≡ Menu

Yet another draft value chart

Last November, I provided an updated version of my own draft value chart where I measured the value provided by each draft pick to his Approximate Value over the course of his first five years. A week later I decided another change was needed. While AV measures the value provided by a player, the marginal value provided by a player is a better measure of the value of a draft pick. As a result, I re-did the chart and only gave players credit for their AV above 2 points of AV.

You can view the values for both of those charts and the Jimmy Johnson chart here. This week, I spoke with Peter Keating of ESPN the Magazine, who is working on an article regarding how teams should value draft picks. Keating asked if I could make two changes to the chart, and I was happy to do so (and thought you guys might be interested). First, I increased the measure of replacement-level AV from 2 to 3 points. Theoretically, this change would reward the best players, as the higher the value used for replacement level, the fewer players that will meet that threshold. The other change was to reduce the number of years measured from five to four, since that matches the length of the typical rookie contract under the new CBA. The chart below shows the raw data and a smoothed curve depicting the marginal AV (over 3) produced by draft picks in the first four years of their career over a 28-year period.

av draft value iii

As it turns out, the chart doesn’t really change even with these tweaks. In fact, if you chart the Jimmy Johnson draft chart, my second draft value chart from November (AV II in the graph below) and today’s iteration (AV III), the two AV lines almost perfectly overlap. Note: to plot all three draft value charts on the same graph, I divided the draft value for each pick by the draft value of the first overall pick for each of the three draft value charts.

av draft charts

The key takeaway isn’t how the two AV draft values compare to each other but how they differ from the Jimmy Johnson chart. The dropoff is much more severe in the Jimmy Johnson draft chart, which in my opinion, has arguably resulted in the exact opposite effect of what was intended. In theory, a standardized draft value chart that each team uses would make trading easier, since everyone would know the value of a draft pick. And in fact, many in the media still use the Jimmy Johnson draft chart (I do not know, but only assume (hope?) that no NFL teams actually use this chart). The problem is that the values associated with the Jimmy Johnson chart bear no relationship to reality. As a result, teams that want to trade down generally demand way too much from teams looking to trade up. It makes no sense to trade up based off of a chart that highly overvalues the top picks, so teams are very reluctant to do so. Conversely, because teams at the top of the draft don’t want to appear “taken” — i.e., by trading a top pick for less than what the media and fans think the top draft picks are worth — teams with high picks may be inflexible about what they will accept via trade.

This year will be an interesting test case, as most draftniks think the value at the top of the draft is very flat (i.e., the difference between the first and 20th picks is small). If that’s the case, then even my AV draft charts wildly overstate the value of the top picks, while the Jimmy Johnson chart would be essentially useless. Whether or not teams budge from the old values associated with top picks will be a fascinating side story to watch.

Finally, for Keating and everyone else out there, the table below shows the marginal draft values (over 3 points of AV) accumulated by the average draft pick for each slot over the first four years of the players’ careers:

Pk
AV
123
220.1
318.4
417.1
516.2
615.4
714.8
814.2
913.7
1013.3
1112.9
1212.5
1312.2
1411.9
1511.6
1611.3
1711
1810.8
1910.6
2010.4
2110.1
2210
239.8
249.6
259.4
269.2
279.1
288.9
298.8
308.6
318.5
328.4
338.2
348.1
358
367.9
377.8
387.6
397.5
407.4
417.3
427.2
437.1
447
456.9
466.8
476.7
486.7
496.6
506.5
516.4
526.3
536.2
546.2
556.1
566
575.9
585.9
595.8
605.7
615.6
625.6
635.5
645.4
655.4
665.3
675.3
685.2
695.1
705.1
715
724.9
734.9
744.8
754.8
764.7
774.7
784.6
794.6
804.5
814.5
824.4
834.3
844.3
854.2
864.2
874.1
884.1
894.1
904
914
923.9
933.9
943.8
953.8
963.7
973.7
983.6
993.6
1003.6
1013.5
1023.5
1033.4
1043.4
1053.4
1063.3
1073.3
1083.2
1093.2
1103.2
1113.1
1123.1
1133
1143
1153
1162.9
1172.9
1182.9
1192.8
1202.8
1212.8
1222.7
1232.7
1242.7
1252.6
1262.6
1272.6
1282.5
1292.5
1302.5
1312.4
1322.4
1332.4
1342.3
1352.3
1362.3
1372.2
1382.2
1392.2
1402.1
1412.1
1422.1
1432.1
1442
1452
1462
1471.9
1481.9
1491.9
1501.9
1511.8
1521.8
1531.8
1541.7
1551.7
1561.7
1571.7
1581.6
1591.6
1601.6
1611.6
1621.5
1631.5
1641.5
1651.4
1661.4
1671.4
1681.4
1691.3
1701.3
1711.3
1721.3
1731.2
1741.2
1751.2
1761.2
1771.2
1781.1
1791.1
1801.1
1811.1
1821
1831
1841
1851
1860.9
1870.9
1880.9
1890.9
1900.9
1910.8
1920.8
1930.8
1940.8
1950.7
1960.7
1970.7
1980.7
1990.7
2000.6
2010.6
2020.6
2030.6
2040.6
2050.5
2060.5
2070.5
2080.5
2090.5
2100.4
2110.4
2120.4
2130.4
2140.4
2150.3
2160.3
2170.3
2180.3
2190.3
2200.2
2210.2
2220.2
2230.2
2240.2
{ 11 comments }
  • James April 4, 2013, 1:24 pm

    That predicts the expected AV of the player picked, but it doesn’t represent the value of the pick. For that you need to account for the rookie pay scale.

    For instance, according to spotrac.com, Luck’s #1 overall contract is worth $22.11 million while Weeden’s #22 contract is $8.08 million. That means Luck is expected to provide 1.04 AV/mil$ while Weeden is expected to provide 1.23 AV/mil$. That means the #22 overall pick is worth slightly more than the 23 to 10 AV difference would imply.

    And if you want to be extremely accurate, the math only gets more complicated since the first round picks have a team option for a 5th year, and the price is based on (1) when they were drafted, (2) their position, and (3) the top contracts after their third year. “If a player is picked 1-10, then the 5th year option is going to be the average of the top-10 players at the respective player’s position. So, if you’re quarterback and go number-two in the Draft, then your fifth year option salary will be the average of the top-ten quarterbacks. The option is exercised after the third season, so it would be based upon the top-ten quarterback salaries at that time. If you’re picked 11-32, then the 5th year club option is the average salary of the top 3rd through 25th player in that position.”

    Reply
  • Ryan April 5, 2013, 11:53 am

    I read about your calculations for AV and all, and I may have missed where you discussed what I am going to ask so I apologize if you have already covered it, but I was wondering if you eliminated outliers when averaging the AV for each draft pick. It appears as though some sort of method has been used because if not we would expect the 199th pick to be worth more than its neighbors because of Tom Brady. The same would go for the 2nd selection, it could be heavily impacted by the careers of guys similar to Ryan Leaf. I’m sure you’ve thought about this, but I just wanted to know how you treated outliers(both positive and negative) in order to avoid skewing a specific pick.

    I also wanted to know if there was a reason you left the AV values rounded to the tenth instead of the hundredth? Just for simplicity?

    Thanks!

    Reply
    • Chase Stuart April 5, 2013, 11:57 am

      Ryan,

      This is a smoothed curve to deal with exactly those issues. If you look at the first chart, that shows the raw data (in blue) with the smoothed curve in black. FWIW, Brady provided 0 AV as a rookie and then 36 points over the next three years. That equates to 18 points of marginal AV. Since there are 28 players in the study drafted at pick 199, he doesn’t even move the needle by a full point of AV.

      Reply
      • James April 5, 2013, 4:21 pm

        So Chase, when are you coming out with Part II to your veteran salary series? I’m a Cowboys fan so I was completely sucked in by your lede, but now I’m left out here waiting for the conclusion!

        Reply
        • Chase Stuart April 5, 2013, 4:22 pm

          Ha. I hope to work on it this weekend. As you can imagine, it’s pretty complicated, and my goal is to spend the time to do it correctly rather than quickly. I’m always open for thoughts on how it should turn out!

          Reply
          • James April 7, 2013, 8:18 pm

            Well, it appears that my first post was lost to the ether so lets try this again.

            Based on your salary cap calculator and replacement level team posts, it appears that a marginal win costs around $12-14 million marginal dollars with about $73 million marginal dollars available per team. My recommendation would be to either force or use as a check a split of $41 million on offense and $32 million on defense. This is based on Brian Burke’s research that showed the ratio of the standard deviation of offense to defense was 1.26 : 1 for yards gained/allowed per play, success rate, EPA, and WPA. http://www.advancednflstats.com/2011/01/top-offenses-top-defenses.html

            Also, Tom Tango posted on his blog that in the NFL it’s about 35 points* per win (not including ~3 base points per team per game). From above, that implies that a marginal point is worth about $370,000 marginal dollars (I may be making a mistake here by not considering rookie contributions to points scored, so please correct me if so). It seems to me that between this information and ANS’s or PFR’s EPA we should be able to get a good idea of how much money each offensive and defensive unit deserves, although EPA will need a constant added so it’s over replacement and not over average. Of course, the hard part will be breaking that down to the individual level, but I would think PFR’s Approximate Value will get you 80% of the way there, and maybe we could even take it further. * – Chase, you probably know this already but for other readers that’s 35 points *scored or allowed* per win, so defense is still included. http://www.insidethebook.com/ee/index.php/site/article/base_scores/

            My biggest concern is synergy between players may not stack linearly like it (mostly) does in baseball. I definitely think it’s possible that a +1 win QB and a +0.5 win WR may be worth more than a +1 win QB and a +0.5 win OLB. This seems especially true for the offensive line’s pass blocking or a defensive line’s run blocking where a team can specifically target a weakness (you’re only as strong as the weakest link), meaning teams will not necessarily be the sum of their parts. Unfortunately, I have no idea how to determine or account for this, but I think that’s the reason why we estimate the NFL’s replacement level is around 0.160 and not 0.320 like in baseball.

            Reply
            • Chase Stuart April 9, 2013, 2:30 pm

              Thanks James. Those are all good comments. Unfortunately, I’m not sure how to best deal with the synergies issue.

              Reply
              • James April 9, 2013, 4:58 pm

                Honestly, if it were me I’d send an email to Tango or Phil because undoubtedly they’ve come across similar issues and thought about it before for baseball and maybe even football. I remember a thread Tango created where he discussed the overlapping ranges of SS and 3B and how that effected their UZR scores and how to account for that but I can’t find it right now, but maybe something along those lines would be helpful.

                Reply

Leave a Comment

Switch to mobile version