American Football

DraftBot 2.1 Commanders Post-First-Wave Free Agency Mock Draft

on

EX Future Science And Technology Museum In Dalian
Photo by VCG/VCG via Getty Images

Addressing the need for more talent

The DraftBot was originally conceived as a simple experiment to demonstrate that the Washington NFL franchise has made drafting a franchise quarterback far more difficult than it needs to be. Using a crude prototype mocked up by some friends at local tech startup BrisVegas Systems, I showed that following a few simple rules and picking off of media draft experts’ big boards would have landed Drew Brees, Aaron Rodgers, Ben Roethlisberger, Lamar Jackson and Tua Tagovailoa, without trading away any draft capital.

That demonstration caught the attention of a certain Australian government agency which bankrolled an ambitious expansion of the DraftBot’s decision making capabilities, as well as some other enhancements, for some different applications that I am not able to talk about due to certain provisions of the Five Eyes security pact. Fortunately, the team at BrisVegas systems saw the value of the NFL draft as a testbed for their technology and have agreed to continue the partnership.

Since its Way Too Early Mock Draft, the DraftBot’s neural network processor has been doing deep data mining on the Commanders’ 2022 season and the 2023 draft class to optimize its strategy for April. It has just completed its analysis of data from the NFL Combine, including an extended analysis of QB hand size, OT arm length, and textual analysis of prospect interviews. Don’t ask me how it got those.

With all of the new data assimilated into its processor, the DraftBot is ready for another trial run following the first wave of free agency. Before we get into that, however, I would like to share some insights that the DraftBot has derived from its analysis of data from the past 23 seasons of football under Dan Snyder’s ownership and how the results will shape its approach to the draft.

Some misconception arose about the DraftBot’s strategy following its Way Too Early Mock Draft. The DraftBot is, in fact, driven entirely by need. It just defines “need” completely differently to most commentators and fans. Based on its deep mining and multifactorial network analysis of data on Washington’s drafts and on-field performance, it has identified three major needs that have remained more or less constant for the past 23 years and in some cases longer:

  1. Franchise quarterback
  2. Elite talent
  3. Depth with upside

The situation at quarterback and what needs to done to address it were the subject of the DraftBot’s debut article and the article that motivated it, so there is no need to rehash it here. Unfortunately, the DraftBot does not feel that it is in a position to address the need for a starting QB in this draft, and is already looking ahead to 2024.

The DraftBot has found that, compared to the competition, Washington’s roster is alarmingly devoid of elite talent, and has been since shortly after Charley Casserly was fired in 1999. For example, Jeremy Reaves recently became the first Washington player named First-Team All Pro since Brandon Scherff in 2020. The last First Team All Pro in Washington before Scherff was punter Matt Turk in 1996. The DraftBot’s analysis has found that the shortage of elite talent has done more to keep Washington out playoff contention than having a particularly weak starter any position, other than quarterback, during most seasons this century.

Unfortunately, again, the top end of the 2023 draft is fairly short of elite prospects compared to many recent drafts. The DraftBot has, however, identified quite a few Day 2 prospects with sleeper potential, and it is busy calculating probabilities on trade back scenarios to gain more picks in that range. Alternatively, it may focus its efforts in this draft on the team’s other major need.

The DraftBot has found that the Commanders have struggled with a chronic inability to replace departing starters from their own depth ranks for as far back as it has looked. This has forced the team to rely more heavily on free agency for team building than is advisable. It has identified a major needs to replace career backups and depth players who are not showing satisfactory development at wide receiver and cornerback in addition to the obvious deficiencies in depth at QB, offensive line and linebacker. It is also concerned about the absence of obvious succession planning at DE, with both starters potentially heading for expensive contract extensions if better options can’t be found.


DraftBot 2.1 Mock Draft

This mock draft was performed using the Pro Football Focus mock draft simulator. I decided to try the PFF simulator this time because the Pro Football Network mock simulator used in the previous mock draft makes players available far later than they are generally projected to go. The PFF draft board is also very different from the DraftBot’s, which makes for some amusing draft grades, which I’ll show at the end. Unfortunately, the PFF simulator does not make it easy to share the full mock results, so I will comment on significant players who were gone and best available players throughout.

The PFF simulator was set to default settings except that drafting for need was set at 50%, as opposed to the default setting which is closer to 75% favoring needs-based drafting. Now, here we go.

The DraftBot is on the board…

Round 1, Pick 16

All but one of the players the DraftBot expected it might see and would have considered taking at 16 were off the board. These included CBs Christian Gonzalez and Joey Porter, OTs Paris Johnson and Broderick Jones. The DraftBot calculated that there was a high probability that the player it liked would still be available in a modest trade-back scenario. It received an offer from Tampa Bay to trade back three spots for a haul of additional picks.

Trade: Tampa receives pick number 16 and selects TE Michael Mayer, Notre Dame. Washington receives picks 19, 82, 154, 174, 178.

The trade netted Washington a surplus of 51 points according to the Rich Hill trade value chart, which is equivalent to the 84th overall pick in the third round. The DraftBot got back the third-round pick that Ron Rivera frittered away for journeyman QB Carson Wentz and added three more picks on Day 3. The best part is the player it liked at 16 was still available.

Round 1, Pick 19

Brian Branch, CB/S, Alabama

6-0, 190 lbs, 4.58 sec 40, RAS 6.10

As I explained in my nickelbacks position roundup, the nickel package has become the Commanders’ true base defense. The Commanders’ primary slot corner in 2022 was safety Bobby McCain, who recorded the second most snaps of any player on defense, splitting time between slot coverage and free safety. In other words, the nickelback/safety role is a starting position in Washington’s defense, and was vacant heading into the draft.

Branch ran slower than expected at the Combine, but there is no indication that he lacked speed against SEC opponents. He has the quicks to cover shifty slot receivers, the length to cover tight ends and the range to play single-high safety. Unlike the player he replaces, Branch is an excellent run defender and finishes tackles with violence. In 2022 he was 4th in the SEC with 58 solo tackles and 2nd with 14 tackles for loss. He is described as a tone setter on defense with no significant weakness to his game, making him one of the safest picks in the draft. Washington adds a plug-and-play starter to the secondary with elite potential.

There was no other player close to Branch on the DraftBot’s board.

Here is how the full first round played out:


Round 2, Pick 47

John Michael Schmitz C, Minnesota

6-3, 301 lbs, 5.35 sec 40, RAS 7.76

Schmitz was one of the stars of the Senior Bowl and is considered by many to be the best center in the 2023 draft. He plays with athleticism, high football IQ and has nearly elite functional strength. Washington has started four players at center in each of the past two seasons. Chase Roullier is the Commanders best offensive lineman when healthy, but has only played 10 games in the in that span and turns 30 this season.

Schmitz will compete with Roullier and free agent acquisition Nick Gates in training camp. He has potential to win the starting job early in his time in Washington, giving the team the flexibility to walk away from Roullier’s expensive contract over the next two years and providing insurance in case he is injured again.

With OT Darnell Washington, TE Darnell Washington, and EDGE Isaiah Foskey off the board, Schmitz was the clear best player available on the DraftBot’s board. The next highest rated players were LB Jack Campbell and TE Luke Musgrave.

Round 3, Pick 82

Tyler Steen, OT, Alabama

6-6, 321 lbs, RAS 9.69

Steen is a raw OT prospect with exceptional athleticism for his size, but is still developing his blocking technique. He posted exceptional scores for a 321 lb lineman in the broad jump (9’1”) and short shuttle (4.59 sec), and also put up 31 reps in the bench press. The DraftBot has noted that Eric Bieniemy seems to have a preference for athletic linemen, and provides a depth piece at OT who fits the mold, with potential to take over swing tackle duties early and push for a starting position in time.

The DraftBot was disappointed that WRs Marvin Mims and Cedric Tillman, RB Devon Achane, and TE Tucker Kraft were selected shortly before this pick. Steen just edged out LB Dorian Williams, QB Tanner McKee, S JL Skinner, DT Zacch Pickens and WR Tyler Scott at this pick.

Round 3, Pick 97

Dorian Williams, LB, Tulane

6-1, 228 lbs, 4.49 sec 40, RAS 8.54

Williams adds a new dimension to the Commanders’ linebacker corps. He is lighter and faster than everyone but Jamin Davis and he excels in coverage. The DraftBot sees him as a fit to Jack Del Rio’s increasingly positionless backfield, providing an option to play the Buffalo Nickel and cover running backs and slot receivers with the ability to play pursuit against the run and spy on mobile quarterbacks. On a team that plays a second linebacker on approximately 60% of defensive snaps, he could prove to be a better option on passing downs in the outside role than current options Cody Barton and David Mayo. He also has the attributes to be a core special teams player.

Other players that the DraftBot liked who were still on the board were S JL Skinner, DT Zacch Pickens, and WR Tyler Scott. Tanner McKee was taken two players after Washington selected Steen at 82.

COLLEGE FOOTBALL: OCT 15 Auburn at Ole Miss
Photo by Kevin Langley/Icon Sportswire via Getty Images

Round 4, Pick 118

Jonathan Mingo, WR, Mississippi

6-2, 220 lbs, 4.46 40, RAS 9.93

The Commanders currently have three quality starters at WR, but there is no depth behind them to provide insurance in case of injury or succession planning at the position. No player on the current depth chart shows much promise of eventually pushing for a starting position. The DraftBot does not approve of career backups and seeks to replace them with young players with upside whenever possible.

Mingo is a big slot receiver with exceptional athleticism and offers something that the Commanders current WRs lack: size. He has good ball skills to make contested catches and is an above average blocker. He joins the team as WR4 where he provides an immediate option as a possession receiver and red zone target. Over time, he provides a potential alternative vision to replace Curtis Samuel with a bigger model at slot receiver.

No other player on the DraftBot’s board was close enough to Mingo at this pick to require a decision.

Round 5, Pick 151

Colby Wooden, IDL/EDGE, Auburn

6-4, 273 lbs, 4.79 sec 40, RAS 9.2

Wooden is a developmental DL prospect with elite athletic traits. He has transitioned through roles at LB to DE and interior defensive line as his frame has grown and developed. He is already a disruptive interior pass rusher, having recorded 11.5 tackles for loss, 6 sacks, 3 passes defended and 3 forced fumbles in 2022. He needs to develop better technique against the run. His best football is clearly ahead of him, but he will require creative defensive coaches to make optimal use of his physical gifts at the next level.

Wooden was available much later than expected, probably due to questions about position fit. The DraftBot is confident that Jack Del Rio and Jeff Zgonina have the creativity to maximize Wooden’s talents to make life difficult for opposing QBs in a year or two. No other player was close at this pick.

Round 5, Pick 154

Emil Ekiyor, G, Alabama

6-2, 314 lbs

Ekiyor plays with solid technique and understanding of the game, and has adequate athleticism and functional strength for his position. He will need to add strength to compensate for his lack of ideal size at the NFL level. He should be able to compete for a position on the depth chart early in his time in Washington and has potential to eventually earn a starting role with further development.

The DraftBot was disappointed to see DE K.J. Henry and OT Wanya Morris go in the preceding 10 picks. It would have been more disappointed to have missed CB Kei’Trel Clark at the previous pick if it had not drafted Brian Branch in the first round. Ekiyor was clearly separated from the next group of players at this pick, which included QB Jaren Hall, TE Brenton Strange, CB Darrell Luter, CB Jakorian Bennett and EDGE Vilami Fehoko.

Round 5, Pick 174

Brenton Strange TE/H-back, Penn State

6-4, 253 lbs, 4.7 sec 40, RAS 8.93

Strange has flown under the radar in a deep TE class. What attracted the DraftBot’s attention to him is that he excels as a lead blocker, both in-line and in space. It thinks these attributes will suit the H-back role, which Eric Bieniemy used in his offense in Kansas City. This skillset could come in handy as an extra blocker in a division where his QB will face Micah Parsons or Kayvon Thibodeaux four times a year.

Strange offers more upside and value as a receiver at the position than Bieniemy is used to. He demonstrated elite explosiveness (36 inch vertical, 10’ 4” broad jump) and great speed (4.7 sec 40, 1.57 sec 10-yd split) at the Combine. Despite modest receiving numbers in his final year at Penn State (32 rec, 362 yds, 5 TD), he has speed and acceleration to separate from linebackers and good ability to gain yards after the catch. Strange could provide a sneaky receiving threat on releases from play action and a good checkdown option in addition to blocking to help protect Washington’s young QB.

Other players in contention for this pick were CBs Darrell Luter, Rejzohn Wright and Jakorian Bennett. The DraftBot concluded that one of those players was almost certain to be available at its next pick and opted to give Bieniemy a new dimension for his offense. It would have taken QB Jaren Hall ahead of Strange, but he was selected nine picks earlier.

Round 6, Pick 178

Darrell Luter CB, South Alabama

6-0, 189 lbs, 4.46 sec 40, RAS 8.35

Luter is small school prospect who first caught scouts’ attention in Senior Bowl practices. He had a breakout junior season in 2021, with nine passes defended and four interceptions, then saw his numbers drop in 2022 as quarterbacks avoided throwing into his coverage. Scouting opinions on his coverage fit are mixed, with the majority saying he excels in zone coverage, flashing excellent instincts, awareness and ball skills. He has adequate speed and elite jumping ability (40.5” vertical, 5th in CB class), which he uses effectively to win on contested throws.

He will compete initially for a position on the CB depth chart with Rachad Wildgoose, Christian Holmes, Cameron Dantzler and Tariq Castro-Fields. In time, he might just push for a starting position.

Luter just edged out fellow CBs Jakorian Bennett (Maryland) and Rejzohn Wright (Oregon St) for this pick. Nothing happened since Washington’s last pick at 174 to alter the DraftBot’s board.

Round 6, Pick 192

Keaton Mitchell, RB, East Carolina

5-8, 179 lbs, 4.37 sec 40, RAS 8.66

Mitchell is an undersized running back who is a threat to take it to the house every time he touches the ball. According to PFF, in 2022 he led the FBS with 54 explosive runs of 10 yards or more. He is PFF’s third-highest graded rusher in the 2023 draft class, having run for 1,325 yards on 179 attempts (7.4 yds/att) and 13 TDs in 2022. He demonstrated elite explosiveness (38” vertical, 10’ 6” broad jump) and speed drills (4.37 sec 40, 1.5 sec 10 yd split) at the Combine and his Relative Athletic Score is only dragged down by his size.

At the NFL level, Mitchell provides an intriguing option on third downs and gadget plays. He is a big-play threat running the ball, and presents an interesting coverage challenge for linebackers on routes out of the backfield. In 2022, he had 27 receptions for 252 yards and 1 TD. However, his size will limit him to outside runs and he adds no value as a blocker.

The reason that the DraftBot could not resist Mitchell at this pick was for his potential in the return game. Dax Milne averaged 7.8 yards per return as Washington’s primary punt returner in 2022, which ranked a respectable, if unexciting, 12th in the NFL. However, his long return of 19 yards ranked 36th. The DraftBot sees a need for more explosiveness in the Commanders’ return game. Mitchell was used sparingly as a kick returner at East Carolina and averaged 26.6 yards per return in 2022. The DraftBot thinks that Mitchell has the decisiveness, footwork and agility to be an explosive threat on punt returns and that this may be his natural fit in the return game.

Three other players that the DraftBot considered at this pick were QB Dorian Thompson-Robinson, DE Durrell Nchami, and RB Eric Gray. TEs Davis Allen or Zack Kuntz might have been the pick, but were taken at picks 190 and 191, respectively.

UCLA v Arizona State
Photo by Bruce Yeung/Getty Images

Round 6, Pick 214

Dorian Thompson-Robinson QB, UCLA

6-2, 203 lbs, 4.56 sec 40, RAS 7.63

Thompson-Robinson is developmental dual threat QB prospect, who joins the Commanders as QB3 on the depth chart, to back up Commanders presumptive 2023 starter, Sam Howell. Thompson-Robinson uses his mobility to extend plays and create opportunities on the ground. In 2022, he completed 266 of 382 passing attempts (60.6) for 3,169 yards and 27 TDs to 10 interceptions. He also rushed for 645 yds and 12 TDs on 118 attempts (5.5 yds/att).

The caveat is that he put up those numbers in Chip Kelly’s option offense, which is favorable to QBs at the college level. He needs to get better at decision making and reading defenses to succeed at the next level, but has shown steady improvement in those areas throughout his time in college.

Thompson-Robinson was a clear-cut choice at 214. No players coming off the board after pick # 192 influenced the decision at 214.

Round 7, Pick 234

John Ojukwu OL, Boise State

6-6, 309 lbs, 5.24 sec 40, RAS 8.71

Ojukwu is a developmental OL prospect with an intriguing combination of size, strength and length. He played OT at Boise State but may be a better fit at guard in the NFL. He has good burst and agility and plays with power and aggression at the point of attack. However, he struggles to maintain leverage and adjust to mirror rushers in pass protection. The DraftBot is pleased to be able to give incoming OC Eric Bieniemy another athletic depth piece to work with.

Ojukwu will compete with draft classmate Emil Ekiyor, Saahdiq Charles, 2nd year player Chris Paul, Nolan Laufenberg and Keaton Sutherland for positions on the OL depth chart or a spot on the practice squad.


PFF Draft Grades

It is not possible to get a meaningful sense of how good or bad a draft class was until two or three years after the fact, let alone before the draft has even taken place, since most players taken after the first two rounds will take a few years to develop, and many players drafted early will start right away due to their draft position, and only later turn out to be busts. But that doesn’t stop anyone from handing out instant draft grades. These are usually based on one of two things: how well the mock drafter did at addressing immediate needs, even in the 7th round, and/or how well the mock drafter did at picking from the grader’s board.

PFF’s mock draft simulator takes the latter approach, which is, of course completely meaningless. Purely in the interest of fun, and nothing else, here is PFF’s report card on the DraftBot’s mock draft.


PFF liked the trade back three spots in Round 1 to net an additional third-round pick and three Day three picks. They absolutely loved the Keaton Mitchell pick, and liked Brian Branch and Colby Wooden. They thought the DraftBot’s neural network short circuited on the Brenton Strange and Darrell Luter picks. I might question whether it’s possible to give a Round 6 pick an F, but what do I know? And on the rest of the picks, it appears that they felt the DraftBot could have got better value.

The DraftBot reply was brief: “Silly biologicals. Their player rankings appear to be based on Combine test scores and subjective, voodoo player grades.”


You must be logged in to post a comment Login