The 2023 Fantasy Football Map
2023 Draft Strategy from a Look Back at the 2022 Fantasy Football Season
“We can never have enough knowledge to allow us to base all our decisions on knowledge. It follows that if a piece of knowledge has proved useful, we are liable to overexploit it to areas where it no longer applies, so that it becomes a fallacy," George Soros.
For those of you who missed my only other post in this project, The Fantasy Football History Project, I recognize that what I am about to do may seem a little strange. In the world of fantasy football, everything is about projections. Sure, data scientist enthusiasts might develop complex play-by-play analytics for the prior season, but it is all designed to evaluate which players are under/overvalued relative to cost for the current season. Studying the past seasons is rarely approached simply for the sake of studying the past, and fleshing out general principles from the patterns observed. You don’t buy last season’s stats.
The principles that come from historical study are called heuristics, simple rules to live by that guide through the fantasy season. The strange fact about heuristics is they often outperform sophisticated models of the upcoming season. These simple rules are robust. They can be generalized to new contexts, while models are more sterile and less adaptive. The draft day projections are appealing in that they project false certainty, but they are also useless on Monday of Week 2.
Real-life fantasy managers deal in evolving decision environments where error (i.e. injuries and busts) must be calibrated. Fantasy championships are won during the season, where FAAB waiver bids, trade requests sent, and start-sit decisions made. To be sure, nobody has foreknowledge of how the season will unfold in advance. None of us can know the terrain. But it does not follow that nothing can be known. We possess the map.
Although fantasy football is about adaption to evolving weekly risk, not season-long values, it is also important to understand the limits of in-season adaption, especially for larger leagues. In these leagues, there are some strategies, like selecting five consecutive picks at the same position, which cannot be overcome. With that in mind, I intend to explore positional breakdowns of draft values that I find puzzling.
Drafters are currently favoring wide receivers (~7 in 1st rd) over running backs (~4) for the first time since 2016, and more or less discarding the late round QB wisdom. It appears to me that it is no longer “late round QB, with exceptions”, top-3 QB in first four rounds, then wait. Instead, the top 3 of Mahomes, Hurts, and Allen in some order have slid solidly into the second round and with the run continuing with next group, including Burrow, Fields, Herbert, and Lamar, within the first four rounds. The combined effect of this is to slide the list of RB names down. David Montgomery in the 7th round? Yes, please!
For years now, consensus has been early RB, and late round QB. Together, they constituted a “map”. Of course, the principle in finance known as reflexivity applies. To the extent that everybody else uncritically accepts that map—the opposite strategy could prove fruitful. In Lamar’s MVP season, I took advantage of league where QB’s went unusually late, and selected him in the late rounds when the idea of him developing as a passer was still under serious skepticism. But reflexivity is not to be confused with a reversal of these fundamental strategies.
And with that, I think we can begin the analysis exploring exactly why is it these strategies among experienced fantasy players developed in the first place. The basic answer is that the season-long point totals, and even weekly averages, gleaned from a fantasy magazine or on Pro Football Reference are misleading. Fantasy football is a weekly game, and points do not come in the same way across positions, and vary widely by position relative to expectation. The headline finding is RB’s are the most predictable with those list of names expected to finish at the top of the position more often than not finishing that way, which is not true for either WR’s or TE’s. The question I am seeking to explore is has the basic decision environment changed so radically that we can justify overturning the map, deprioritizing RB’s in top 3 picks of drafts? To me, in a 12 team, and especially 14 or 16 team league, you shouldn’t feel secure without two top 20 RB’s in first three picks. Mathematically, not all teams will have the luxury of being secure, but you should jump at the opportunity if it presents itself.
The Macro Map: Positional Analysis
In any sort of analysis, nothing can be discovered without first devising a set of arbitrary rules. For every flex position during the fantasy season from weeks 1-17, we want to find if that player finished as a stinker, bust, boom, sleeper, or whether they met expectations. Additionally, we want a ratio for good (met, boom, or sleeper) to bad (stinker or bust) outcomes. Expectations are determined by FantasyPros rankings, whether in any particular week community the aggregation of fantasy experts thought a player was a good start, a weak start but still in consideration, and dart throws only considered if you are desperate. Those categories coincide with top 20, 21-40, 41+ for RB’s, top 24, 25-48, 49+ for WR’s, and top 8, 9-16, and 17+ for TE’s. They are arbitrary, but they relate to RB30, WR36, and TE 12 reference value-based drafting reference points. Bench grades actively hurt your roster that week, while players who finish as weak starts at least give you some points. The following chart summarizes how stinker, bust, boom, sleeper, and met were assigned. For instance, in Week 1, CeeDee Lamb was ranked as the WR6, but finished as the WR76, so he would fall in the dark red stinker section of the chart.
The first part of the analysis section is to find some suggestive evidence for why it’s generally better to have a roster top-loaded with RB’s, which held even in 2022 with many top-end WR season finishes. RB’s are more likely to deliver on expectation than WR’s, and WR’s, while more variable than RB’s, are still more likely to deliver on expectation than TE’s. This result makes sense. Good RB’s usually get 60+% of RB fantasy opportunities, while the very best WR’s only get a 25% target share, and all except Kelce and Andrews (more on them later) get even less than that at the TE position. Each opportunity of course is more valuable for pass-catchers, but because there are less of them the weekly finish has more variance. To demonstrate, here is a table of the fantasy finishes from weeks 1-17 who finished as a strong start at each flex position and rankings category (the green column in the above chart). The goal of fantasy football is to play as many strong starts as possible. So another way of interpreting this chart is the percent likelihood of a successful result at each rankings tier and position.
Recall the title of my first chart, fantasy happiness= reality-expectations. The implication of this chart is that assuming you must accept tradeoffs (i.e. you can’t have Travis Kelce, Cooper Kupp, and Austin Ekeler), you are likely to be happier if you play multiple solid top 20 RB’s, within the 24ish range at WR, and fish at the bottom of the barrel for a TD at TE.
And here is the full set of separate charts for each position, taking into account all combinations of possible finishes, strong and weak starts, and bench. The significant point to add is that strong start RB’s (15.3 %) are also significantly less likely than WR’s to finish with stinkers (23.8%), and WR’s less likely than top-8 TE’s (38.2%).
The Micro Map: Individual Player Analysis
Now that we have established the outline by position, we are ready to look at the aim of fantasy analysis, looking at performance by individual players. The last section discovered that all other things being equal, RB’s are more likely than pass-catchers to perform at least up to expectation. All other things are not equal. And you have to zoom out from the macroscopic positional level to the microscopic player level to discover which things are not equal, and make notes for draft day.1
The guiding principle remains constant from the positional to player level. For the first two rounds of your draft, you want as many players as possible with very few “stinker” performances, such that they were rated as benches on the week.
This list clarifies the first and second round targets. If a player had one stinker or less and a good/bad ratio over 1 (to account for players like Javonte who did not play enough games), they are a 1st round target. This includes the expected list of RB’s: Austin Ekeler, CMC, Saquon, Chubb, Josh Jacobs, Derrick Henry, and Bijan Robinson. I’m intentionally leaving out some names for later discussion. As expected, there are fewer pass-catchers on this list. But the pass-catchers who are on this list, because it’s so unusual for them to provide consistent production, are more valuable than all except the absolute best RB’s—Ekeler and CMC. Kupp was on pace for another all-time great season and Kelce lapped the field at the TE position. Kupp is my WR1 and I do not hesitate to take Kelce in the middle of the first round. Tyreek Hill is also worthy of a first round pick. Jefferson and Chase are of course both worthy of first-round picks, but note Jefferson is not on this list, and Chase has far more bust performances than is typical for a top 5 pick by ADP.
The next list of names is just as interesting. The exercise of identifying top round targets works much less well for names further down the list, but you can spot the names which I believe are 2nd round targets: Stefon Diggs, AJ Brown, Amon-Ra St. Brown, Davonte Adams, Jonathan Taylor, Travis Etienne, Chris Olave, and even Joe Mixon. The first four names are obvious, but they belong in the 2nd round, not the first. Add Tony Pollard, Dalvin Cook, Breece Hall, and Najee Harris from the previous list. I’m confident Dalvin Cook will get work wherever he signs, although there are certainly little signs of decline. Even though he is younger, the signs of decline have been much bigger for Joe Mixon—it is worth noting that, like Mike Evans, a good portion of his total points came in just one game—but they just have nobody on the depth chart. Any clarification on his situation and I’m very willing to take him.
I see a lot more risk in a rookie rushing QB-led offense associated with drafting JT than the typical drafter. I would make a similar argument for another player who I also view as talented and even less questions about the quality of the offense, DeAndre Swift. I won’t be owning any shares of either this year. Fantasy football is dumb in this respect, but both are gambles that they just won’t catch enough passes to stabilize weekly floor. Many talented backs have unfortunately fallen into this bin, including historically, Nick Chubb. JT feels like Nick Chubb; nobody doubts he is a good player, but needs 18 TD upside to finish number 1 overall. I’m increasingly in favor of Breece Hall over Najee Harris in the back end of the 2nd round. Like Ricky Bobby said in Talledega Nights, “if you’re not first, you’re last.” For what it’s worth, real football analysts like Breece over Najee. Do note that later in that conversation, they also confessed to the classic fantasy football bias of favoring a talented young player without the resume over proven production but just isn’t going to ever win you a week.
There is one name who is very much in consideration in the second round, who is way far down on this list because he had 8 stinkers in a promising rookie campaign, Garrett Wilson. The Vegas prop bet of 1150 yds and 8.5 TD’s is behind only Cooper Kupp, Davante Adams, Jamaar Chase, and Tyreek Hill. In the same way that I evaluate Breece Hall with the reptilian part of my fantasy football brain, Wilson as top 5 receiver feels right to me. And the very high number of stinkers from Wilson should come down for the simple reason that stinkers are related primarily to consistency of QB play—Chase had less than Jefferson who had less than Adams is related in this way Burrow>Cousins>Carr. Put me down for the 2020 and 2021 MVP, Aaron Rodgers> > > > >Zach Wilson. Carr’s consistency issues leads me to the other most tantalizing young WR on this list, Chris Olave. To be sure, betting Carr > > Red Rifle also seems safe, and the real football analysts are captivated by his tape. For all these reasons, an intriguing strategy this year for those who can execute it is to draft Bijan, then Garrett Wilson, then Chris Olave.
In a similar vein, there is another name, who is also in consideration in the second round, but is just off this list because all of his bad plays were stinkers, and none were standard busts, Rhamondre Stevenson.2
To finish off the young receivers list, notice I’m excluding Jaylen Waddle as a Round 2 target. If he doesn’t return to 100 catch season, he will be too inconsistent, more inconsistent than the other number 2 tied to a good offense, Tee Higgins. To be honest, I should probably jump Jonathan Taylor above Tyreek Hill as well. The Dolphins offense was just so electric last season. The Dolphins are Niners East, and Mike McDaniel is Shanahan East. It is stupid to bet against either.
Here is a summary of my first two rounds, in a very loose order as I talk things through and discover new information. First round: CMC, Ekeler, Kupp, Kelce, Bijan, Saquon, Jefferson, Chase, Nick Chubb, Henry; followed by the RB’s that I don’t want to take but will if I get stuck at the end, Jacobs and Jonathan Taylor (8 RB’s/ 4 pass-catchers). Second round: Tyreek Hill, Stefon Diggs, CeeDee Lamb, Davante Adams, Tony Pollard, AJ Brown, Breece Hall, Najee Harris, Garrett Wilson, Travis Etienne, followed by a list of RB’s who have second round grades but big question marks until free agents are signed, Rhamondre Stevenson and Joe Mixon. I will not participate in the Hurts/Mahomes/Allen run so long as above list is available.
How to Approach the Rest of the draft
As for the players who were not discussed on the above lists, there are two categories:
Players who will be worse than they were last season, but not as bad as their slides in drafts implies. James Conner had zero stinkers last season! Why? He caught passes and was used at the goal-line. Think of him the same way fantasy players used to spend a sixth rounder on James White. On the same team, Hopkins is still very good at football and is being drafted as if he weren’t. These are the 2022 stats with Colt McCoy, useful in PPR:
Hollywood was also very good last season. Nobody is salivating over the Baker Mayfield-led Bucs, but Godwin will also be fine. Refer back to the above table on Hopkins—PPR friendly receivers if they are the #1 are somewhat resistant to inconsistency at the QB position. “If they are the #1” caveat is important. I do not want to touch Diontae Johnson or Courtland Sutton. Lockett or Pittman is an interesting choice in that range, and I expect the good/bad ratio for both players to hold. Lockett is in the category of unexciting veteran who contributes on championship rosters because he doesn’t bust.3
Then there are players who are rising in value because their production is expected to jump, either from nothing or from useful to top-end. I am still in on Ken Walker, 2nd round draft investment in Zach Charbonnet aside. He is an ideal target if you went WR/WR, TE/WR. And if for some reason you waited even longer, Damion Pierce is very enticing. The new crop of top 10 RB’s from a dynasty perspective to replace Zeke, Cook, and Fournette is Hall, Walker, and Pierce. Montgomery is the last of the non-speculative RB assets for drafters who took a shot at QB or TE in the middle rounds. He will do what Jamaal Williams did last season, blow-up weeks interspersed with disappointing finishes (Williams—7 starts, 4 OK, 5 stinkers). If you want to bet on Sean Payton tapping his offensive mind for Russell “Hustle and Bustle” Wilson resurgence, Jerry Jeudy safely beat out Sutton in their WR contest last season.
Cam Akers, Raachad White, Christian Watson, Isiah Pacheco, and Alexander Mattison were good with limited fantasy expectations. Will they still be good now that their NFL role and fantasy expectations are both expected to increase?
This post started on heuristics. One of the most famous fantasy football heuristics is TE’s break out in Year 3. I want one share of Kyle Pitts for this reason.
Now for the fun part…..
We will get to see this season who is right. Today, FantasyPros posted a similar draft strategy article which offers a diametrically opposed view. I recommend following the logic on how the author, Fitzmaurice, argues “WR’s are stable assets, while RB’s are unstable” because I think it is the consensus perception that is driving draft values and rankings that look like this.
I argued in this post that the first two rounds should be reversed with 8 RB’s/4 pass-catchers going in the first 12 picks, and more balanced split in the second round. The fallacy Fitzmaurice fixates on is average points. Categorize individual games into strong starts, weak starts, and stinkers like I’ve done here and you get the opposite result—RB’s are the stable assets from a weekly perspective. Nothing is more valuable in fantasy football than a RB like Austin Ekeler was last year, a player who is never bad. Revealingly, Fitzmaurice massively undervalues him relative to consensus.
Though, in a sense, the author is right. Taking two RB’s with your first two picks as I’m likely to do given the current costs is a strategy that is guided by risk aversion. The aim is to go 8-5 and be part of the pack that is in playoff contention. If the aim is to go 11-2 and cruise to winning your league, you should take Justin Jefferson/Chase, followed by Garrett Wilson (we agree on Wilson), and then hope you hit on the RB’s I listed off in the middle rounds. I admit that it’s a more appealing group than usual and the strategy might just work, because the range of outcomes for all of those players, even Isiah Pacheco, is top-10. That would mean you would have the prized RB others paid up for AND the WR studs who will provide the week-winning performances.
Matthew Berry’s number one rule is to minimize weekly risk. Everybody comes to that question slightly differently because they have different tolerances for how far they are willing to turn the risk dial. But everybody also agrees that if for whatever reason you don’t hit in the middle rounds—Jerrick McKinnon is still too involved, Cam Akers is Week 1 MIA Akers, Mattison was just a small sample size darling, or Arizona and Tampa can’t move the ball—and you don’t have two startable RB’s, you are screwed until you spend FAAB on the hot handcuff moved into starting role. In large leagues, they are owned too, so you are just screwed. I draft guys like Najee Harris who will plod their way to 1000ish yards, 8 TD’s, and 45 catches because I want to avoid that scenario. The same is not true for WR’s. There will be a list of 50 receivers who may not hit even a 1000 yards, but will have 5 week stretches which are useable. FAAB on those guys is usually 10-15 percent if you need them in a pinch.
Happy drafting to everybody who has friends like me who are obsessed with fantasy football and start way too early!
The link is is to an instructional video on theoretical physics. But the idea about emergence really has broad relevance. The macro (positional) theory and the micro (individual player) theory both describe reality, it is just a question of which theory is more fit for purpose.
Think of Stevenson as the anti-Nick Chubb, who had a handful of busts but no stinkers. A handful of stinkers but no busts is worth a one round downgrade.
At least not since a very bad stretch of games he had two years ago, where he went MIA.