Photo

BSL: 2015 NFL Player Rankings (including Ravens)


  • Please log in to reply
8 replies to this topic

#1 BSLMikeRandall

BSLMikeRandall

    Sr. Ravens Analyst

  • Members
  • PipPipPipPipPip
  • 19,039 posts

Posted 28 January 2016 - 01:23 PM

BSL: http://baltimorespor...nkings-z-score/

 

Using my Z-Score metric to rank the qualifying NFL players on the 2015 season. Look at how the Ravens measure up. Look at the rookie class. Also a look at Cam Newton, the man who will likely win the league MVP. How did he measre up to his peers. Should he win the award, or should it go to another player? 

 


  • BSLChrisStoner and Mackus like this
@BSLMikeRandall

#2 Mackus

Mackus

    HOF

  • Members
  • PipPipPipPipPip
  • 41,010 posts

Posted 28 January 2016 - 01:51 PM

This is exactly how I do my fantasy football rankings.  Project stats for the whole league (the hardest part, I usually download a few and merge rather than do my own), apply the league scoring settings to the players stats to get fantasy points, then determine how many standard deviations above or below average each player is.

 

It's great for fantasy football, because all that matters is numbers, not their context.

 

It's value to real football is limited much like any offensive stat is, it gives no weight to the context of the stats.  You're also assuming that each of the 4 statistics you look at for each position are all equally as important and the only 4 things that matter.  I've got no idea which stats matter the most, and how they rank compared to others, so this is as good of an attempt as any, but I'd be curious in knowing how you determined the recipe for each position.

 

For baseball, the goal is almost always the same, so you really only have to adjust for park factors.  In basketball, the goal of each possession is almost always the same, so if you only have to account for tempo or make the stats per possession.  For football, there are just so many variables.  The goal of every play is not the same, some are designed to get you the 1 yard you need, others are low percentage chances of getting all of the yards, some WR go deep a lot so you wouldn't expect many catches but would expect high YPC and vice versa.  Some offenses run 60/40 others pass 60/40.  Some teams have great offensive lines that help their backs and QBs produce better, others vice versa.  Some WR get to catch passes from Tom Brady or Aaron Rodgers, others have to have ducks hurtled at them by current-era Peyton Manning or Ryan Tannehill.

 

I have no idea how to resolve all of those concerns with football when trying to compare multiple positions across the league.  I doubt it's possible.  I like seeing people make attempts, though.  It's definitely an interesting analysis, I just wouldn't use it to make any sort of conclusions.


  • Ricker Says, BSLMikeRandall and BSLGabeFerguson like this

#3 BSLMikeRandall

BSLMikeRandall

    Sr. Ravens Analyst

  • Members
  • PipPipPipPipPip
  • 19,039 posts

Posted 28 January 2016 - 02:22 PM

This is exactly how I do my fantasy football rankings.  Project stats for the whole league (the hardest part, I usually download a few and merge rather than do my own), apply the league scoring settings to the players stats to get fantasy points, then determine how many standard deviations above or below average each player is.
 
It's great for fantasy football, because all that matters is numbers, not their context.
 
It's value to real football is limited much like any offensive stat is, it gives no weight to the context of the stats.  You're also assuming that each of the 4 statistics you look at for each position are all equally as important and the only 4 things that matter.  I've got no idea which stats matter the most, and how they rank compared to others, so this is as good of an attempt as any, but I'd be curious in knowing how you determined the recipe for each position.
 
For baseball, the goal is almost always the same, so you really only have to adjust for park factors.  In basketball, the goal of each possession is almost always the same, so if you only have to account for tempo or make the stats per possession.  For football, there are just so many variables.  The goal of every play is not the same, some are designed to get you the 1 yard you need, others are low percentage chances of getting all of the yards, some WR go deep a lot so you wouldn't expect many catches but would expect high YPC and vice versa.  Some offenses run 60/40 others pass 60/40.  Some teams have great offensive lines that help their backs and QBs produce better, others vice versa.  Some WR get to catch passes from Tom Brady or Aaron Rodgers, others have to have ducks hurtled at them by current-era Peyton Manning or Ryan Tannehill.
 
I have no idea how to resolve all of those concerns with football when trying to compare multiple positions across the league.  I doubt it's possible.  I like seeing people make attempts, though.  It's definitely an interesting analysis, I just wouldn't use it to make any sort of conclusions.



I've been wondering how I could apply it to fantasy football, and it sounds like you have it down.

Nothing is perfect. It took a lot of tinkering to basically come up with a group of stats that matter a lot, carry weight, is fair to the different types of QBs, RBs,
or receivers. Something that the end result is the good players have a good score and poor ones have a bad score.

I had thought about weighting things like TDs to make that count more, because a TD directly effects the score of the game. But I don't have the intellect to even know where to begin to make that adjustment in the formula. Adding a coefficient in there. Multiply by 2, 3, 1.5, I don't know.

I just consider it another tool for the box. I mean, Jerry Rice had more best seasons than anyone according to this metric. Tom Brady often topped the lists too. It's probably not perfect. But it works I guess.
@BSLMikeRandall

#4 jkough1

jkough1

    HOF

  • Members
  • PipPipPipPipPip
  • 7,235 posts

Posted 28 January 2016 - 02:25 PM

Gordon was the worst and Gurley was the best in terms of RBs? Was there some discussion about this before the season by some posters?



#5 Mackus

Mackus

    HOF

  • Members
  • PipPipPipPipPip
  • 41,010 posts

Posted 28 January 2016 - 02:40 PM

I've been wondering how I could apply it to fantasy football, and it sounds like you have it down.

 

Yeah this process works great, but the bulk of the impact is if your preseason projections were accurate or not.  That's why I like to use composites so I'm not basing my draft on a few sleepers I really think will be great (or that any on stat projection really likes).  Get the median expectation from everybody, as long as I get lucky and avoid the busts, I'm able to ensure I am getting good value at each pick.  If your projections end up stinking for many of the guys you drafted (like mine this year), then you're gonna struggle even if the method is a really sound one.

 

I've been wondering how I could apply it to fantasy football, and it sounds like you have it down.

Something that the end result is the good players have a good score and poor ones have a bad score.

 

See this is the thing.  You are at least somewhat picking the data so that it matches your pre-conceived notions of who is good and who is bad, using those notions to check your various iterations to gauge which is the best.  This is effectively a version of fitting your question to your desired answer.  Eliminating those notions completely is nearly impossible, but the stat needs to reveal the good players and not the other way around.

 

I had thought about weighting things like TDs to make that count more, because a TD directly effects the score of the game. But I don't have the intellect to even know where to begin to make that adjustment in the formula. Adding a coefficient in there. Multiply by 2, 3, 1.5, I don't know.

 

If I had the time and effort and interest, I think the way to begin would be to look at whole teams and see how certain team stats correlated with scoring lots of points (since you're only looking at offense).  The higher the correlation any one statistic has with points scored, the more important it probably is.  I'd also compare overall numbers and the numbers if I exclude situations that are obviously atypical, like when the margin is 3 scores or more and the team ahead is usually just pounding the ball to run clock and the team behind is throwing all the time and willing to take major turnover risks because they have to.  Once you have a better idea of which statistics lead to scoring, you can break it down into how individuals lead to accumulating that stat for their team.

 

It's really, really difficult with football since there are so many moving pieces and so many different effective strategies.  The stats can't get you nearly as close to a meaningful answer as they can in other sports.  I think the PFF method of trying to grade individuals based on if they did their job or not on each play might be the best way to go about it, but that too contains tons of uncertainty and room for error.


  • BSLMikeRandall likes this

#6 Icterus galbula

Icterus galbula

    Half-Member, Half-Amazing

  • Members
  • PipPipPipPipPip
  • 6,208 posts
  • LocationThe Big Easy

Posted 28 January 2016 - 03:25 PM

I do it too Mackus. There is a free excel file that football guys.com puts out in July and I just edit their own projections to my liking as the season nears. Spits out positional rankings and an overall.
  • Mackus likes this

#7 BSLGabeFerguson

BSLGabeFerguson

    Ravens Analyst

  • Members
  • PipPipPipPipPip
  • 8,063 posts
  • LocationCity of Angels

Posted 30 January 2016 - 03:08 PM

Gordon was the worst and Gurley was the best in terms of RBs? Was there some discussion about this before the season by some posters?

That does ring a bell.


  • BSLRobShields and Icterus galbula like this
@gabefergy

#8 BSLGabeFerguson

BSLGabeFerguson

    Ravens Analyst

  • Members
  • PipPipPipPipPip
  • 8,063 posts
  • LocationCity of Angels

Posted 30 January 2016 - 03:10 PM

This is great analysis Mike. I tend to agree with Mackus that just looking at stats with no context has its limitations, but from a pure production standpoint, this tells a very complete story.


  • BSLChrisStoner and BSLMikeRandall like this
@gabefergy

#9 BSLMikeRandall

BSLMikeRandall

    Sr. Ravens Analyst

  • Members
  • PipPipPipPipPip
  • 19,039 posts

Posted 30 January 2016 - 07:46 PM

This is great analysis Mike. I tend to agree with Mackus that just looking at stats with no context has its limitations, but from a pure production standpoint, this tells a very complete story.



Thanks Gabe.
@BSLMikeRandall




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

Partners