RADIO
Glenn analyzes auto-drafting vs. live-drafting using the 2023 ADP and the "Teams in a Vacuum" (TIAV) concept. "Old Maid" picks are randomly inserted into the fantasy football draft process as part of the Teams In A Vacuum analysis.
“I kicked your butt and I auto-picked!” How many of us so-called “experts” have heard that?
And to add insult to injury, before this statement, I warned the guy, “This is an advanced league where you pick defensive players instead of team defenses. Auto-draft will never be able to handle that.”
Damn that artificial intelligence! So how is it that auto-drafted teams beat some of our teams that were meticulously built by hand after hours of research?
Editor's Note: The FFPC Baby Gorilla Tournament is now open, featuring a $100,000 grand prize and a $675,450 total prize pool! This 12-team, Tight End Premium contest uses a 20-round draft format, with the overall winners determined by total points scored during Weeks 15–17. Get $25 to use toward your first entry by signing up through our link. Grab your team now! Sign Up Now!
Teams In A Vacuum Simulating Auto-Draft
How can auto-drafted teams beat hand-picked teams? Let’s start with the first round. In fantasy sports, there’s a saying: "You can’t win your league with your first pick, but you can definitely lose it.”
Table 1 – 2023 Average Draft Position With 10-week Fantasy Point Totals
| Standard League Scoring | Point-Per-Reception Scoring | |||||||
| ADP | Name | Position | Fan Points | ADP | Name | Position | Fan Points | |
| 1 | Christian McCaffrey | RB1 | 182 | 1 | Justin Jefferson (IR) | WR1 | 109 | |
| 2 | Justin Jefferson (IR) | WR1 | 73 | 2 | Christian McCaffrey | RB1 | 220 | |
| 3 | Austin Ekeler | RB2 | 86 | 3 | Ja'Marr Chase | WR2 | 180 | |
| 4 | Ja'Marr Chase | WR2 | 111 | 4 | Austin Ekeler | RB2 | 108 | |
| 5 | Travis Kelce | TE1 | 84 | 5 | Tyreek Hill | WR3 | 224 | |
| 6 | Tyreek Hill | WR3 | 155 | 6 | Travis Kelce | TE1 | 141 | |
| 7 | Bijan Robinson | RB3 | 102 | 7 | Cooper Kupp | WR4 | 68 | |
| 8 | Cooper Kupp | WR4 | 44 | 8 | Bijan Robinson | RB3 | 131 | |
| 9 | Nick Chubb (IR) | RB4 | 19 | 9 | Saquon Barkley | RB4 | 106 | |
| 10 | Saquon Barkley | RB5 | 83 | 10 | Stefon Diggs | WR5 | 201 | |
| 11 | Derrick Henry | RB6 | 106 | 11 | CeeDee Lamb | WR6 | 198 | |
| 12 | Stefon Diggs | WR5 | 128 | 12 | Nick Chubb (IR) | RB5 | 23 | |
| 13 | CeeDee Lamb | WR6 | 130 | 13 | A.J. Brown | WR7 | 204 | |
| 14 | A.J. Brown | WR7 | 137 | 14 | Davante Adams | WR8 | 141 | |
| 15 | Patrick Mahomes | QB1 | 172 | 15 | Tony Pollard | RB6 | 113 | |
| 16 | Davante Adams | WR8 | 84 | 16 | Amon-Ra St. Brown | WR9 | 170 | |
Getting the obvious out of the way, whether you auto-draft or pick yourself, injuries happen. Because of two consecutive injury-plagued seasons, many fantasy players with the first pick will not draft Christian McCaffrey. Using the Draft Sharks Preseason Injury Predictor, Christian McCaffrey (82%) and Tyreek Hill (91%) had higher injury predictions than Nick Chubb (63%) and Justin Jefferson (40%). Injuries are not predictable.
Disregarding injuries, what does auto-draft do that’s so special? In the words of Kyle Reese from The Terminator, “It can’t be bargained with. It can’t be reasoned with. It doesn’t feel pity, or remorse, or fear.” Auto-draft doesn’t have an opinion or bias about players, it picks the next one from the list. Anyone who’s participated in a live draft that had people auto-drafting must have noticed as players are taken out of order, a queue of “old maids” accumulates and auto-draft comes along and gobbles them up. In this particular PPR league, Tyreek Hill was auto-picked in the eighth spot by the guy who now claims brilliance.
Old Maids Are Created Randomly
So how do you simulate one team auto-drafting while the others use their brains to draft? You write software that uses Average Draft Position (ADP) to draft teams into leagues. The standard Teams In A Vacuum (TIAV) drafts teams into different league sizes, correlates the fantasy points scored by each player, and tallies each team's score to produce a winner. To simulate the undrafted old maids, the software randomly swaps the draft position of the last team with one of the other late draft positions.
For example, in a 10-team league, Team 10 randomly swaps its pick with Team 7, Team 8, or Team 9. The limitation in the size of the range is saying that pick 10 is never going to be so lucky that one of the top ADP players will drop to the last pick in the round. An example snake draft may have Team 10 with picks 7, 11, 29, 31, 48, 51, and 68. Similar old-maid simulations are done for 12-team and 16-team leagues.
Table 2 – 2023 Example Auto-Draft Winning Teams Using 10-week Fantasy Point Totals
| 2023 | 12-Team Standard League | 2023 | 12-Team Point-Per-Reception League | |||||
| Draft Rank | Player Name | Fan Points | Final Pos Rank | Draft Rank | Player Name | Fan Points | Final Pos Rank | |
| 11 | Derrick Henry | 106 | RB6 | 10 | Stefon Diggs | 201 | WR5 | |
| 13 | CeeDee Lamb | 130 | WR6 | 13 | A.J. Brown | 203 | WR7 | |
| 32 | Kenneth Walker | 119 | RB13 | 35 | Lamar Jackson | 183 | QB4 | |
| 37 | Jonathan Taylor | 64 | RB15 - FLX | 37 | Keenan Allen | 205 | WR18 -FLX | |
| 58 | Trevor Lawrence | 123 | QB8 | 59 | Cam Akers | 46 | RB22 | |
| 62 | Drake London | 60 | WR25 | 63 | Rachaad White | 132 | RB23 | |
| 95 | David Njoku | 47 | TE10 | 136 | Tyler Higbee | 54 | TE13 | |
| Team 12 | Total | 649 | Team 12 | Total | 1024 | |||
The standard TIAV algorithm snake drafts team comprised of the following positions:
- 1 QB
- 2 RBs
- 2 Wide Receivers
- 1 Tight End
- 1 Flex -- best available Running Back, Wide Receiver, or Tight End when the other position is filled
Because there is a random component to this simulation:
- Only 2023 fantasy points are used
- Leagues are drafted 1000 times to see if auto-draft has a pattern of winning
Table 3 – Auto-Draft Winning Team Totals Over 1,000 Drafts
| League Type | Number Of Teams | Expected Win Percentage | Auto-Draft Winners |
| Standard | 10 | 10% | 4.2% |
| 12 | 8.33% | 3.25% | |
| 16 | 6.25% | 3.4% | |
| Point-Per-Reception | 10 | 10% | 10.9% |
| 12 | 8.33% | 24% | |
| 16 | 6.25% | 5.6% | |
| Total | 8.20% | 7.99% |
Although the simulation gets a variety of results, the cumulative number of wins for auto-draft is close to an exact match of the expected number of wins. Although most league sizes do not produce enough auto-draft wins to reach the expected number, 12-team PPR was a sweet spot in the 2023 draft.
As seen in Table 1, if the last team in a 12-team PPR league was lucky enough to grab an old maid of Stefon Diggs (201) or CeeDee Lamb (198) plus its normal wrap-around pick A.J. Brown (204), Team 12 will win a disproportionate number of times. This is true for an auto-drafted team or a live-drafted team; however, how many people believe in the WR-WR strategy for their first two picks? Once again, auto-draft doesn’t care!
No matter how much studying we do, simulations we run, or heated arguments we have with our competitors, we must acknowledge that fantasy football is a game of chance. If you randomly picked a number from 1 to 16, Table 1 shows you are just as likely to pick an underachiever as an overachiever. Auto-draft is making its selections without emotion.
But we are not machines. Fantasy football is about opinions. The idea is, you are trying to predict which players will score the most points. Look at the composite rankings, listen to the experts' projections, and then decide for yourself.
I would like to close with the words of Andy Behrens, who I met at this year’s Fantasy Football Expo. If you are drafting in a Yahoo league, two minutes before the draft starts, a video of Andy pops up and he instructs you who you should draft with your first pick. I said, “Thanks for the input but I rarely agree with your video.” Andy’s response, “You’re supposed to have your favorites. That’s the fun part!”
Download Our Free News & Alerts Mobile App
Like what you see? Download our updated fantasy football app for iPhone and Android with 24x7 player news, injury alerts, rankings, starts/sits & more. All free!
More Fantasy Football Analysis
