substantive material change in the situation and re-rank/draft accordingly...
on the no better than a coin flip take...
a few years ago, i subbed a few times for the weekly IDP upgrades/downgrades... on the upgrades, there were a few simple guidelines... restrict the handful of positional picks to outside a certain range (say top 20 - not much use in the captain obvious point that patrick willis is likely to do well that week...
), that would score above a specified range... it soon became apparent that the odds were often stacked against a high % of hits... in some weeks, out of possibly dozens of DL/LB/DBs, only a few might have satisfied the criteria... getting a couple right and a couple wrong could involve a degree of difficulty not obvious by the seemingly 50% hit rate... use DBs for an example... 32 teams (non-bye week) play 5-6 regularly... but even sticking with 4 starters = 128 possible to be chosen from... if only a few satisfy the criteria in any given week, even a 50% hit rate (on a much more lower percentage chance play) could be much better than appears at first glance...the value/under-value play articles aren't structured the same, but just trying to make the point that 50% could mean different things depending on the context... if a value call not appreciably exceeding ADP (but not missing, either) is by definition a miss, this could lead to a situation where... out of lets say 15 eligible TEs to choose from (in the targeted ADP range we are tasked with), during the course of the season, seven do about what they were expected to (by the "hit" criteria, this is a "miss"*), five outperform and three underperform... by my reckoning, you have about a 1-3 chance of getting a value hit... about 1-5 chance of an undervalued hit... than to do better than a 50% score on the basis of one hit and one miss among your two TE value or undervalued picks (basically, two hits = 100%), those odds would be 1/3 X 1/3 OR 1/5 X 1/5? this would entail a degree of difficulty not obvious from the coin flip analogy?
* clearly if we are keeping score, there should be degrees of misses (jeff pasquino alluded to this, i think)... if a value play with TE 8 ADP finishes TE7... that is a far less problematic "miss" than a case where he finished TE30... making a case for adding push to hit and miss designations... maybe even a scale of DEGREES of hits and misses... football largely isn't binary (in some things... either you are on IR or not)... why should our analysis be? imo, our quantitative (and scouting) analysis needs to have a level of sophistication commensurate with the complexity of the subject matter under investigation...
** the +/- range for what constitutes a hit or miss would almost certainly have to be recalibrated by position... for position like WR, a big range (like 10) might make more sense... for a position like TE, that would seem too high... if there are only 15 POSSIBLE (see above), and some of the top ones are probably thrown out (not too many calls of the TE1 being a value play at his ADP
)... there could be a case where most of the top 15 fall within their expected ADP... for the few that jumped up from TE15+ range... it is important to note we are barred from making these value picks... also, if after the first top ADP TEs (that are also getting tossed out for the most part), many of the other TEs are just getting reshuffled inside the top 15... than the range within which even hits are judged misses, would nearly overlap with the choices possible to us... this would lead inexorably to a situation where you would almost be wrong no matter what.*** another problematic aspect to account for in this kind of exercise, is when words are mixed with numbers... even in the same class of getting a hit right... if some say that the player should crush or destroy their ADP, but they don't... does that actually count as a miss?
Last edited by a moderator:
