deadstick
Footballguy
PREFACE: Not looking for help on who to keep, just an evaluation of the analysis.
I have recently been working on an analytical approach to calculating keeper "values" for players in leagues where you lose a draft pick to keep a player. I started looking at this for one of my 10-team leagues that allows us to keep 5 players (any position) at a cost of a draft pick one round earlier than where you took that player in the previous year's draft. If a player was a FA, he costs you a 10th round pick. We're in year 4, and you can continue keeping players so some guys (Like Larry Johnson) have been kept for a number of years, and each year the pick that the owner loses gets one round earlier (10th in 2004, 9th in 2005, 8th in 2006, 7th in 2007).
So, my team has the following possible keeper candidates:
1) Frank Gore: 5th round pick
2) Maurice Jones Drew: 10th round pick
3) TJ Houshmandzadeh: 8th round pick
4) Larry Fitzgerald: 3rd round pick
5) Michael Turner: 10th round pick
6) Deion Branch: 10th round pick
7) Jerricho Cotchery: 10th round pick
4 of my keepers were easy for me: Gore, Jones-Drew, Houshmandzadeh and Turner. I really liked the idea of having Fitzgerald as a stud WR, but Branch for a 10th rounder was intriguing as well, which lead me to try to determine which was better value. I did a number of mocks with both approaches using Draft Dominator and found that my team was nearly identical in scoring in each case. I also did this for one other team's keepers for a similar scenario.
As a VERY rough approximation, I determined that Larry Fitzgerald for a 3rd round pick is about equal to Deion Branch for a 10th round pick. If I kept Fitz, the QBs fall to the point where I get a guy like Kitna later, and in the 10th round am looking at someone like Ladell Betts.
If I don't keep Fitz, I get a top-5 QB in the 3rd round (like Bulger), but my #3 WR ends up being TE Antonio Gates.
In our scoring system, with FBG projections, the 2 different starting lineups are:
QB: Kitna
RB: S Jackson, Gore
WR: Holt, Fitz, Housh
QB: Bulger
RB: S Jackson, Gore
WR: Holt, Housh, Gates (Branch, another keeper, is my #4 WR)
And that's about a wash in terms of scoring. One could argue the relative dynasty-type worth of a guy like Kitna but I'm not interested in that for this. Another facet would be the fact that Branch would be good value as a keeper again in 2008 (for a 9th rounder), but Fitz would cost a 2nd and likely not be worth it, definitely not worth a 1st rounder in 2009 (no WRs hardly ever are in this system).
I took the VBD values of all potential keeper candidate players in the league (normalized by adding 100, so a guy with a VBD of -9 is now 91, while a guy with a VBD of 50 is now 150, just to worth with postitive numbers), and also computed the difference in the "round pick to keep" vs. ADP.
I then looked for a relation between VBD and round pick difference to try and determine a "keeper value".
As a VERY rough approximation, I used the example of Fitzgerald and Branch. DD's VBD for Fitz is 39 (normalized to 139) and Branch's is 0 (normalized to 100). Fitz's ADP is round 2, while Branch's is round 6. My cost to keep Fitz is a 3rd rounder, cost to keep Branch is a 10th rounder. So, set the 2 equal, with an unknown variable (PV = Pick Value, which was 3-2 = 1 for Fitz and 10-6 = 4 for Branch):
VBD(PV + x), so
139(1+x) = 100(4+x)
x = 6.7
So, my equation for keeper value was:
KV = VBD(PV + 6.7)
Throwing this into a spreadsheet for all the keepers in my league, it looks to be intuitively correct. The biggest flaw being that 6.7 is just some random number that works out for this (and one other spot-check I did for another similar situation).
Here's the top 10 keeper values for my league (which looks fairly accurate in my mind for a 10-team league, 20-round draft, keep 5, start 1 QB, 2 RB, 3 WR)
1) RB Travis Henry: 16th round, KV = 3084
2) RB Larry Johnson: 8th round, KV = 2945
3) RB Willie Parker: 9th round, KV = 2558
4) RB Frank Gore: 5th round, KV = 1958
5) RB Maurice Jones Drew: 10th round, KV = 1945
6) WR Braylon Edwards: 14th round, KV = 1758
7) RB Brandon Jacobs: 17th round, KV = 1720
8) RB Cedric Benson: 9th round, KV = 1626
9) RB Laurence Maroney: 6th round, KV = 1605
10)WR Marques Colston: 10th round, KV = 1537
Re-evaluating my keeper candidates, ranked by KV, I would keep the first 5 (however, I will be keeping Turner due to the dynasty aspect of him which isn't recognizable by his 2008 projections):
1) Frank Gore: 1958
2) Maurice Jones Drew: 1945
3) TJ Houshmandzadeh: 1402
4) Larry Fitzgerald: 1070
5) Michael Turner: 10th round pick
6) Deion Branch: 1070
7) Jerricho Cotchery: 951
So, does anyone have a more scientific approach (such as instead of 6.7, use the # of skilled position starters) and a proof to back it up? Any other comments?
I have recently been working on an analytical approach to calculating keeper "values" for players in leagues where you lose a draft pick to keep a player. I started looking at this for one of my 10-team leagues that allows us to keep 5 players (any position) at a cost of a draft pick one round earlier than where you took that player in the previous year's draft. If a player was a FA, he costs you a 10th round pick. We're in year 4, and you can continue keeping players so some guys (Like Larry Johnson) have been kept for a number of years, and each year the pick that the owner loses gets one round earlier (10th in 2004, 9th in 2005, 8th in 2006, 7th in 2007).
So, my team has the following possible keeper candidates:
1) Frank Gore: 5th round pick
2) Maurice Jones Drew: 10th round pick
3) TJ Houshmandzadeh: 8th round pick
4) Larry Fitzgerald: 3rd round pick
5) Michael Turner: 10th round pick
6) Deion Branch: 10th round pick
7) Jerricho Cotchery: 10th round pick
4 of my keepers were easy for me: Gore, Jones-Drew, Houshmandzadeh and Turner. I really liked the idea of having Fitzgerald as a stud WR, but Branch for a 10th rounder was intriguing as well, which lead me to try to determine which was better value. I did a number of mocks with both approaches using Draft Dominator and found that my team was nearly identical in scoring in each case. I also did this for one other team's keepers for a similar scenario.
As a VERY rough approximation, I determined that Larry Fitzgerald for a 3rd round pick is about equal to Deion Branch for a 10th round pick. If I kept Fitz, the QBs fall to the point where I get a guy like Kitna later, and in the 10th round am looking at someone like Ladell Betts.
If I don't keep Fitz, I get a top-5 QB in the 3rd round (like Bulger), but my #3 WR ends up being TE Antonio Gates.
In our scoring system, with FBG projections, the 2 different starting lineups are:
QB: Kitna
RB: S Jackson, Gore
WR: Holt, Fitz, Housh
QB: Bulger
RB: S Jackson, Gore
WR: Holt, Housh, Gates (Branch, another keeper, is my #4 WR)
And that's about a wash in terms of scoring. One could argue the relative dynasty-type worth of a guy like Kitna but I'm not interested in that for this. Another facet would be the fact that Branch would be good value as a keeper again in 2008 (for a 9th rounder), but Fitz would cost a 2nd and likely not be worth it, definitely not worth a 1st rounder in 2009 (no WRs hardly ever are in this system).
I took the VBD values of all potential keeper candidate players in the league (normalized by adding 100, so a guy with a VBD of -9 is now 91, while a guy with a VBD of 50 is now 150, just to worth with postitive numbers), and also computed the difference in the "round pick to keep" vs. ADP.
I then looked for a relation between VBD and round pick difference to try and determine a "keeper value".
As a VERY rough approximation, I used the example of Fitzgerald and Branch. DD's VBD for Fitz is 39 (normalized to 139) and Branch's is 0 (normalized to 100). Fitz's ADP is round 2, while Branch's is round 6. My cost to keep Fitz is a 3rd rounder, cost to keep Branch is a 10th rounder. So, set the 2 equal, with an unknown variable (PV = Pick Value, which was 3-2 = 1 for Fitz and 10-6 = 4 for Branch):
VBD(PV + x), so
139(1+x) = 100(4+x)
x = 6.7
So, my equation for keeper value was:
KV = VBD(PV + 6.7)
Throwing this into a spreadsheet for all the keepers in my league, it looks to be intuitively correct. The biggest flaw being that 6.7 is just some random number that works out for this (and one other spot-check I did for another similar situation).
Here's the top 10 keeper values for my league (which looks fairly accurate in my mind for a 10-team league, 20-round draft, keep 5, start 1 QB, 2 RB, 3 WR)
1) RB Travis Henry: 16th round, KV = 3084
2) RB Larry Johnson: 8th round, KV = 2945
3) RB Willie Parker: 9th round, KV = 2558
4) RB Frank Gore: 5th round, KV = 1958
5) RB Maurice Jones Drew: 10th round, KV = 1945
6) WR Braylon Edwards: 14th round, KV = 1758
7) RB Brandon Jacobs: 17th round, KV = 1720
8) RB Cedric Benson: 9th round, KV = 1626
9) RB Laurence Maroney: 6th round, KV = 1605
10)WR Marques Colston: 10th round, KV = 1537
Re-evaluating my keeper candidates, ranked by KV, I would keep the first 5 (however, I will be keeping Turner due to the dynasty aspect of him which isn't recognizable by his 2008 projections):
1) Frank Gore: 1958
2) Maurice Jones Drew: 1945
3) TJ Houshmandzadeh: 1402
4) Larry Fitzgerald: 1070
5) Michael Turner: 10th round pick
6) Deion Branch: 1070
7) Jerricho Cotchery: 951
So, does anyone have a more scientific approach (such as instead of 6.7, use the # of skilled position starters) and a proof to back it up? Any other comments?
