Lessons Learned From Optimized Auction Lineups (Fantasy Football)
I recently did a series of articles breaking down Average Auction Value Tiers (AAVT) based on a series of real money auctions I participated in between February and July. AAVTs are handy in creating budgets, looking for arbitrage opportunities, and developing nomination strategies but I wanted to take it further. I wanted to know how to budget for the best possible starting lineup using Average Auction Values(AAVs) from real auction drafts. I also wanted to know what would happen if I wanted to budget differently or how to adjust if one of the players in the optimized lineup got too expensive. The results blew me away.
The lineups the optimizer produced made me rethink how I approach auction and snake drafts. Even though the optimized results were for auction drafts, most of the lineups are roughly achievable in snake drafts. With these lineups in mind, I viewed some recent snake drafts and recognize some drafters making critical mistakes in the early rounds. Here are the results and the lessons learned from these optimizations.
I pulled projections from the Ultimate Draft Kit*. Each of the Fantasy Footballers has their own set of projections and they all finished 2019 in the top-20 most-accurate rankers according to FantasyPros, but Jason Moore finished highest as the sixth-most accurate expert ranker so I used his projections.
*I displayed the total projected points, but you must access the UDK to view projections for individual players.
I used AAVs from my 12-team leagues that start with a $200 budget. The leagues use PPR scoring with a starting lineup consisting of: QB/2RB/3WR/TE/Flex.
I created an optimizer program that would generate the best possible combination of players within budget and starting roster constraints. I ran the optimizer to maximize the starting eight positions with no more than a $185 budget. The remaining $15 or more would normally be used to fill in bench spots, kickers, and defenses.
I also upped Clyde Edwards-Helaire’s auction value to a conservative $42(around Dalvin Cook‘s AAV). CEH’s AAV had been around $25, but his projection skyrocketed with the news of Damien Williams sitting out the season and his auction prices will reflect that.
A few QBs were projected near QB1 territory but weren’t top-12 in AAV so I included a ‘QB-X’ with a top-12 QB projection and a $1 AAV. Think of a QB like Ben Roethlisberger–Matthew Stafford–Cam Newton–Baker Mayfield–Jared Goff. Whichever late-round QB you like.
[lptw_table id=”164482″ style=”default”]
I never thought an optimized lineup would exclude most of the top RBs in favor of the top WR in Michael Thomas. I was also surprised to see how the budget broke down for the position groups. WRs led with $89, RB $50, QB $26, and TE $19. I expected to see our one dollar ‘QB-X’ with at least $150 go to the WR and RB position. I didn’t expect the optimizer to pay up at the TE and QB position. Then again, Jason’s projection for Lamar Jackson is quite high.
I ran the optimizer again with the lowest Jackson projection (Andy Holloway…still quite high though).
Andy’s Projection for Lamar Jackson
[lptw_table id=”164492″ style=”default”]
Almost the same team, but the optimizer still choose one of the higher projected QBs and used the extra money to upgrade at TE. I like using the Late-Round-QB (LRQB) popularized by J.J. Zachariason, which roughly translates to QBs that are $6 or less so I added a constraint for the third run.
[lptw_table id=”164493″ style=”default”]
And this is why folks should do auction drafts. Auction values are likely to increase as we get closer to the year so I think this Christian McCaffrey-Michael Thomas combo will be a huge stretch. Obviously snake-drafters can’t get the McCaffrey-Thomas combination (or Thomas-Kenyan Drake), but McCaffrey-Kenny Golladay, Thomas-Josh Jacobs, or Davante Adams-Drake are possible. A pattern is definitely emerging with some of the same RBs and WRs popping up in lineups.
Michael Thomas increased to $55
Christian McCaffrey increased to $65
[lptw_table id=”164501″ style=”default”]
Not a very practical lineup for snake-drafters, but the shift to a stronger WR corps is interesting.
Michael Thomas increased to $55
Christian McCaffrey increased to $65
Must draft three RBs
[lptw_table id=”164503″ style=”default”]
The total projection takes a nose-dive and it’s pretty rough to trust David Johnson and Todd Gurley, but it’s not fair to compare a lineup that forces three RBs to the other runs without all the extra constraints.
Must draft three RBs
[lptw_table id=”164504″ style=”default”]
The total score popped up again, but jamming three RBs into the lineup isn’t giving us the projected points we had while starting four WRs.
Implications for Snake Drafts
Most of these lineups are roughly achievable in regular snake drafts, but they ignore strict adherence to a lot of trendy draft strategies. The buzz around fantasy football circles has been about certain strategies like Zero-RB(not drafting an RB until the sixth round in most cases), Zero-WR(not drafting a WR until the sixth round), Robust-RB(drafting three or four RBs in a row), LRQB, and Drafting with Upside/Risk. The key appears to be balanced especially in the first four rounds, but there are elements that we should take away from each of these strategies.
Zero-RB focuses on creating huge positional advantages at the WR and TE position with the intent of taking advantage of the chaos that is more likely to strike the RB position like injuries or depth chart changes(We’re not rooting for injuries, it’s just an unfortunate reality). The optimizer clearly preferred targeting early-round RBs, but it still prioritized spending draft capital on the WR position with lineups that always featured four WRs. While the optimizer didn’t target low-tier RBs, the balanced teams it did produce would have needed to focus on RBs in the second half of the draft and waiver wire to survive injuries, deal with bye weeks, and find ‘league-winners.’
The optimizer focused draft capital on WRs normally found after the fifth round of snake drafts like the Zero-WR strategy does, but it still favored targeting stud WRs normally drafted in the first two rounds. The idea behind Zero-WR is to jam rosters with early-round RBs, TEs, and QBs while taking advantage of WR value in the mid- to late rounds. The problem is, at some point you start to lose value from the RB position compared to the other position groups and breakout WRs are difficult to find in later rounds. Still, the optimizer showed that two early-round RBs are a must and that a stud QB and/or TE go a long way in an optimized lineup.
Starting a draft with two RBs is perfectly fine, but you’re starting to lose value pushing for a third RB over a stud WR, TE, or QB. There are some upside-RBs available after the fourth round, but you probably need to punt TE and QB to the later rounds while you hammer WRs.
LRQB highlights the weakness of the optimizer, which is its inability to recognize risk/reward and chaos. If we knew that Lamar Jackson would produce 350 or more fantasy points, then yes; he probably is an auto-pick in the first few rounds, but we don’t know that. His rushing ability is without question, but his 36 passing TDs and 9% TD-rate seemed primed for some regression. More importantly, there is some QB going outside the top-12 that is going to have a huge season. It happens every year, we just don’t know exactly who yet. By drafting Jackson(or Patrick Mahomes) early, we pass up on a stud at one of the other positions and give up on a QB lottery ticket later. We also know that streaming QBs is a viable strategy that produces high-end QB results and carries with it nearly zero injury (COVID) risk.
The optimizer prioritizes value, but can’t recognize upside or risk. The optimizer is banking on RBs like David Johnson, Todd Gurley, and James Conner to produce. They’re all projected to have very useful seasons, but they also carry massive risks and limited upside. The risk drafting one of these guys early gets compounded if you wait until the seventh round or later to draft your third RB. If I pass on RBs early, I’m probably shooting for either a high-upside RB like D’Andre Swift or someone with a guaranteed role like James White or Devin Singletary as my third RB over another WR.
Applying Optimized Lineups to Draft Position
Snake Drafts can normally be divided into three sections in 2020; The First Four, Middle, and the Back-Turn. The First Four section has by far the largest advantage and it’s not just because of the players they draft first. If you look at draft boards, most of the top TEs and QBs are getting drafted by teams in the first handful of draft positions. Teams in the Middle and Back-Turn sections are going to need to figure out a way to gain positional advantages because they will not be able to construct their teams the way teams from the First Four are constructed and expect to win.
The First Four
You’re almost certainly starting with an RB. There’s an argument for Michael Thomas in this group, but most drafts will start with four consecutive RBs especially with the CEH being hyped out of his mind. These first few positions are so powerful because they can take advantage of all the talent that falls to them in the second and third rounds that include elite TEs and QBs. There’s really no “wrong” start except for a RB-TE-QB start. You would pretty much be forced to draft an RB in the fourth round and then your WR corps becomes a disaster.
The Middle Section
Unfortunately, you probably miss out on the elite RBs, which means it will be tough to beat teams from the First Four unless you build your team differently. The good news is, you’ll have plenty of opportunities to draft RBs in the second, third, and fourth rounds leaving you plenty of options to start your draft depending on how you view those RBs. Your best bet to create an optimized lineup is to prioritize drafting two WRs and two RBs with your first four picks. The fifth round is loaded with talent at all the position groups. By starting with two RBs and two WRs, you can let the draft come to you in the fifth. Be careful starting a draft WR-WR or WR-TE because it kicks off a Zero-RB slippery slope.
The Back-Turn Section
In past years, this was the best place to draft a Zero-RB type team because the WRs available around the first turn were capable of putting up 300+ PPR seasons, but that upside probably isn’t there (with the possible exception of a Davante Adams-Tyreek Hill start). What we do have are useful RBs available in the first four rounds and the best TE in the league. Grabbing an RB in the first or second round is probably a must. RB-RB or RB-TE might be the best way to separate your team from the teams that pick early. Some combination of RB-WR can work as well, but a WR-WR start will jam you up at the RB position and you’ll have to punt QB and TE.
Final Thoughts and Balanced Draft Tenets
After running the optimizer, I started looking at draft boards (auction and snake) differently. I don’t think drafters think enough about how decisions made in the early parts of drafts affect them later. You can easily identify where teams go wrong and how teams that skew too heavily towards certain position groups early tend to throw a bunch of ‘stuff’ at the wall to see what sticks later.
If you want to achieve something following a balanced (optimized) draft approach, use the following tenets:
- Draft two RBs in the first four rounds
- Draft at least one WR in the first four rounds
- Plan on drafting at least three WRs between the fifth and eighth rounds
- Only draft a QB if you already have two RBs or will draft a second RB in the next round
- Only draft a TE in the second round if you already drafted an RB
- Do not draft a TE and QB back-to-back before the fourth round
- Prioritize drafting a top-tier TE over a QB
- Avoid a WR-WR or WR-TE start to the draft
- Try to get one of the first four picks, but if you always get stuck with a late pick-do more auction drafts
- Marvin Jones or Marquise Brown in the Flex for the win
Love the auction content. I have same problem with any auction stuff I read – prices are not in line with my leagues prices. We are in our 13th (i think) year as a money league, so everyone takes it pretty serious. And Lamar and MAhomes both went for $45. And CMC went close to $70 i believe. I always read auction prices and think “our league will go way higher”. So it’s tough to get good auction advice out there. Ballers are the best in general, maybe it’s just hard to give auction advice because league prices have so much variance?
I appreciate the auction content and I’d love for there to be more of it. But, I just need to say this…and I absolutely love the Fantasy Footballers, but I cannot take any of the auction values seriously. CMC will be going for $70+ come draft day. Davante Adams for $38? I mean…at minimum $50. All of the Ballers price projections for each player seem way off. The elite guys are way under-priced and then middle tier guys are way over priced. It seems like little effort is put into auctions and all of the attention is put on snake drafts. The UDK is awesome but the auction piece is basically worthless.
Where is Michael Thomas going for 49 dollars? I have only seen him go above 60 in every mock I have done.
Funnily enough, my summer project was this same basic idea using a genetic algorithm to optimize an auction-based starting lineup, and it too kept coming up with Lamar and Mahomes at starting QB.