SITE NEWS: We are moving all of our site and company news into a single blog for Sports-Reference.com. We'll tag all PFR content, so you can quickly and easily find the content you want.
Also, our existing PFR blog rss feed will be redirected to the new site's feed.
Pro-Football-Reference.com ยป Sports Reference
For more from Chase and Jason, check out their work at Football Perspective and The Big Lead.
ShareThisAn AV-based draft chart
This is my current best attempt at an Approximate value-based draft chart.
I looked at all drafts from 1980--1999. I figured 1980 was late enough to be "modern," 1999 was early enough to have allowed most of the players to have finished their careers, and 20 years was enough to get a decent sample size.
Then, for each player, I computed what I call his "draft value":
Draft Value = (Total approx value of first four seasons) + (90% of AV of fifth season) + (80% of AV of sixth season) + ... + (10% of AV of 13th season)
The basic idea is this. When you draft a player, you're getting roughly his first four years, and then some uncertainty. This was just a quick way for me to capture the fact that there is some chance that your draftees will be helping some other team and not yours in five, six, seven years time. Also, teams presumably value 2008 production more than 2015 production.
Then I looked at the draft values of all players selected in a certain slot, and bunged them down into a chart. Well, sort of. Let me take a line and explain it:
Slot Top3 80 60 40 20 ========================= 22 124 52 39 26 14
This is the line corresponding to the #22 pick in the draft. In order to make things a little smoother, I actually grouped together all the picks from #17 through #27. So what I've got now are 20 years worth of players drafted 17 through 27. That's 220 players. Top3 refers to the average of the three players in that group with the highest draft value. 80 represents the 80th-percentile player in the group, 60 represents the 60th-percentile player, and so on.
So roughly speaking, the Cowboys (who hold the #22 pick), could conceivably get a 124 draft value player if they get really lucky. They have about a 20% chance of getting a 52 or better, a 40% chance of getting a 39 or better, a 60% chance of getting a 26 or better, and an 80% chance of getting a 14 or better.
To give you a sense of what that means, here are some players at various levels and their draft values:
Peyton Manning 140 Ray Lewis 115 Jason Taylor 101 Chris Doleman 94 Michael Dean Perry 81 Eric Moulds 71 Keena Turner 60 Bart Oates 55 Frank Wycheck 50 Jeff Hartings 45 Tyrone Poole 40 Eric Barton 35 Dexter McCleon 30 David Givens 25 Cory Schlesinger 20
Alright then, here is the full chart. Enjoy:
Slot Top3 80 60 40 20 ========================= 1 133 81 61 46 21 2 133 77 59 45 19 3 133 77 60 45 20 4 133 75 59 43 20 5 133 75 59 42 20 6 133 73 58 40 20 7 127 69 57 40 20 8 117 66 57 40 19 9 111 65 53 38 18 10 108 64 50 36 17 11 112 63 50 35 17 12 118 61 45 33 17 13 118 60 45 33 17 14 119 59 45 32 16 15 119 59 45 31 16 16 119 58 44 29 14 17 119 58 44 29 15 18 119 57 43 29 15 19 119 54 43 28 14 20 119 54 42 27 14 21 120 54 41 27 14 22 124 52 39 26 14 23 122 53 39 25 12 24 122 53 38 25 12 25 122 50 37 24 11 26 122 50 35 21 9 27 122 48 34 19 7 28 123 49 32 19 6 29 123 49 32 18 6 30 123 48 31 18 6 31 123 50 34 19 6 32 123 49 34 19 6 33 116 50 34 19 5 34 112 50 31 19 5 35 114 49 30 17 5 36 114 49 31 17 5 37 114 48 31 19 6 38 114 49 32 19 7 39 110 49 31 19 7 40 110 48 31 18 7 41 110 48 31 19 7 42 110 47 31 18 6 43 110 47 29 16 6 44 103 45 29 17 6 45 103 45 29 16 5 46 96 44 30 17 5 47 96 44 30 16 5 48 96 44 29 14 4 49 96 44 28 13 4 50 96 42 27 13 4 51 90 41 24 10 3 52 90 41 23 10 3 53 90 40 23 9 3 54 89 40 23 10 4 55 87 40 21 9 4 56 85 39 20 9 4 57 79 35 18 9 4 58 80 34 20 9 4 59 80 35 20 10 4 60 80 35 21 10 4 61 81 35 20 10 4 62 81 35 21 10 4 63 81 35 20 10 4 64 81 34 20 10 4 65 81 34 18 9 3 66 74 33 20 9 3 67 76 33 20 9 3 68 85 35 21 9 3 69 93 36 18 8 3 70 93 35 17 7 2 71 93 33 16 7 2 72 93 33 17 7 2 73 93 33 16 7 2 74 93 33 14 6 2 75 93 31 14 6 1 76 93 31 14 6 1 77 93 30 13 4 1 78 92 30 12 4 1 79 83 27 12 4 1 80 71 28 12 4 1 81 76 30 12 3 1 82 73 29 10 3 1 83 73 29 10 3 1 84 83 27 12 3 1 85 83 27 10 3 0 86 83 30 12 3 1 87 84 29 11 3 1 88 84 30 13 3 1 89 84 27 9 3 1 90 84 27 9 3 1 91 86 27 11 3 1 92 83 27 12 3 1 93 83 28 13 3 1 94 83 28 12 3 1 95 74 30 11 3 1 96 74 30 11 4 1 97 74 30 11 4 1 98 71 30 11 4 1 99 72 27 11 4 1 100 72 29 11 4 1 101 72 27 11 4 0 102 66 26 11 3 0 103 66 24 9 3 0 104 64 23 8 4 0 105 63 22 8 3 1 106 63 20 9 4 1 107 62 21 10 3 0 108 67 21 11 4 0 109 74 26 13 4 0 110 72 26 13 4 0 111 72 23 13 3 0 112 72 23 11 4 0 113 84 24 11 3 0 114 84 24 11 3 0 115 84 24 11 3 0 116 84 24 11 3 0 117 84 23 9 3 0 118 84 21 7 2 0 119 82 18 7 2 0 120 77 18 7 2 0 121 77 17 5 2 0 122 77 17 6 2 0 123 77 18 5 1 0 124 66 18 5 2 0 125 65 17 5 2 0 126 69 17 5 2 0 127 66 17 5 1 0 128 65 18 5 1 0 129 65 19 5 1 0 130 65 19 5 1 0 131 65 17 5 1 0 132 64 18 5 1 0 133 64 18 6 1 0 134 64 18 6 1 0 135 62 18 6 1 0 136 62 19 6 1 0 137 56 18 6 1 0 138 56 17 6 1 0 139 56 16 5 1 0 140 56 17 6 1 0 141 57 17 6 1 0 142 57 17 6 1 0 143 57 16 5 1 0 144 57 15 4 1 0 145 69 15 4 1 0 146 69 14 4 1 0 147 69 13 4 1 0 148 69 14 4 1 0 149 87 14 4 1 0 150 87 14 4 1 0 151 87 13 3 1 0 152 84 14 4 1 0 153 83 15 4 1 0 154 86 17 4 1 0 155 86 16 4 1 0 156 76 16 4 1 0 157 76 17 4 1 0 158 76 17 4 1 0 159 76 16 3 0 0 160 57 16 3 0 0 161 66 17 3 0 0 162 66 17 3 0 0 163 66 17 3 0 0 164 66 16 2 0 0 165 62 14 2 0 0 166 62 15 2 0 0 167 62 13 2 0 0 168 62 11 2 0 0 169 62 11 2 0 0 170 62 10 2 0 0 171 62 11 2 0 0 172 52 10 2 0 0 173 52 10 2 0 0 174 59 10 2 0 0 175 59 10 2 0 0 176 60 12 3 0 0 177 60 11 3 0 0 178 60 12 3 0 0 179 60 9 2 0 0 180 60 10 2 0 0 181 60 10 2 0 0 182 64 11 2 0 0 183 64 12 2 0 0 184 64 11 2 0 0 185 57 8 2 0 0 186 57 8 2 0 0 187 66 8 2 0 0 188 66 8 2 0 0 189 66 8 2 0 0 190 66 10 2 0 0 191 74 10 2 0 0 192 74 10 2 0 0 193 69 8 2 0 0 194 70 7 1 0 0 195 70 8 1 0 0 196 70 8 2 0 0 197 70 9 2 0 0 198 72 11 2 0 0 199 72 10 2 0 0 200 72 9 2 0 0 201 72 8 2 0 0 202 68 9 2 0 0 203 74 11 2 0 0 204 74 12 2 0 0 205 74 12 2 0 0 206 74 12 2 0 0 207 74 12 2 0 0 208 74 10 2 0 0 209 64 10 2 0 0 210 64 11 2 0 0 211 64 10 2 0 0 212 64 10 2 0 0 213 63 10 2 0 0 214 56 9 2 0 0 215 56 9 2 0 0 216 56 10 2 0 0 217 68 10 2 0 0 218 76 9 2 0 0 219 76 9 2 0 0 220 76 9 1 0 0 221 76 9 1 0 0 222 77 9 1 0 0 223 77 9 1 0 0 224 77 9 1 0 0
This entry was posted on Friday, April 25th, 2008 at 4:24 am and is filed under Approximate Value, NFL Draft. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.
"I actually grouped together all the picks from #17 through #27". If you used this method for picks 1-4 the table could be misleading. E.g. pick 1's data will include pick 1-5 - it is thusly "dragged down" and the data looks like pick 1 is not as valuable as it should be.
Omar, I did use that method at the top, and I considered the difficulty you mentioned. But ultimately I decided the extra players in the sample was worth the tradeoff.
For what it's worth, here are the top 6 if you strictly use the draft slots instead of grouping:
Hmmm, it looks like you're right. I am short-changing the top few picks.
How do you use this to determine trade value for draft picks?
It might be interesting to see the 50 percentile value and also the average value. Skpping from 60 to 40 misses what I'd consider the most important data point. Although an argument could be made that sub par picks aren't as damaging as their AV would suggest due to the fact that you stay at the top of the draft until you get your Peyton Manning.
Bad picks are expensive in dollar terms, yes, but it's probably better for a team to draft Peyton Manning and Ryan Leaf rather than two average players as they can move on from the bad pick. This is also why QBs are taken so early, the upside of a franchise QB is too much to ignore. I'm sure most teams going back to 98 would have been willing to blindly pick one of Manning or Leaf. Usually the split isn't that dramatic, but the point still stands, boom or bust, it's worth the chance.
I've been through this but don't understand how you derive the weightings. Many of them seem arbitrary, which is ok, but how do you judge whether their right or some others better?
Which weights are you referring to, Bruce? Are you talking about the construction of the AV numbers themselves?
I think so, as for example your weighting on the ln of the draft choice slot number. I assume you did regression analysis to get that?