Kool-Aid Larry
Footballguy
#jackiegleason
I honestly hate to be negative, especially since you guys put out all this time and effort, even getting a podcast out the day after thanksgiving, but since you asked......
first of all, I like your idea to broaden the utility of the feature from spitballing random guys, although I also liked the random spitballing, but I've always been fundamentally opposed to that kind of analysis, and have never really put a lot of faith in it.
so, you might throw a rating out there for cle browns vs DL, but on one side there could be a joe thomas, and on the other a chump thomas --- the side my guy plays on will make a pretty big difference, and averaging the numbers really doesn't help anyone.
there is sometimes a common element, like a qb who gets the ball out quick, or a team that runs 40x and throws 20x, so I know peyton manning is a bad match up regardless of side, but this is very hit and miss throughout the league.
to move over to safeties, as examples, there might be a throwing team that's a good fs play, or maybe a running team that gives points to box safeties, etc
whereas points given to LB includes both inside LB as well as 3-4 OLB, who are more like ends.
another problem I have with it is that it's very circumstantial data and has a fairly ephemeral shelf life.
I might remember this wrong, so let's call it a hypothetical, but you might say tenn is a poor match for DL, but they've lost their starting LT, and now the current tackles are both Q tomorrow and might not play.
I'd think if they're down to their 4th and 5th tackles the match up would be substantially more favorable, although peyton could be back there, blahblahblah.
for example, and this is a pretty poor example because det is already giving up sacks --- but I thought jared allen was a good spot play with reiff out, even though he hasn't done much on the year.
and this is all before even considering the small data set of opponents faced that can drastically skew results.
edit: did you mention kc being a good play for corners -- how is that even possible?
I honestly hate to be negative, especially since you guys put out all this time and effort, even getting a podcast out the day after thanksgiving, but since you asked......
first of all, I like your idea to broaden the utility of the feature from spitballing random guys, although I also liked the random spitballing, but I've always been fundamentally opposed to that kind of analysis, and have never really put a lot of faith in it.
so, you might throw a rating out there for cle browns vs DL, but on one side there could be a joe thomas, and on the other a chump thomas --- the side my guy plays on will make a pretty big difference, and averaging the numbers really doesn't help anyone.
there is sometimes a common element, like a qb who gets the ball out quick, or a team that runs 40x and throws 20x, so I know peyton manning is a bad match up regardless of side, but this is very hit and miss throughout the league.
to move over to safeties, as examples, there might be a throwing team that's a good fs play, or maybe a running team that gives points to box safeties, etc
whereas points given to LB includes both inside LB as well as 3-4 OLB, who are more like ends.
another problem I have with it is that it's very circumstantial data and has a fairly ephemeral shelf life.
I might remember this wrong, so let's call it a hypothetical, but you might say tenn is a poor match for DL, but they've lost their starting LT, and now the current tackles are both Q tomorrow and might not play.
I'd think if they're down to their 4th and 5th tackles the match up would be substantially more favorable, although peyton could be back there, blahblahblah.
for example, and this is a pretty poor example because det is already giving up sacks --- but I thought jared allen was a good spot play with reiff out, even though he hasn't done much on the year.
and this is all before even considering the small data set of opponents faced that can drastically skew results.
edit: did you mention kc being a good play for corners -- how is that even possible?
Last edited by a moderator: