We, as fantasy football managers, can be heavily swayed by anecdotal evidence and personal experience, and that can cause us to ignore larger trends that will help us find long-term success.
Maybe you drafted Josh Allen early in your drafts last year and are convinced that an early-round quarterback is an absolute must to win your league. Maybe you selected Christian McCaffrey and swore off of early-round running backs and would rather play the late-round and waiver-wire guessing game for your backs (which, by the way, is harder to do than you might think).
And while median fantasy football projections help get your bearings as we enter draft season, they don’t do a good job explaining the range-of-outcome reality that exists with fantasy prognostication.
Fantasy football is volatile. Injuries happen. Players are less (or more) efficient than expected. This means that — with, say, a reasonable discrepancy in touchdown rate from expectation — players can easily fall from a QB6 projection to a QB13 output, and we won’t think twice about it.
But in order to put some math behind this and help uncover the actual odds certain players return us with top-tier fantasy results, I went ahead and simulated the NFL season 10,000 times to see which running backs (and quarterbacks, wide receivers, and tight ends) are most likely to lead the league in fantasy points. This can help us uncover who has tangible, predictable league-winning upside for the 2022 season.
The Process and Expectations
It’s very easy to get stuck in neutral when constructing fantasy football projections. Yes, we want to be accurate, but trying to find a rigid answer to a rate stat isn’t going to help us very much.
A yards-per-carry average of 5.0 compared to 4.0 is a pretty big gap, but even over 100 attempts, that’s an extra 100 yards rushing, and that’s 10 fantasy points, far less than a point per game. We can get bogged down in efficiency numbers, and while I’m not saying we should try not to predict efficient players, it’s ultimately futile in the big picture of how fantasy football works.
So, instead, I have realized there is much more value in painting with broader strokes for an exercise such as this.
To get started, I went back and studied historical deviations and ranges of outcomes for players based on numberFire’s preseason fantasy football projections. This helped me tailor projections for the simulations.
From there, I accounted for historical injury rates among positions, replacements for injuries when they occur in the simulations, and — of course — actual fantasy point variance because, even if a player plays all 17 games, he may not score his exact projection.
One of the strange things about projecting running backs is that, if you don’t anticipate injuries, then it’s incredibly easy to over-project rushers with featured roles. Jonathan Taylor over a 17-game season is going to look fantastic compared to virtually every other running back, and the same goes for Christian McCaffrey. Full workloads for these guys are untouchable by most anyone else in the league.
That being said, we know there’s no guarantee either of them — or any back — plays all 17 games this season, so we need to account for variance if we’re really trying to see what expectations we can lay out for each player at the position.
In other words, we want these simulations (or I do, at least) to mirror history to feel accurate. Here is how buckets of backs have fared since 2012 based on average draft position, a proxy for preseason rankings in the projections.
|RB1 to RB12||7.6%||29.4%||54.6%||70.6%|
|RB13 to RB24||0.0%||6.8%||19.7%||51.3%|
|RB25 to RB36||0.0%||0.9%||9.6%||31.3%|
What does this mean? On average, backs taken inside the top 12 at the position have finished as top-five performers at a 29.4% clip. Put another way, take a random back in the top-12 at the position, and he should be around 30% likely to be a top-five back. (The top-12 in my simulations have a top-five rate of 29.4%.)
So, the odds in the simulations — if our math (i.e. baseline projections and standard deviations) is in order — should reflect these to a degree. We don’t want the model to suggest that the RB4 has a 90.0% chance to be a top-24 performer.
It’ll feel like he’s a lock to do that, but that’s not how history shows it to be once we account for injuries and performance. (Don’t worry; I checked the numbers, and they’re on par with historical results, so we should be good.)
Simulation Results and Analysis
Here are the results of the 10,000 season simulations and each running back’s average draft position in FanDuel’s best ball fantasy football formats.
A caveat here is that a lot of these are going to feel “wrong” across positions. It feels wrong to think that there’s a 30% chance that stud running back won’t even be in the top 24 by season’s end, but based on historical trends — and the mathematics behind the range of outcomes — it’s accurate. Some of them will get injured, and others will overperform.
The goal is to project 2022 not as the flowery, flawless preseason we all hope to see but as the utter mess that it always winds up being.
|Kenneth Walker III||36||0.1%||1.2%||7.1%||23.3%|
These numbers reflect historical expectations, perhaps, better than even any other position, which is kind of wild because of how volatile running back seems. The reality, though, is that when studs are healthy, they are the studs. When they are injured, though, it’s far from a guarantee that backups produce like we’d hope — let alone like true studs.
I spent a long time making sure that the simulations properly accounted for things such as missed games, and that’s what is most appealing in this conversation about running backs.
If we look at situations when a starter actually is injured and misses a significant portion of time, those don’t just magically result in top-12 or top-24 seasons for backups.
Since 2016, in 28 instances during which a team’s top projected back missed at least eight full games, those teams’ second-string backs accounted for two top-12 seasons (7.1%), and just five (17.9%) were top-24 backs.
Isolated within the context during which a starter gets injured for at least half of the season, those are pretty solid hit rates.
However, we aren’t drafting backup running backs with the foregone knowledge that their starters will be out half the season.
In fact, top-projected backs for each team since 2016 have averaged 12.6 games. It’s rare that we get injuries to such a heavy degree that it’ll alter full-season outcomes. In the full sample of seasons (regardless of whether a starter is hurt), only around 3.0% of team RB2s end up generating top-12 seasons.
So, basically, one backup running back this year across the 32 teams will have a top-12 year. That’s why the historical outcomes and simulations look the way that they do.
As for the individual analysis, the projections system at numberFire is heavy on Taylor (19.5%), McCaffrey (16.0%), and Derrick Henry (14.0%) with Austin Ekeler (9.3%) in a bit of his own tier in terms of pure RB1 upside. This makes sense because while one of these guys may stumble, they probably all won’t, and since 2016, each of the eventual RB1s has had a top-10 projection entering the season based on numberFire’s projections.
That’s why the rest of our top 10 — Najee Harris (5.1%), Dalvin Cook (4.6%), Joe Mixon (3.3%), D’Andre Swift (3.0%), Aaron Jones (3.0%), and Saquon Barkley (3.0%) — looks the way that it does for RB1 odds. In total, the top-12 have an average top-12 simulation rate of 7.2%, comparable to the historical rate of 7.6%.
Running back is a position that’s easy to complicate. When injuries don’t derail seasons, elite things happen from studs, and even when those injuries occur, we don’t see backups vault into the top-12 very often because there are still a long list of backs with elevated roles.
Given the harsh realities of the waiver-wire running back landscape, we must acknowledge that the league-winning backs still primarily come early in our fantasy football drafts.