The DraftTwitter 100
In January 2017, a group of talented amateur scouts got together to embark on an experiment to see if DraftTwitter, that amorphous blob of football talent evaluation that resides in the Twitterverse, could rank NFL player prospects with the "accuracy" of pro scouts.
Keeping Track
The DraftTwitter Top 100 ("DT 100") was not meant to be compared to the actual NFL draft. Despite the "Best Player Available" draft strategy followed by some teams, many base their selections on team player personnel needs. So. while DraftTwitter's goal was to rank the 100 best prospects, teams like the Bears traded up to get the player they felt best fit their specific needs. So, how can we compare ourselves to the draft if the criteria used for selection are inconsistent?
Position Breakout
Instead of comparing The DT 100 to NFL's first through one hundredth picks, we can simply break out ranking by position. For example, here are the comparative rankings for the 8 RBS included in the DT 100 vs. their actual spot selected (Pos Pick, below):
| Player | Pos Pick | DT 100 | School |
| Leonard Fournette | 1 | 1 | Louisiana State |
| Christian McCaffery | 2 | 2 | Stanford |
| Dalvin Cook | 3 | 3 | Florida State |
| Joe Mixon | 4 | 4 | Oklahoma |
| Alvin Kamara | 5 | 5 | Tennessee |
| Kareem Hunt | 6 | 7 | Toledo |
| D'Onta Foreman | 7 | 6 | Texas |
| James Conner | 8 | Pittsburgh | |
| Samaje Perine | 9 | 8 | Oklahoma |
| Tarik Cohen | 10 | North Carolina A&T |
In the above example, the DT 100 matched up nicely, selecting the first 5 RBs in the exact order they were drafted, including nailing the complex Joe Mixon as RB4. The remaining RBs were each only off by one selection, proving the DT 100 RB class the strongest forecast by the project.
Obviously, we were pretty accurate with our RB picks, but how does one present this quantitatively?
The "V" Word Again
Not sure what some of you were thinking, but the "V" word we use time and again is Value. And what ultimately determines value is what someone is willing to give up to obtain something compared to another. While comparing one player's level of contribution to another is a nice way to start an interminable argument, it is not a good way to compare players. The Opportunity expense to have a player on your football team is ultimately measured in one way - contract value. By comparing a drafted player's projected slotted contract value to that of the position DraftTwitter predicted he would be picked, we find out if we over valued a player or got a great "bargain". (This assumes the actual draft order is "perfect" which it is since that is the standard we have until players actually start ringing up stats over the next few years).If we add contract values to the table above, it looks like this:
DraftTwitter Top 100 - RB Rankings
| Leonard Fournette | 1 | 1 | Louisiana State | $27,286,806 | 0% |
| Christian McCaffery | 2 | 2 | Stanford | $17,322,323 | 0% |
| Dalvin Cook | 3 | 3 | Florida State | $6,374,002 | 0% |
| Joe Mixon | 4 | 4 | Oklahoma | $5,465,848 | 0% |
| Alvin Kamara | 5 | 5 | Tennessee | $3,868,505 | 0% |
| Kareem Hunt | 6 | 7 | Toledo | $3,382,934 | 3% |
| D'Onta Foreman | 7 | 6 | Texas | $3,276,778 | -3% |
| James Conner | 8 | Pittsburgh | $3,181,736 | ||
| Samaje Perine | 9 | 8 | Oklahoma | $3,018,197 | -5% |
Leonard Fournette was the 4th overall pick, earning him a salary of $27MM. Since we had him as the first RB selected, we agree that his salary should have been identical and so he is at even value. Going down to Kareen Hunt, we had him as RB7 (with a contract value of $3,276,778) however the league told us he's RB6 (at a $3,382,934 value) so the NFL valued him 3% more than DT 100. On average, DT RB selections had a variance of only -1%., meaning we were pretty much dead on with our picks from a value standpoint compared.
In absolute values, the range went from 1% (RBs and Safeties) to 29% (the NFL valued the average LB about 29% more than we did):
Here is the table of how we performed at each position:
| Position | Under/ (-Over) |
| CB | 10% |
| DE | 13% |
| DT | 5% |
| LB | 29% |
| OL | 2% |
| QB | 3% |
| RB | -1% |
| S | 1% |
| TE | -4% |
| WR | -2% |
- LB includes OLB and ILB
- OL includes C, G and T
- S includes FS and SS
The Good
Out of the Top 100 players selected, DraftTwitter picked 26 correctly within their position group. The most accurate group was the RB group, in which we picked 63% correct.
| Position | Bullseyes |
| CB | 27% |
| DE | 27% |
| DT | 27% |
| LB | 44% |
| OL | 8% |
| QB | 20% |
| RB | 63% |
| S | 10% |
| TE | 17% |
| WR | 54% |
The Bad
We were off the most, by far, with the LB group:
| Player | Pos | Pos Rank | DTR | School | Total Value | |
| Haason Reddick | OLB | 1 | 2 | Temple | $13,538,340 | 23% |
| Jarrad Davis | ILB | 2 | 6 | Florida | $11,015,687 | 106% |
| T.J. Watt | OLB | 3 | 7 | Wisconsin | $9,295,615 | 97% |
| Reuben Foster | ILB | 4 | 1 | Alabama | $9,071,121 | -33% |
| Tyus Bowser | OLB | 5 | 9 | Houston | $5,591,978 | 52% |
| Ryan Anderson | OLB | 6 | 10 | Alabama | $5,339,704 | 47% |
| Raekwon McMillan | ILB | 7 | 8 | Ohio State | $4,709,050 | 5% |
| Zach Cunningham | OLB | 8 | 3 | Vanderbilt | $4,482,010 | -52% |
| Duke Riley | LB | 9 | 11 | Louisiana State | $3,684,055 | 16% |
The numbers were all over the place: DT` 100 had Reuben Foster at LB1, but the NFL thought he was no better than LB4, so we got him at a huge 33% discount to his NFL value. On the other hand, the NFL didn't watch the same tape DraftTwitter did because they wrote a $11MM check to Jarrad Davis at LB2 compared to the $5.3MM check DraftTwitter had ready for him. That is a 106% premium to where DT had Mr. Davis. The closest we came to agreeing with the NFL on LBs was Raekwon McMillan who came at only a 5% premium compared to DT 100's.
The Ugly
The RB table shows how good DT 100 was with that player position. We were also 1% off for Safeties but despite the nearly identical differential compared to RBs (off -1%), our skill at picking Safeties was pretty far worse. What do I mean?
| Jamal Adams | S | 1 | 1 | Louisiana State | $22,367,629 | 0% |
| Malik Hooker | S | 2 | 2 | Ohio State | $12,655,428 | 0% |
| Jabril Peppers | S | 3 | 4 | Michigan | $10,385,025 | 52% |
| Budda Baker | S | 4 | 3 | Washington | $6,853,308 | -34% |
| Marcus Maye | S | 5 | 8 | Florida | $6,575,816 | 45% |
| Marcus Williams | S | 6 | 6 | Utah | $6,260,481 | 0% |
| Justin Evans | S | 7 | 10 | Texas A&M | $5,213,582 | 23% |
| Obi Melifonwu | S | 8 | 5 | Connecticut | $4,519,852 | -31% |
| Josh Jones | S | 9 | 7 | North Carolina State | $4,242,361 | -19% |
| John Johnson | S | 10 | Boston College | $3,361,028 | ||
| Delano Hill | S | 11 | Michigan | $3,287,014 | ||
| Tedric Thompson | S | 12 | Colorado | $3,030,488 | ||
| Eddie Jackson | S | 13 | 9 | Alabama | $3,024,247 | -29% |
Although DT 100 got 3 selections perfect, the other picks suffered from the values distributed over a larger number of picks and, thus, a wider range of contract values. So, overall the group performance was quite good (at 1% off) the fluctuation in numbers makes us realize we lucked out on this one when the differential was only 1%.
Superlatives
The overall most overvalued pick - According to our method, Adoree Jackson was by far the most overvalued pick out there. Tennessee picked him at #18 overall as CB3, with a contract value of $11.3MM. DT 100 had him as CB11 which actually went 66th overall at a contract value of $3.8MM. Based on the above, Jackson was a whopping 192% overvalued by the league.
According to Kent Lee Platte's (aka MathBomb) Relative Athletic Score maps a player's potential to his athletic testing profile. Here is his RAS for Jackson:
His height and weight were his only "red" metrics and his overall RAS is very promising.
According to Kent Lee Platte's (aka MathBomb) Relative Athletic Score maps a player's potential to his athletic testing profile. Here is his RAS for Jackson:
Here are some other players who were overvalued by > 50%:
- Taco Charleton 98% (Actual pick: DE5 [#28 overall]; DT100 DE6 [#51])
- TJ Watt 97% (Actual pick: LB3[#30]; DT100 LB7 [#54])
- Adam Shaheen 86% (Actual pick: TE5 [#45]; DT100 TE6 [#100])
- Garett Bolles 57% (Actual pick: OL1 [#20]; DT100 OL3 [#34])
- Tyus Bowser 52% (Actual pick: LB5[#47]; DT100 LB9 [#75])
The best bargain pick - Unfornately for Bucky Hodges, DT100 had him at TE4 which went at pick #44 overall with a contract value of $6MM but the NFL wasn't as high on him. He didn't get called until the 201st pick as TE13 (appropriate) with a contract value of $2.5MM, massively undervalued at 59%. but also the potentially biggest bargain of this draft.
Take a look at his RAS score:
Take a look at his RAS score:
The guy was 14 out of 463 - better than 96.7% of all qualifying TEs in Kent's database. But he lost out on a lot of money with his drop through the rounds.
Here are some other players who were undervalued by > 30%:
- Jake Butt -55% (Actual pick: TE8 [#145 overall]; DT100 TE5 [#45])
- Zach Cunningham 52% (Actual pick: LB8[#57]; DT100 LB3 [#30])
- Deshaun Watson 52% (Actual pick: QB35 [#12]; DT100 QB1 [#2])
- Caleb Brantley 35% (Actual pick: DT14 [#185]; DT100 DT4 [#65])
- Obi Melifonwu 31% (Actual pick: S8 [#56]; DT100 S5 [#39])
Looking Forward
With 2017 results recapped, we can now look forward to the 2017-18 NCAA college football season. Throughout the season, we will post weekly Wide Receiver Return on Investment calculations and specialty articles on prospects across all positions in FBS, FCS, Division II and III. Occasionally, I might throw in an NFL progress report on a success of a particular draft class. And in January 2018, we will look to launch another DraftTwitter Top 100 survey to all interested parties.
Thank you and feel free to drop me a line at Boombearjr@gmail.com.
No comments:
Post a Comment