Hey, the offensive linemen are the biggest guys on the field, they’re bigger than everybody else, and that’s what makes them the biggest guys on the field. -John Madden
Make sure you get your PSPR up by the weekend this week because I'm going to do my *objective* rankings and we'll see who's system is a better predictor of the rest of the season.
When you noticed that your PSRP had me ranked 8th for two weeks in a row, I would have thought it would have made YOU call its accuracy into question. But, if you choose to continue to defend and use a failed system, by all means I'm going to come up with a better one and turn it into a competition between the two. :)
It definitely seems flawed. I think just looking at Points Forced is a better indication of who the best teams are, but another method could be to incorporate something like standard deviation into it so we can see which teams consistently put up good points rather than having skewed average scores from one or two big weeks (like my Week 1).
Remember this is not a measure of a teams overall competitiveness, rather an estimation of their ranking, based on W-L records, had the points been PA more evenly.
Regardless of what it's supposed to be measuring, it's clearly not working. I'm in 3rd place and have the 3rd highest point total and Tony Sr is in 4th with the 4th highest point total. Why would distributing the PA more equally bump me down to 8th place and him down to 7th place?
The problem is that the system penalizes a team simply for having a low PA total, and rewards teams simply for having a high PA total. John H is 8th for PF, why would leveling out the PA bump him up to 3rd? It doesn't make sense to me. I think the system's broke man.
If the PA were distributed evenly, then the rankings would simply go by PF. If everyone's opponents scored a total of 995.93, then why would I be ranked 8th scoring 1079.98 PF, but John H would be 3rd scoring 964.66?
I think the problem is that in the final formula the PA/Game - Avg. PA/Game penalizes or rewards teams simply for having a big spread between PA/Game and Avg. PA/Game.
"Clearly not working" by making an erroneous *estimation*? I think you're taking this "system" a little too seriously.
What this is saying is that if you take into account the fact that if one have faces a lower PA, then they may not have the W-L record that they currently do, thus putting them lower on the rankings. You can call it penalization, I am just using a formula to articulate a fact we usually start seeing at this point in the season- some teams have suffered greatly because of a large disparity they have experienced in PA. On the flipside. last year, Tony Sr. was consistently a top 3 team, while not putting up competitive PF, but absorbing an extremely low PA.
The theory behind this is clear from the formula, add the difference in PA from the average PA for the league to that teams PF. This takes into account the points they have scored and the disparity they may have experienced in PA.
Is it perfect? Probably not. All it is, is an attempt to take into account matchup history into a players W-L record, given their actual matchups and strength of team.
As I've said before, the sample size gets bigger each week, and I believe the final results at the end of the year will match our intuitions. The sample size is still small. If you don't like this, I mean, disregard it, doesn't matter to me. I'm just doing it because I still believe it is somewhat accurate as time goes, and can help teams gauge their performance in another way, possibly more objective way than the standard W-L record.
:) Not taking it seriously at all. I just like to analyze stuff like this. Yes, I'm aware that makes me a huge nerd, maybe I should have been a systems analyst or something like that.
Not taking it seriously in the sense that I couldn't care less that this system has me ranked 8th. BUT, the fact that it has me ranked 8th made me look at it a little closer, as I would have thought it would have you as well, and I don't think it is accomplishing what you hoped/thought it would.
I was going to leave it alone because I didn't want to be overly critical of something you clearly enjoyed working on, but since you're getting smart, I'll go ahead with my critique.
I look forward to your critical analysis, and seeing how it relates to the claims that I have made. Though, I am inclined to think that it will instead me addressing to claims that have never been made by me. Either way, bring it! :)
:) I understand that you don't claim that the system "predicts" anything or even really ranks anything. But what you said was that the system is... "rather an estimation of their ranking, based on W-L records, had the points been PA more evenly."
And the problem is, it doesn't do that. It doesn't even out the PA, it gives greater than average PA to the teams currently with lower than average PA, and gives lower than average PA to teams that currently have greater than average PA.
Just take an unbiased look at it for a second and look at John H's rank (don't mean to pick on John, but his team happens to be the best example). He's got the 5th lowest PF, yet he jumps up to 3rd in your rankings, when the only thing your rankings manipulate is PA. How can that possibly be accurate? The only way John H's team could possibly be in 3rd place with his current PF is if he had significantly below average PA. The same is true in the opposite way for Tony Sr and I. The only way we would be ranked as low as 7th or 8th with our current PF is if we had vastly above average PA.
So, I'm just not sure what the point is of an alternative rankings system if it doesn't at least have some validity to it.
I think I understand what your goal was (correct me if I'm wrong)- to incorporate PF and PA into the rankings instead of just W/L. I just don't think this system accomplishes what you were trying to do.
I was actually thinking about this today, and what I think needs to happen is to somehow figure out the proper pythagorean expectation formula for our scoring system. This will rank teams based on PF and PA instead of W/L. But I have no idea how to figure out the formula.
Being in politics is like being a football coach. You have to be smart enough to understand the game, and dumb enough to think it's important. -Eugene McCarthy
Make sure you get your PSPR up by the weekend this week because I'm going to do my *objective* rankings and we'll see who's system is a better predictor of the rest of the season.
ReplyDeleteYou dare call into question the accuracy of the PSRP?!
ReplyDeleteWhen you noticed that your PSRP had me ranked 8th for two weeks in a row, I would have thought it would have made YOU call its accuracy into question. But, if you choose to continue to defend and use a failed system, by all means I'm going to come up with a better one and turn it into a competition between the two. :)
ReplyDeleteIt definitely seems flawed. I think just looking at Points Forced is a better indication of who the best teams are, but another method could be to incorporate something like standard deviation into it so we can see which teams consistently put up good points rather than having skewed average scores from one or two big weeks (like my Week 1).
ReplyDeleteOr one or two bad weeks, for that matter...
ReplyDeleteRemember this is not a measure of a teams overall competitiveness, rather an estimation of their ranking, based on W-L records, had the points been PA more evenly.
ReplyDeleteRegardless of what it's supposed to be measuring, it's clearly not working. I'm in 3rd place and have the 3rd highest point total and Tony Sr is in 4th with the 4th highest point total. Why would distributing the PA more equally bump me down to 8th place and him down to 7th place?
ReplyDeleteThe problem is that the system penalizes a team simply for having a low PA total, and rewards teams simply for having a high PA total. John H is 8th for PF, why would leveling out the PA bump him up to 3rd? It doesn't make sense to me. I think the system's broke man.
If the PA were distributed evenly, then the rankings would simply go by PF. If everyone's opponents scored a total of 995.93, then why would I be ranked 8th scoring 1079.98 PF, but John H would be 3rd scoring 964.66?
ReplyDeleteI think the problem is that in the final formula the PA/Game - Avg. PA/Game penalizes or rewards teams simply for having a big spread between PA/Game and Avg. PA/Game.
Am I wrong?
"Clearly not working" by making an erroneous *estimation*? I think you're taking this "system" a little too seriously.
ReplyDeleteWhat this is saying is that if you take into account the fact that if one have faces a lower PA, then they may not have the W-L record that they currently do, thus putting them lower on the rankings. You can call it penalization, I am just using a formula to articulate a fact we usually start seeing at this point in the season- some teams have suffered greatly because of a large disparity they have experienced in PA. On the flipside. last year, Tony Sr. was consistently a top 3 team, while not putting up competitive PF, but absorbing an extremely low PA.
The theory behind this is clear from the formula, add the difference in PA from the average PA for the league to that teams PF. This takes into account the points they have scored and the disparity they may have experienced in PA.
Is it perfect? Probably not. All it is, is an attempt to take into account matchup history into a players W-L record, given their actual matchups and strength of team.
As I've said before, the sample size gets bigger each week, and I believe the final results at the end of the year will match our intuitions. The sample size is still small. If you don't like this, I mean, disregard it, doesn't matter to me. I'm just doing it because I still believe it is somewhat accurate as time goes, and can help teams gauge their performance in another way, possibly more objective way than the standard W-L record.
:) Not taking it seriously at all. I just like to analyze stuff like this. Yes, I'm aware that makes me a huge nerd, maybe I should have been a systems analyst or something like that.
ReplyDelete"Not taking it seriously" haha of course you are not. LOLZ ROFL
ReplyDeleteNot taking it seriously in the sense that I couldn't care less that this system has me ranked 8th. BUT, the fact that it has me ranked 8th made me look at it a little closer, as I would have thought it would have you as well, and I don't think it is accomplishing what you hoped/thought it would.
ReplyDeleteI was going to leave it alone because I didn't want to be overly critical of something you clearly enjoyed working on, but since you're getting smart, I'll go ahead with my critique.
Coming soon. :)
I look forward to your critical analysis, and seeing how it relates to the claims that I have made. Though, I am inclined to think that it will instead me addressing to claims that have never been made by me. Either way, bring it! :)
ReplyDelete*instead be addressing claims that*
ReplyDelete:) I understand that you don't claim that the system "predicts" anything or even really ranks anything. But what you said was that the system is... "rather an estimation of their ranking, based on W-L records, had the points been PA more evenly."
ReplyDeleteAnd the problem is, it doesn't do that. It doesn't even out the PA, it gives greater than average PA to the teams currently with lower than average PA, and gives lower than average PA to teams that currently have greater than average PA.
Just take an unbiased look at it for a second and look at John H's rank (don't mean to pick on John, but his team happens to be the best example). He's got the 5th lowest PF, yet he jumps up to 3rd in your rankings, when the only thing your rankings manipulate is PA. How can that possibly be accurate? The only way John H's team could possibly be in 3rd place with his current PF is if he had significantly below average PA. The same is true in the opposite way for Tony Sr and I. The only way we would be ranked as low as 7th or 8th with our current PF is if we had vastly above average PA.
So, I'm just not sure what the point is of an alternative rankings system if it doesn't at least have some validity to it.
I think I understand what your goal was (correct me if I'm wrong)- to incorporate PF and PA into the rankings instead of just W/L. I just don't think this system accomplishes what you were trying to do.
I was actually thinking about this today, and what I think needs to happen is to somehow figure out the proper pythagorean expectation formula for our scoring system. This will rank teams based on PF and PA instead of W/L. But I have no idea how to figure out the formula.