Ranking Dynasty RB's: The 80/20 Principle and Quality Years Remain (1 Viewer)

[Englishteacher]

*******NEW Rankings March 14, 2016******

Scroll Down

THIS THREAD IS UNDER CONSTRUCTION. CHECK FOR NEW ADDITIONS.

This thread will attempt to do the following:

1. Give suggestions on how to value your dynasty years (i.e., year one 20%, year two 16%, year three 12.8%) for when you are weighing first year(s) production versus future potential in dynasty play as a whole (even across positions). (For this, see post 13)

"That would break your valuation % by year down like this: 20, 16, 12.8, 10.2, 9.2, 7.4, 5.9, 4.7, 3.8, 3, 2.4, 1.9, 1.5, 1.2."

2. Give a rough approximation of the depreciation of a running back as he progresses through his career in dynasty terms. Or, roughly what percentage of his career contribution he's made in any given year in longevity terms. (for an example, see post 16)

3. Combine the 80/20 principle (below) and Quality Years Remaining with a performance metric to create a RB dynasty ranking. In addition, the 80/20 principle would predict aggregate deterioration of said performance skills and subsequent devaluation through time. See below.

Kudos's to Bruce Hammond for introducing me to Quality Years Remaining. LINK: http://forums.footballguys.com/forum/index.php?showtopic=244365

The 80/20 principle I'm sure many of you have heard of. I believe I picked it up from The One Minute Manager by Johnson & Blanchard. *** Link to The One Minute Manager: http://www.amazon.com/dp/0688014291/?tag=googhydr-20&hvadid=36325575518&hvpos=1t1&hvexid=&hvnetw=g&hvrand=6398134454746903254&hvpone=&hvptwo=&hvqmt=b&hvdev=c&ref=pd_sl_3g6h9xzhgk_b *** Basically, "80% of your good comes from 20% of your work". You know, the most important work you do, about 20% of it, constitutes the bulk of your quality production/output/results. Priorities are important. First things first. Delaying gratification. After allocating your most attention, effort, and resources to the most important things, taking care of the "big rock first rather than the pebbles and sand", its a principle of diminishing returns.

However, who DOESN'T like immediate gratification? I mean, don't you want to win your FF championship in 2014 rather than wait till 2020 when HAL takes over (hope I got that movie reference right)?

So, getting started.

What I've done is combined the QYR and 80/20 principles here and applied them to fantasy football. I'm sure this isn't any different than what you guys already do. But, I took a few minutes and broke down the math in combining the two. Here goes, enjoy!

First, you do want to win in 2014 right? Fill the championship mug up with a cold beverage and grin at all your league mates? That should be priority one. We want our ice cream NOW don't we?

The first year is the year we want to win the most so we assign it the highest value. If you get around to it in a dynasty start up draft, maybe a rule of thumb would be to assign 80% of your resources to guys that will help you win year one. What's the output of your work? A Championship! That's great! But you know what's greater than winning a championship? Winning MULTIPLE championships! But you can't win multiple without getting the first one. So, the percentage of importance of the first championship, in your dynastic run, is 20. You put 80% of your draft picks into winning your first championship and took flyers on some high upside flyers later. The 80/20 rule. Remember, you can't win two, three, four, etc, without the first though. So, subsequent years are not as important as you first ring (or cheap $15 trophy for must of us).

Le's move on, what does this have to do with RB dynasty rankings? Well, following, I'm showing you he math and how it applies to RB's of all ages. RB's are easier to do this with because the run game is easier to master than the pass. You guys know well the 3rd year breakout rule for WR's and TE's and that QB's take time o develop. But, RB's are more plug and play, so this works nicely.

Here goes!

AGE SEASON 1 YEAR VALUATION % AGGREGATE VALUATION CONTRIBUTED %

23 1 20 20

24 2 16 36

25 3 12.8 48.8

26 4 10.2 59

27 5 9.2 68.2

28 6 7.4 75.6

29 7 5.9 81.5

30 8 4.7 86.2

31 9 3.8 90

32 10 3 93

33 11 2.4 95.4

34 12 1.9 97.3

35 13 1.5 98.8

36 14 1.2 100

Obviously, you'd rather have Eddie Lacy than Fred Jackson in dynasty despite vague ballpark production in 2013.

Next, a table showing a RB's physical ability or likelihood of performing to optimum skill capacity through time.

AGE SEASON 1 YEAR DEPRECIATION % AGGREGATE DEPRECIATION % % of ORIGINAL PERFORMANCE CAPACITY

23 1 0 0 100

24 2 1.2 1.2 98.8

25 3 1.5 2.7 97.3

26 4 1.9 4.6 96.4

27 5 2.4 7.0 93

28 6 3.0 10.0 90

29 7 3.8 13.8 86.2

30 8 4.7 18.5 81.5

31 9 5.9 24.4 76.6

32 10 7.4 31.8 68.2

33 11 9.2 41.0 59.0

34 12 10.2 51.2 48.8

35 13 12.8 64.0 36.0

36 14 16.0 80.0 20.0

Later, I will show how you can use the two tables together using a performance metric (individual player grade) in a given year to predict overall remaining value in a career with a specified number that you can compare with other players (rank by highest number).

I'm going to leave this first post off here, let you guys comment, and also work on another post in this thread to fill you in on some ways I'm using this to value RB's.

Have at it!

 

[Englishteacher]

Just to get the ball rolling on some discussion, a 27 year old RB entering in his fifth season will have contributed 68.2% of his career value as opposed to a 23 year old RB in his first year who is giving you his first 20% contribution.

 

[Englishteacher]

If you take a 36 year old RB, he is likely to give you 1.2% the production in his 14th year as he would have been able as a 23 year old rookie. You could look at it from an effectiveness standpoint as well. More 23 year old contribute than 36 year olds. If you want to extrapolate the numbers, it can predict, abstractly, on field production as well because RB's lose production as the get older. This would be a "mileage" or"juice remaining" predictor.

Obviously, you'd expect a 23 year old to be more productive than a 36 year old in the same season.

It's the principle of diminishing returns. Metaphorically, as soon as you drive the car off the lot, it loses value. Speaking about football, we'd rather have a stud rookie RB than a washed up vet.

If you use the above numbers to predict effectiveness or contribution on the same player at different points in his career, you can some up with a predictive model.

Assuming the same player, under islolated or controlled circumstances, gained 1,000 yards in his rookie season at age 23. What yardage total would you guess he'd get at age 36? That's impossible to predict in real life but going by my numbers, I'd estimate his contribution to be 12 yards at age 36 in his 14th season to give an example of how these numbers could be used for projections. He's expected to contribute 1.2% of his "prime" physical self, so 1,000 yards times 1.2% is 12 yards. Very abstract I know but I'm illustrating that the 80/20 principle and QYR can be used together.

Later today, I am going to combine the 80/20 principle, QYR, a player grade from PFF (I won't use the name, only the number), and the overall model to give an assigned value to 1 player based on his individual skill. What this will create is a value across several evaluation techniques to assign a grade to the talent and talent left through the course of his career.

Stay tuned if I haven't lost you already.

I've already come up with a numerical dynasty value for 1 player but it took about a half hour to come up with by hand. My project will be to come up with dynasty rankings based on individual talent using 80/20, QYR, and player grades. The incorporation of player grades gives more than just a way to value your year to year dynasty assignments.

 

[Englishteacher]

Crickets. Bueller?

Let me know if I am speaking another language still. Please, ask questions. I could use the help

Does this help you guys at all? Do you find it interesting? Does it make any sense?

How do you guys value your QYR? I used 80/20 here to say year one is 20%. What figure do you use? 40? 50?

 

[Dinsy Ejotuz]

Look up work on a similar topic by ZWK. He's got a spreadsheet posted here that looks at expected FF value remaining as a function of most recent season and age. It's a more sophisticated version of what you're doing here (I think), using real world data. He did a great job with it.

 

[Christo]

So should I draft Adrian Peterson or Eddie Lacy?

 

[lod01]

How do LeVeon Bell and Giovani Bernard look?

 

[Englishteacher]

So should I draft Adrian Peterson or Eddie Lacy?

How do LeVeon Bell and Giovani Bernard look?

Guys, have only run the data on one player so far as I just developed the idea. I'll be able to come up with a quantifiable dynasty ranking with more time. I could develop a spreadsheet but that would take time. I did the first player, a RB on my six man dynasty squad, by hand with a calculator. Obviously that's not going to work over 50-60 prospects time wise. It'll be much easier to do once the idea is finished and I have a spreadsheet. The early work on this is very labor intensive. However, the math on my real life test subject added up. I think this works quite well.

What we have so far is a way to value years for your dynasty rankings and a system that can incorporate actual player performance grades to predict total value over a career at a given point in time (based on individual talent, not situation).

What I expect is that a player like Zac Stacy, predicted to have a big 2014, will be devalued due to his lack of overall talent in the long run. That's not saying he won't have a big 2014 but you don't want to invest in a player heavily that won't have lasting value.

I expect quite a few surprises in the way this shakes out compared to a lot of dynasty rankings. The 80/20 principle is sound enough that it can cross over into football and I liked the way the math added up as displayed in my first post. My next post will be to display the math involved in the real life test subject I ran and open up the floor to criticism and discussion.

To you guys' point though, I will have a dynasty ranking in time.

 

[Englishteacher]

Look up work on a similar topic by ZWK. He's got a spreadsheet posted here that looks at expected FF value remaining as a function of most recent season and age. It's a more sophisticated version of what you're doing here (I think), using real world data. He did a great job with it.

I certainly appreciate the tip. If you have the link bookmarked or did a search given what keywords you know on the topic, I'd enjoy the link. I'll search on my own as well.

I'm wondering, did he combine this with performance metrics at all?

What I've done so far is not rocket science. It was merely a way for me to value dynasty years using the 80/20 principle, an idea I like a lot. I think it came up with a relatively decent way, for a layperson, to value dynasty years and career value.

For instance, my real life test subject that I've done the math on, is 27 years old and will be playing his fifth season. I determined that at the conclusion of the 2014 season, he will have contributed 68.2% of his career value to me.

Next, I've combined a skill performance metric to these numbers that I'll be able to compare him to other players with.

I'll show the math on the test player and I'd be happy to hear your thoughts.

Again, thanks for your input. The crickets were getting loud in here and I was beginning to think nobody understood my pig latin

 

[ZWK]

Look up work on a similar topic by ZWK. He's got a spreadsheet posted here that looks at expected FF value remaining as a function of most recent season and age. It's a more sophisticated version of what you're doing here (I think), using real world data. He did a great job with it.

Thanks, Rob. That work is here.

 

[Englishteacher]

Look up work on a similar topic by ZWK. He's got a spreadsheet posted here that looks at expected FF value remaining as a function of most recent season and age. It's a more sophisticated version of what you're doing here (I think), using real world data. He did a great job with it.

Thanks, Rob. That work is here.

Will be checking this out for sure.

 

[Englishteacher]

First, if you look at "Aggregate Valuation Contributed", you can predict what percentage of "good" or a player's career he has contributed. At age 36, highly improbable (but possible by Jackson or Sproles), he will have contributed 100% of his career. In year one, age 23, he will have contributed 20% of his career. The 20% (of the 80/20 principle) is multiplied by 80% for Year 2, giving you, .2 X .8 = .16. 20% + 16% = 36%. Or, by year two, he has finished 36% of his career.

AGE SEASON 1 YEAR VALUATION % AGGREGATE VALUATION CONTRIBUTED %

23 1 20 20

24 2 16 36

25 3 12.8 48.8

26 4 10.2 59

27 5 9.2 68.2

28 6 7.4 75.6

29 7 5.9 81.5

30 8 4.7 86.2

31 9 3.8 90

32 10 3 93

33 11 2.4 95.4

34 12 1.9 97.3

35 13 1.5 98.8

36 14 1.2 100

 

[Englishteacher]

Additionally, when planning a dynasty and you're wondering how to value given years, this works as well.

What do you guys think of placing 20% of your emphasis on year 1? 16% on year two? And, continuing to take 80% of that for each subsequent year?

That would break your valuation down like this by year: 20, 16, 12.8, 10.2, 9.2, 7.4, 5.9, 4.7, 3.8, 3, 2.4, 1.9, 1.5, 1.2.

23 1 20 20

24 2 16 36

25 3 12.8 48.8

26 4 10.2 59

27 5 9.2 68.2

28 6 7.4 75.6

29 7 5.9 81.5

30 8 4.7 86.2

31 9 3.8 90

32 10 3 93

33 11 2.4 95.4

34 12 1.9 97.3

35 13 1.5 98.8

36 14 1.2 100

 

[Tanner9919]

keep up the good work,OP..!

I"m not sure what it all means and where it's headed, but one thing that hasn't been factored in is

changes in coaching philosophy, changes in coaching staff from year to year, QB capability ( D. Martin did nothing with Freeman at QB but his replacements found gold with Glennon at QB), changes in O-line ( see Ravens last year vs 2012), changes in the team a given RB plays for - see Thomas Jones and how he put it all together late in his career - changes in division opponents' defenses - the AFC East is getting stronger defensively, while the NFC West 'arrived' a couple of years ago,and the NFC East is in decline and in full disarray..

number crunching is great and all,but there are so many outside factors that determine things, that you might be doing this all in vain..

Ben Tate should do better in Cle than he did in Houston, simply because he's the starter now.

conversely, it's a LOCK that once a RB hits the 370-carry mark in a season, he's finished.

see LJ, Dickerson, J. Anderson,etc..add to that list one Arian Foster.

I think ADP does better this year, with Norv Turner ,than he did last year..

 

[Englishteacher]

keep up the good work,OP..!

I"m not sure what it all means and where it's headed, but one thing that hasn't been factored in is

changes in coaching philosophy, changes in coaching staff from year to year, QB capability ( D. Martin did nothing with Freeman at QB but his replacements found gold with Glennon at QB), changes in O-line ( see Ravens last year vs 2012), changes in the team a given RB plays for - see Thomas Jones and how he put it all together late in his career - changes in division opponents' defenses - the AFC East is getting stronger defensively, while the NFC West 'arrived' a couple of years ago,and the NFC East is in decline and in full disarray..

number crunching is great and all,but there are so many outside factors that determine things, that you might be doing this all in vain..

Ben Tate should do better in Cle than he did in Houston, simply because he's the starter now.

conversely, it's a LOCK that once a RB hits the 370-carry mark in a season, he's finished.

see LJ, Dickerson, J. Anderson,etc..add to that list one Arian Foster.

I think ADP does better this year, with Norv Turner ,than he did last year..

Appreciate it very much. If you have any questions about any points, explanations, or applications, shoot.

Of course, all your above points are very true. I'm more going for how to value dynasty years, depreciation of a running back, and overall talent level (which I think is the best predictor of performance in the long run). You've certainly added to the conversation though which I'm happy about. Your comments, which I'd like to discuss more, illustrate many blind spots and why you wouldn't be able to rely on this model when finished 100% without some tweaking for your individual final rankings. My stuff is more a starting point or guideline that gives you valuation of a point of a RB's career combined with his overall talent level.

 

[Englishteacher]

keep up the good work,OP..!

I"m not sure what it all means and where it's headed, but one thing that hasn't been factored in is

changes in coaching philosophy, changes in coaching staff from year to year, QB capability ( D. Martin did nothing with Freeman at QB but his replacements found gold with Glennon at QB), changes in O-line ( see Ravens last year vs 2012), changes in the team a given RB plays for - see Thomas Jones and how he put it all together late in his career - changes in division opponents' defenses - the AFC East is getting stronger defensively, while the NFC West 'arrived' a couple of years ago,and the NFC East is in decline and in full disarray..

number crunching is great and all,but there are so many outside factors that determine things, that you might be doing this all in vain..

Ben Tate should do better in Cle than he did in Houston, simply because he's the starter now.

conversely, it's a LOCK that once a RB hits the 370-carry mark in a season, he's finished.

see LJ, Dickerson, J. Anderson,etc..add to that list one Arian Foster.

I think ADP does better this year, with Norv Turner ,than he did last year..

Since you're the second person who has mentioned ADP (who is 29 years old), using my model, he'll have contributed 81.5% of his career value by year's end. See below for the bolded, corresponding season:

AGE SEASON 1 YEAR VALUATION % AGGREGATE VALUATION CONTRIBUTED %

23 1 20 20

24 2 16 36

25 3 12.8 48.8

26 4 10.2 59

27 5 9.2 68.2

28 6 7.4 75.6

29 7 5.9 81.5

30 8 4.7 86.2

31 9 3.8 90

32 10 3 93

33 11 2.4 95.4

34 12 1.9 97.3

35 13 1.5 98.8

36 14 1.2 100

 

[Englishteacher]

One more shameful bump for the lunch crowd.

 

[Gandalf]

I think it's nice to have a system. But player evaluation is more an art than a science. How many years a player has left is so dependent on the individual person. I mean Barry Sanders checked out early just because. Ray Rice completely fell apart. Meanwhile Fred Jackson has one awful injury after another and just keeps coming back for more. To say a player at 30 has used up 86.2% (really not 86.7% or 88%?) can lead to wrong decisions.

I dunno. I think while interesting this could be an exercise in madness and yields little in terms of actual help with building your fantasy team.

 

[az_prof]

I think it's nice to have a system. But player evaluation is more an art than a science. How many years a player has left is so dependent on the individual person. I mean Barry Sanders checked out early just because. Ray Rice completely fell apart. Meanwhile Fred Jackson has one awful injury after another and just keeps coming back for more. To say a player at 30 has used up 86.2% (really not 86.7% or 88%?) can lead to wrong decisions.

I dunno. I think while interesting this could be an exercise in madness and yields little in terms of actual help with building your fantasy team.

I tend to agree.

 

[Englishteacher]

I think it's nice to have a system. But player evaluation is more an art than a science. How many years a player has left is so dependent on the individual person. I mean Barry Sanders checked out early just because. Ray Rice completely fell apart. Meanwhile Fred Jackson has one awful injury after another and just keeps coming back for more. To say a player at 30 has used up 86.2% (really not 86.7% or 88%?) can lead to wrong decisions.

I dunno. I think while interesting this could be an exercise in madness and yields little in terms of actual help with building your fantasy team.

I tend to agree.

You guys are correct. But, when projecting into the future it gets hazy anyway. Sure, 86.2% exactly is an absurd number, its a rough estimate. Also, it'd be good to reevaluate every year with performance metrics to see drop offs and then use what you know about the players to adjust accordingly.

What do you guys think about the year valuations for dynasty mentioned in Post 13 for dynasty planning? That would break your valuation down like this by year: 20, 16, 12.8, 10.2, 9.2, 7.4, 5.9, 4.7, 3.8, 3, 2.4, 1.9, 1.5, 1.2.

 

[Englishteacher]

keep up the good work,OP..!

I"m not sure what it all means and where it's headed, but one thing that hasn't been factored in is

changes in coaching philosophy, changes in coaching staff from year to year, QB capability ( D. Martin did nothing with Freeman at QB but his replacements found gold with Glennon at QB), changes in O-line ( see Ravens last year vs 2012), changes in the team a given RB plays for - see Thomas Jones and how he put it all together late in his career - changes in division opponents' defenses - the AFC East is getting stronger defensively, while the NFC West 'arrived' a couple of years ago,and the NFC East is in decline and in full disarray..

number crunching is great and all,but there are so many outside factors that determine things, that you might be doing this all in vain..

Ben Tate should do better in Cle than he did in Houston, simply because he's the starter now.

conversely, it's a LOCK that once a RB hits the 370-carry mark in a season, he's finished.

see LJ, Dickerson, J. Anderson,etc..add to that list one Arian Foster.

I think ADP does better this year, with Norv Turner ,than he did last year..

These will be talent/performance rankings based on individual player grades. I'd rather have a true talent in dynasty than a 1 year player in a great situation.

 

[Englishteacher]

I think it's nice to have a system. But player evaluation is more an art than a science. How many years a player has left is so dependent on the individual person. I mean Barry Sanders checked out early just because. Ray Rice completely fell apart. Meanwhile Fred Jackson has one awful injury after another and just keeps coming back for more. To say a player at 30 has used up 86.2% (really not 86.7% or 88%?) can lead to wrong decisions.

I dunno. I think while interesting this could be an exercise in madness and yields little in terms of actual help with building your fantasy team.

I think Fred Jackson is trying to make enough money in the NFL to feed all the poor in the world.

 

[Englishteacher]

Hello folks! These are some preliminary findings of the top 24 or so graded runners by PFF in 2015 ranked for FF dynasty purposes according to my formula.

Player / Dynasty Value

Todd Gurley          94.84

T.J. Yeldon            83.21

Thomas Rawls       81.06

Karlos Willaims      77.96

Duke Johnson        77.14

Leveon Bell            75.14

Devonta Freeman  63.46

Lamar Miller           54.73

Carlos Hyde           50.37

Eddie Lacy             50.32

Spencer Ware        50.18

Giovani Bernard     49.04

Charles Sims          43.05

Dion Lewis              40.66

Doug Martin            35.94

Chris Ivory               25.77

Lesean McCoy        25.03

Jonathan Stewart    20.35

Marshawn Lynch    16.06 (Retired)

Tim Hightower        15.21

Jamaal Charles      14.59

Matt Forte              11.27

Rashad Jennings   10.88

DeAngello Williams 5.88

Still about 125 other 2015 graded runners to put into the spreadsheet; but this is a teaser for how the top runners in 2015 fit into the formula.

Enjoy! Cheers!

 

[Arodin]

Looks high on Yeldon and Duke, low on Bell, Miller and Hyde (unless situation is being factored in).

Interesting early list.

 

[Dr. Octopus]

Hello folks! These are some preliminary findings of the top 24 or so graded runners by PFF in 2015 ranked for FF dynasty purposes according to my formula.

Player / Dynasty Value

Todd Gurley          94.84

T.J. Yeldon            83.21

Thomas Rawls       81.06

Karlos Willaims      77.96

Duke Johnson        77.14

Leveon Bell            75.14

Devonta Freeman  63.46

Lamar Miller           54.73

Carlos Hyde           50.37

Eddie Lacy             50.32

Spencer Ware        50.18

Giovani Bernard     49.04

Charles Sims          43.05

Dion Lewis              40.66

Doug Martin            35.94

Chris Ivory               25.77

Lesean McCoy        25.03

Jonathan Stewart    20.35

Marshawn Lynch    16.06 (Retired)

Tim Hightower        15.21

Jamaal Charles      14.59

Matt Forte              11.27

Rashad Jennings   10.88

DeAngello Williams 5.88

Still about 125 other 2015 graded runners to put into the spreadsheet; but this is a teaser for how the top runners in 2015 fit into the formula.

Enjoy! Cheers!

So this formula seems to leave out a player's role. Yeldon's actual production just likely took a hit with the signing of Ivory and backup RBs like Carlos Williams or Spencer Ware and 3rd down backs like Duke Johnson or Charles Sims seem to get overvalued.

 

[spider321]

Ewwww.

Time to get a new formula.

 

[Englishteacher]

Looks high on Yeldon and Duke, low on Bell, Miller and Hyde (unless situation is being factored in).

Interesting early list.

Yeldon was helped by being a Top 10 runner and his age. Duke was also graded high and was young. Situations for both are less than enviable at the current moment.

Bell was the top ranked runner by PFF but is a bit older than, say, Gurley, who was very young. Bell will of course benefit from a high-octane Steelers' offense.

Miller was also a Top 10 runner but a little older.

Hyde simply didn't have the graded talent to be further up the list although he should get every carry he can handle in SF. 

All noteable comments and I appreciate the critique. This formula grades 100% on talent at the moment and it will be interesting to see how it would shake out in the long run. Regardless of the shock factor it carries, it is against the grain of most rankings, it would be a great list to "trade down" on in a startup, and accumulate picks to draft against an expected ADP of many players. You're going to pay a premium for Bell, Miller, and Hyde while in six months or a year, when some of the head-scratchers break out, you'd be well ahead of the curve, stronger at other positions, and have two handfuls of RB studs rather than a marginalized and leveraged, say, Bell and Miller. Over the course of your dynasty, would you rather have all your eggs in one basket with Bell and Miller? Or, have a selection of studs like Yeldon, Duke, Thomas Rawls, Karlos Williams, and Spencer Ware whose situations just haven't materialized yet? The talent is there on those players though.

 

[Englishteacher]

So this formula seems to leave out a player's role. Yeldon's actual production just likely took a hit with the signing of Ivory and backup RBs like Carlos Williams or Spencer Ware and 3rd down backs like Duke Johnson or Charles Sims seem to get overvalued.

Yes, this is a 100% talent and age list for dynasty.

 

[Englishteacher]

Yes, this is a 100% talent and age list for dynasty.

Ivory is older. McCoy may be on the decline. Charles is on the decline. While Duke and Sims are stronger as receivers than runners, their PFF  grade indicates they hold enough on their own in both to be very good.

With a break or two, all of these guys are likely future studs.

If you're talking dynasty investments, not bad at all. And, much better than other marginal players, i.e. Langford and Matt Jones, that are presently being hyped and overvalued that will be replaced before the train really leaves the station.

Think Tre Mason and Zac Stacy, neither had the true talent to be long term dynasty producers yet many FF players invested in them hook, line and sinker, only to lose their shirts because they "had a window". 

 

[fruity pebbles]

David Johnson?

 

[Biabreakable]

In the case of Tre Mason I do think he is capable of being a featured RB. If you look at Jeff Fishers draft history (as I did prior to Zac Stacy getting the opportunity briefly) you see that with the Titans he used high picks at RB every year after George retired until they landed Chris Johnson who became their full time starter for awhile.

1996 George 14th overall

Zero RB drafted until 2000 where Mike Green is drafted in the 7th round. Then no RB were drafted until 2004 where they took Troy Flemming (more of a H back iirc) in the 6th round.

That is nine years of not using more than a 6th round pick at the RB position.

2005 they took Damien Nash in the 5th round even though George retired. I think they had a free agent temporary solution in Travis Henry.

2006 2nd round pick LenDale White. 2007 2nd round pick Chris Henry.2008 1st round pick Chris Johnson.

There is another nice thread I think started by Gianmarco looking at RB careers by draft position. What he found is that most non 1st round RB are not drafted to be starters or featured RB but to be depth or COP RB. There are several exceptions to this of course, but I wouldn't look at Mason as being a failure. 

Mason had some pretty serious deficiencies in pass protection that he has had to work on as well. That limitation reduced how often he could be counted on as a starter and for roles in a RBBC.

All of that is not to say that people should have expected Mason to be the next Eddie George or Chris Johnson. From a talent perspective those players have talent, in Zac's case it is proven talent. Talent however that is defined is obviously not enough to become a long term starter in the NFL. See Joseph Randle and Christine Michael.

Overall I appreciate the thread and the ideas that went into it. Not sure about the math and calculations. I do like that you end up with some numbers that show some clear separation between players, just have not dug into what is behind those calculations enough to make sense of your method yet.

 

[Arodin]

Yeldon was helped by being a Top 10 runner and his age. Duke was also graded high and was young. Situations for both are less than enviable at the current moment.

Bell was the top ranked runner by PFF but is a bit older than, say, Gurley, who was very young. Bell will of course benefit from a high-octane Steelers' offense.

Miller was also a Top 10 runner but a little older.

Hyde simply didn't have the graded talent to be further up the list although he should get every carry he can handle in SF. 

All noteable comments and I appreciate the critique. This formula grades 100% on talent at the moment and it will be interesting to see how it would shake out in the long run. Regardless of the shock factor it carries, it is against the grain of most rankings, it would be a great list to "trade down" on in a startup, and accumulate picks to draft against an expected ADP of many players. You're going to pay a premium for Bell, Miller, and Hyde while in six months or a year, when some of the head-scratchers break out, you'd be well ahead of the curve, stronger at other positions, and have two handfuls of RB studs rather than a marginalized and leveraged, say, Bell and Miller. Over the course of your dynasty, would you rather have all your eggs in one basket with Bell and Miller? Or, have a selection of studs like Yeldon, Duke, Thomas Rawls, Karlos Williams, and Spencer Ware whose situations just haven't materialized yet? The talent is there on those players though.

This makes sense, and being a non-traditional way to look at things, is worthy of consideration. 

It does seem however to weaken the "100% talent" approach in also weighing age factors and downgrading older backs.  While that may impact "pure talent" for backs nearing the end of their shelf life like Forte or Foster, it doesn't seem likely that Miller, for instance, is going to drop in talent due to age this coming season.

Appreciate your willingness to engage constructively with critiques of the rankings.  It helps refine what exactly you're measuring, and how to balance it with other factors, like situation.

 

[Biabreakable]

What is your definition of talent as it pertains to these rankings?

Talent is a word very subject to opinion and there are many shades of grey. I am never quite clear what people are talking about when they are using the word talent in regards to football players.

Perhaps a better definition of talent could lead to an improved analysis? To get the right answers we have to ask the right questions.

 

[Englishteacher]

What is your definition of talent as it pertains to these rankings?

Talent is a word very subject to opinion and there are many shades of grey. I am never quite clear what people are talking about when they are using the word talent in regards to football players.

Perhaps a better definition of talent could lead to an improved analysis? To get the right answers we have to ask the right questions.

Hi Biabreakable. For talent, I simply used player grades from the 2015 season from PFF. Then inserted them into a formula, above in post 1, that quantifies talent applied, a career, for dynasty purposes. While subjective, the PFF grades have a criteria if you are familiar with them.

While Leveon Bell had the highest talent grade for the 2015 season, Todd Gurley, who is considerably younger but somewhat comparable in talent, or "player grade", received a higher overall dynasty grade because of what he'll do in their projected remaining careers.

 

[Englishteacher]

David Johnson?

Haven't finished the spreadsheet yet. Only the Top 24 graded runners per PFF in 2015, of which, David Johnson was not.

 

[Biabreakable]

Ah ok PFF grades.

Could you give me an example of a player, what their PFF grade was and how you calculated this player for dynasty based off of the PFF grade? 

I am not sure what you are combining their grades with besides age?

How would you calculate fantasy points from a PFF grade?

 

[Englishteacher]

In the case of Tre Mason I do think he is capable of being a featured RB. If you look at Jeff Fishers draft history (as I did prior to Zac Stacy getting the opportunity briefly) you see that with the Titans he used high picks at RB every year after George retired until they landed Chris Johnson who became their full time starter for awhile.

1996 George 14th overall

Zero RB drafted until 2000 where Mike Green is drafted in the 7th round. Then no RB were drafted until 2004 where they took Troy Flemming (more of a H back iirc) in the 6th round.

That is nine years of not using more than a 6th round pick at the RB position.

2005 they took Damien Nash in the 5th round even though George retired. I think they had a free agent temporary solution in Travis Henry.

2006 2nd round pick LenDale White. 2007 2nd round pick Chris Henry.2008 1st round pick Chris Johnson.

There is another nice thread I think started by Gianmarco looking at RB careers by draft position. What he found is that most non 1st round RB are not drafted to be starters or featured RB but to be depth or COP RB. There are several exceptions to this of course, but I wouldn't look at Mason as being a failure. 

Mason had some pretty serious deficiencies in pass protection that he has had to work on as well. That limitation reduced how often he could be counted on as a starter and for roles in a RBBC.

All of that is not to say that people should have expected Mason to be the next Eddie George or Chris Johnson. From a talent perspective those players have talent, in Zac's case it is proven talent. Talent however that is defined is obviously not enough to become a long term starter in the NFL. See Joseph Randle and Christine Michael.

Overall I appreciate the thread and the ideas that went into it. Not sure about the math and calculations. I do like that you end up with some numbers that show some clear separation between players, just have not dug into what is behind those calculations enough to make sense of your method yet.

I agree on Tre Mason. In fact, I owned him in a six man keeper-dynasty and was excited about his prospects. However, he was on the cusp in analysis like this, BEFORE Gurley's drafting of being cut by me in my league because I identified him as a guy that could be replaced as a starter in either the draft or free agency.

It was a tough decision that came down to either holding Carlos Hyde, behind Frank Gore at the time, or Mason, who had a year's worth of starter's experience and stats and ranked higher by most experts in dynasty. Where are they now?

Further, this list helps identify posers that we are going to identify as high ranking dynasty assets for the short term solutions at an NFL team's position group that they are.

I'll throw out a couple of examples of players that I think are being highly overvalued in dynasty right now, Matt Jones and Jeremy Langford. Both are young and have beelines to starting jobs. Great fantasy asset, right? The next Tre Mason's. They're going to be replaced, if not this season, soon enough. WHat you sacrifice in perhaps some short term gain using this system is less volatility in the long run because you draft better talent, over situation.

 

[Englishteacher]

Ah ok PFF grades.

Could you give me an example of a player, what their PFF grade was and how you calculated this player for dynasty based off of the PFF grade? 

I am not sure what you are combining their grades with besides age?

How would you calculate fantasy points from a PFF grade?

That will take me a little bit but I am on it.

 

[Englishteacher]

That will take me a little bit but I am on it.

Last    First    PFF Grade  20        21    22    23      24         25        26         27         28         29         30         31       32        33       34    35    36    
                                         0.2       0.2    0.2   0.16  0.128    0.102    0.092    0.074    0.059    0.047    0.038    0.03    0.024    0.019    0.015    0.012    Dynasty Value
Bell    Leveon    94            0          0       0    15.04  12.03    9.58    8.64    6.95         5.54        4.41    3.57      2.82    2.25    1.78    1.41    1.12             75.14

Sorry about the formatting but if you read the first few posts and familiarize yourself with some of the concepts, it will help.

Leveon Bell received the top overall 2015 player grade by PFF. It was a 94.0. See left. 

Player ages are across the top. 20, 21, 22, 23, 24, 25, ...... 36.

Below player ages are compuations based on the 80/20 rule, or Pareto Principle, (see OPost), in which, a player's overall dynasty value decreases the older he gets.

The 0.2, 0.2, 0.2, 0.16, 0.128, 0.092 going across to and ending in 0.012 are multipliers for determining a single's year's value. I'll explain that. Hold on.

Go down from age 23 for Bell. You multiply his player grade (94)  times .16, or 16%,and it gives you a total of his age 23 year old year in the league (which will be 2016), or 15.04.

The math: 94 X 16% = 15.04... this is Bell's total contribution to his career dynasty value. It will happen in one year, 2016, when he is 23 years old. IN other words, 16% of his dynasty value will occur in 2016.

If you add up all the individual years value, i.e., 15.04 + 12.03 + 9.58 + 8.64....................... + 1.12 = 75.14 or Total Dynasty Value.

Please fire away with questions.
 

 

[Englishteacher]

Haven't finished the spreadsheet yet. Only the Top 24 graded runners per PFF in 2015, of which, David Johnson was not.

While he had a quality rookie season, I think from my analysis, Johnson is overvalued right now in Dynasty Circles. I'm seeing him go #3 behind Gurley and Bell and while he is decently talented and has a good situation I bet he doesn't live up to that over the course of his career. I'd expect to see Arizona either bring in a complement or replacement either this year or soon enough.

 

[fruity pebbles]

Very interesting reading, thanks for the time put into this.

 

[Englishteacher]

This makes sense, and being a non-traditional way to look at things, is worthy of consideration. 

It does seem however to weaken the "100% talent" approach in also weighing age factors and downgrading older backs.  While that may impact "pure talent" for backs nearing the end of their shelf life like Forte or Foster, it doesn't seem likely that Miller, for instance, is going to drop in talent due to age this coming season.

Appreciate your willingness to engage constructively with critiques of the rankings.  It helps refine what exactly you're measuring, and how to balance it with other factors, like situation.

I agree, and I'll come back to this further.

I tend to favor talent over situation in dynasty. That's the short answer. For redraft, that's a different beast.

 

[Englishteacher]

Very interesting reading, thanks for the time put into this.

Appreciate it.

 

[ZWK]

I think your formula is putting too much weight on good performance over a small sample size. Three reasons to be concerned about players who had good production on a small sample size:

1. Small sample sizes are less predictive. It's a lot easier to have your numbers inflated by good luck (of various sorts) on 100 touches than on 500 touches. And a single year is not enough data to throw out what you thought before that season: it's not that uncommon for RBs to bust after one big year (e.g., Spiller, Hills, Derrick Ward) or to bounce back after a down year (e.g., Forte, Martin).

2. Role players: a player who is really good in a particular role won't necessarily be good at taking on a bigger role. Darren Sproles has put up great efficiency numbers as a runner (career 5.0 YPC), but he's never had more than 100 carries in a season because he is only a good runner in a relatively narrow set of situations.

3. Ability to carry the load: some players can't hold up to a large workload. Often their coaches know this, and don't give them a big workload.

 

[Biabreakable]

Englishteacher thanks for the explanation.

Seems simple enough. Use PFF end of year grades that are compiled from from their weekly grades. Then apply the deprecation by age as a percentage, starting with age 23 at 100% and diminishing from that point forward. For ages 20 to 23 there is no depreciation. This suggests that the players value begins to decline at 24 years old simply because they have one less year remaining in their entire career moreso than a decline because of age. The discount should become more dramatic than just one less year at age 29 for a RB.

What I am not clear on is how you did this calculation. Because in your Bell example you are talking about multiplying by 16% (or .16).

In the original post you say:

AGE SEASON 1 YEAR DEPRECIATION % AGGREGATE DEPRECIATION % % of ORIGINAL PERFORMANCE CAPACITY

23 1 0 0 100

24 2 1.2 1.2 98.8

25 3 1.5 2.7 97.3

26 4 1.9 4.6 96.4

27 5 2.4 7.0 93

28 6 3.0 10.0 90

29 7 3.8 13.8 86.2

30 8 4.7 18.5 81.5

31 9 5.9 24.4 76.6

32 10 7.4 31.8 68.2

33 11 9.2 41.0 59.0

34 12 10.2 51.2 48.8

35 13 12.8 64.0 36.0

36 14 16.0 80.0 20.0

So I am not sure where this 20% number I see you using as a possible baseline or something?

In any case wouldn't Bells value for 2016 be 98.8% of his age 23 value? So wouldn't this be 92.872 when you multiply the percentage with the PFF grade? Why are you multiplying by 16%?

The PFF grade is not a career grade. But if you assume that most players on average follow this curve of production by age over the course of their careers then you could apply the percentage discount to any kind of ranking. It will not tell you how PFF will grade these players in this coming year however or how they will perform in FF.

It would be interesting to look at how closely PFF grades are to FF performance. Without having fantasy points projected it makes it difficult to compare players based on performance. While Clinton Portis may have graded very high in PFF rankings because of his blocking and other qualities they grade on, he wasn't putting up the same numbers in Washington as he was with Denver.

I think draft position is another important data point to take into consideration. Of all of the different metrics and ways you can evaluate players, draft position seems to be the most predictive.

 

[Englishteacher]

I think your formula is putting too much weight on good performance over a small sample size. Three reasons to be concerned about players who had good production on a small sample size:

1. Small sample sizes are less predictive. It's a lot easier to have your numbers inflated by good luck (of various sorts) on 100 touches than on 500 touches. And a single year is not enough data to throw out what you thought before that season: it's not that uncommon for RBs to bust after one big year (e.g., Spiller, Hills, Derrick Ward) or to bounce back after a down year (e.g., Forte, Martin).

2. Role players: a player who is really good in a particular role won't necessarily be good at taking on a bigger role. Darren Sproles has put up great efficiency numbers as a runner (career 5.0 YPC), but he's never had more than 100 carries in a season because he is only a good runner in a relatively narrow set of situations.

3. Ability to carry the load: some players can't hold up to a large workload. Often their coaches know this, and don't give them a big workload.

Thanks for the comments ZWK.

1. I agree small sample sizes are less predictive. What's less predicitve, I feel, is actual production and fantasy points than a measured performance using a grading system. To me, this is a more important barometer for long term success than opportunity (guy in front of you goes down and you get a shot), situation (you have a great offensive line or QB, which can change), or offensive philosophy ( run oriented team versus pass oriented team). To answer your original point on small sample sizes, I had thought that once PFF had three years of doing Player Grades (this was their first year doing grades as opposed to Premium Statistics) of doing a 3 year average weighted more strongly towards the most recent year, similar to how FF players will sometimes average three years worth of fantasy points (this would help negate a players lagging grade due to an injury like an ACL tear or an inflated grade perhaps too). Also, players in roles are able to specialize, but PFF uses a variety of factors in determining their original grade (run, receiving, blocking, etc.). In addition, the player grade should be more predictive than it's predecessor, premium stats, as it's more a measure of skill than amasses opportunity and compiling stats. In fact, what my model does is helps a FF player be able to abstractly "plug and play" and visualize what a certain player might be able to do given opportunity.

To support the above, I'd be buying on players like TJ Yeldon (owners scared of Chris Ivory) and selling on those like Jeremy Langford and Matt Jones (not that talented and dynasty value will drop once competition or a replacement is brought in). 

2. While I agree that TRUE role players  are tricky in valuing, the really good ones will put up substantial fantasy years too. To your point, Darren Sproles has put up back end RB1 numbers in his career. Another specialist, Danny Woodhead, put up RB 1 numbers in my league in 2015.

3. This is true. I agree. However, talented players will have work over non-talented ones, giving them value.

I'm not saying this list is an end all or I'm a messiah; I think it's a vastly more informative starting point however than actual production where we see a lot of posers to long term success that too many FF players (myself included) take the bait on. With hope, this approach should be a starting point rather than thinking Matt Asiata is a future dynasty asset for you when he scores in ADP's absence when he's JAG.

I hope this addresses some of your valid points ZWK, As always, I very much enjoy reading everything you write and have great respect for you.

 

[ZWK]

Thanks for the comments ZWK.

1. I agree small sample sizes are less predictive. What's less predicitve, I feel, is actual production and fantasy points than a measured performance using a grading system. To me, this is a more important barometer for long term success than opportunity (guy in front of you goes down and you get a shot), situation (you have a great offensive line or QB, which can change), or offensive philosophy ( run oriented team versus pass oriented team). To answer your original point on small sample sizes, I had thought that once PFF had three years of doing Player Grades (this was their first year doing grades as opposed to Premium Statistics) of doing a 3 year average weighted more strongly towards the most recent year, similar to how FF players will sometimes average three years worth of fantasy points (this would help negate a players lagging grade due to an injury like an ACL tear or an inflated grade perhaps too). Also, players in roles are able to specialize, but PFF uses a variety of factors in determining their original grade (run, receiving, blocking, etc.). In addition, the player grade should be more predictive than it's predecessor, premium stats, as it's more a measure of skill than amasses opportunity and compiling stats. In fact, what my model does is helps a FF player be able to abstractly "plug and play" and visualize what a certain player might be able to do given opportunity.

To support the above, I'd be buying on players like TJ Yeldon (owners scared of Chris Ivory) and selling on those like Jeremy Langford and Matt Jones (not that talented and dynasty value will drop once competition or a replacement is brought in). 

2. While I agree that TRUE role players  are tricky in valuing, the really good ones will put up substantial fantasy years too. To your point, Darren Sproles has put up back end RB1 numbers in his career. Another specialist, Danny Woodhead, put up RB 1 numbers in my league in 2015.

3. This is true. I agree. However, talented players will have work over non-talented ones, giving them value.

I'm not saying this list is an end all or I'm a messiah; I think it's a vastly more informative starting point however than actual production where we see a lot of posers to long term success that too many FF players (myself included) take the bait on. With hope, this approach should be a starting point rather than thinking Matt Asiata is a future dynasty asset for you when he scores in ADP's absence when he's JAG.

I hope this addresses some of your valid points ZWK, As always, I very much enjoy reading everything you write and have great respect for you.

I brought up the small sample size issue because the majority of your top 15 has less than 250 career touches. Spencer Ware is a 6th rounder who has been in the league 3 years and has less than 500 career yards from scrimmage, and he's sitting at #11.

One way to deal with small sample sizes is with regression to the mean. e.g, If a player has a grade of 90 on 1200 snaps then you might guess that his true talent level is 80, and if he has a grade of 90 on 400 snaps then you might guess that his true talent level is 70. You can calculate these numbers by giving each player credit for an additional 400 fake snaps with a grade of 50, and then calculating his average grade (including those fake snaps and his actual snaps). The larger the sample size, the closer your guess is to the grade that he earned from his play. You can pick other numbers rather than 400 fake snaps - the smaller the number of "fake snaps" that you include, the more you let players stand out with only a small sample size.

And (non-rookie) players can get a larger sample size if you use multiple years of data. PFF did have grades in the past, though they used a different scale (rather than giving ratings out of 100), but it shouldn't be too hard to transform them into a rating out of 100 to match their current format. For players in their first couple years, you also could use their draft position to make up a "fake grade" for a fake previous season, to account for the fact that Gurley & Gordon came into the league with a higher estimated talent than Langford & Rawls.

(The use of things like fake snaps, and fake grades for a fake previous season, is actually the correct way to do this sort of analysis according to Bayesian statistics. Though they use more dignified-sounding names.)

 

[Englishteacher]

I brought up the small sample size issue because the majority of your top 15 has less than 250 career touches. Spencer Ware is a 6th rounder who has been in the league 3 years and has less than 500 career yards from scrimmage, and he's sitting at #11.

Very true about Ware. Of course, he's been sitting behind two HOF'ers.

 

[Englishteacher]

And (non-rookie) players can get a larger sample size if you use multiple years of data. PFF did have grades in the past, though they used a different scale (rather than giving ratings out of 100), but it shouldn't be too hard to transform them into a rating out of 100 to match their current format. 

Unfortunately, PFF doesn't allow access to their premium statistics anymore as they've shifted their concept, just player grades are available. I had a subscription in the last several years but the premium statistics accumulated stats while the player grades average them.

 

[Englishteacher]

 For players in their first couple years, you also could use their draft position to make up a "fake grade" for a fake previous season, to account for the fact that Gurley & Gordon came into the league with a higher estimated talent than Langford & Rawls.

(The use of things like fake snaps, and fake grades for a fake previous season, is actually the correct way to do this sort of analysis according to Bayesian statistics. Though they use more dignified-sounding names.)

I'm somewhat familiar in name and concept with regression to the mean, fake snaps, projecting with draft position, and Bayesian stats.

Gordon had a pretty brutal rookie season. Some of the scouting reports I remembered reading on him were that he had a bust potential and that his college production, while impressive, could have been inflated due to volume and team style. In addition, other reports said his playing style might now translate well to the NFL. I'm steering clear of him, myself.