Finding 2020's League-Winning Fantasy Running Backs Through Season Simulations
This one might get ugly, but we're going to do it anyway.
After running NFL season simulations to find quarterback upside already, I'm moving on to the hyper volatile running back position. Elite running backs are fantasy game-changers, so why not figure out how to find those with true upside odds?
It seems pretty basic.
The problem? When you really peel back the camera and look at historical bust rates at the position, drafting running backs seems really risky. But on the other hand, it's super hard to replace early-round backs from the waiver-wire or later in the draft.
Put another way: these league-winning assets are risky, and we need to figure out the risk/reward in this scenario.
If you're like me and you aren't clairvoyant, then we have to embrace the fact that we can be wrong -- easily -- when it comes to running backs. But, I can't stress this enough: we should not be overly confident that we find stud running backs off the waiver wire.
Even if we don't know which running backs will stay healthy and productive, simply knowing the percentages can allow us to draft accordingly when the time comes.
Projections, Simulations, and Historical Precedent
I mentioned this in my quarterback piece and will do it again here, but it's really easy to get bogged down in details when building out fantasy projections.
Will Saquon Barkley's 4.62 yards per carry rate climb back up toward the 5.00 mark he had in 2018 because the New York Giants drafted an offensive tackle fourth overall? Does it really matter? How much does it really matter? (The difference between those two rates over 250 carries is 9.5 fantasy points for the season.) What are the odds you nail his yards per carry down to the hundredth -- or even the tenth -- anyway?
It's incredibly easy to get caught up in the minor stats that are really tough to pinpoint, and that's why I find that the most value from projections comes from understanding opportunity available within a team. Are you projecting Melvin Gordon for 215 carries? What does that really leave behind for Phillip Lindsay? Things like that. That matters.
And, no, I'm not saying that we should just randomize rate stats for players, but we can't predict them easily, and for running backs, volume matters a ton when we're looking at a per-carry averages that fit a tight window between four and five, generally. No back is going to average 12.0 yards per carry over a 200-carry sample.
Moving on: I've built and audited my projections for this season. Median projections. I use numberFire's Net Expected Points (NEP) data, success rates, average target depth, and a few more advanced stats to project expected basic rate stats. Yards per carry in my model stems from Rushing NEP per carry and Rushing Success Rate -- not from last year's yards per carry. It's been more reliable this way.
But, hey, you want to know a dirty secret about running back projections? Most of them aren't median projections. Even the ones that are meant to be. A good portion of RB1s are projected for a high-end outcome as their baseline because, if they stay healthy, they can flirt with 300 touches and 1,500 yards.
There's only so much room to go up from there and so, so, so (so) much room to go down from that volume projection.
We see that reflected in year-end, look-back studies in bust rates, but we rarely fully account for the range of outcomes entering the season for running backs. Before anyone yells at me that the ZeroRB drafting philosophy is espousing this, I get it. I just want to get actual numbers on these dudes.
Are some players more volatile than others? That's probably fair to say, but has anything really shown to help predict performance over a four-year sample of preseason projections to actual fantasy points scored? What I'm asking here is: do backs who expected to catch a lot of passes show a higher level of consistency than those who are primary rushers?
No. Nothing that we can really hang onto.
Before getting started (I know, it's long, but it's important), here is what the past seven years have shown us has happened with running back picks (the average draft position data comes from MyFantasyLeague).
|Finish Tier Odds
by ADP (2013-2019)
|Worse Than 36
Just a bit of a warning: the sims spit out data that might look stupid, but for the most part, they jive with what we've actually seen from historical hit rates. If anything, they may err on the side of optimism in most cases.
However, my goal here is to balance the actual projections and ranges of outcomes to identify which fantasy backs actually have high-end, projectable upside based on historical variance.
To find that out, here's what I did: I took my baseline, "median" simulations and accounted for historical variance and standard deviations in year-long projections in order to simulate the 2020 season 10,000 times. This will bake in injury rates for running backs.
The goal, effectively, is to project 2020 not as the flowery, flawless preseason we all hope to see but as the utter mess that it always winds up being.
Okay, so we know from the table above that the top-six preseason backs are actually top-six backs less than half the time but are top-12 backs around 60% of the time. I get that it looks weird projecting someone like Saquon Barkley as only two-thirds likely to be a fantasy RB1, but that's actually generous, historically, if we're trying to capture all types of risk involved in early running backs.
After tons of tweaks and adjustments, I got results that replicate historical rates well, and that's what the goal was here. The table here is sorted by the odds a specific player finishes as a top-12 running back in the simulations.
|Player||Top-6||Top-12||Top-24||Worse Than RB36|
You can argue with the general rank of a player here, but what matters is the idea behind the data, the historical trends. and likelihood behind it all.
Thought Experiments and Potential Ramblings
The sims show that top-6 and top-12 performances come from early-round picks. That shouldn't surprise us at all, and it's not exactly groundbreaking.
I've found similar information when examining waiver-wire running back performance. Those early-round picks aren't particularly safe, but if you don't take the risk with them, then your odds of getting a high-end performance are quite low.
And if you're thinking like a ZeroRB drafter, you might think something like this: "I don't need the RB12 for the full season. I just need the RB12 on a per-game basis while he's healthy, and then I'll ditch him for another back who isn't hurt." You're not wrong, but you're kind of wrong. The idea is right. The execution isn't that easy.
I'm once again going to point you to my latest study on waiver-wire running backs. Waiver pickups can't generally come close to replacing high-end running back performances, but if you're seeking RB3-level returns from waiver-wire backs, that's more realistic. That checks out in the preseason projections, as well.
What exactly separates some of these backs from the others? For one, projectable workload.
Snap rates tell us a ton about fantasy points for running backs. Very few backs averaged elite snap rates for extended periods last year, and that's why someone such as Christian McCaffrey was such a stud. That and the fact that he was just a stud.
Also, if you dig back into the history books, you'll find that it's touchdowns that really separate the elite assets in a year from the rest of the pack. Outliers do come in the yardage department, but it's the touchdown rates and totals that give us the game-changing performances over the bigger picture.
For that reason, projected touchdown totals and projected team touchdown totals factored into the ranges of outcomes for individual players.
And, again, you can flat out disagree with where a name is on this list, but you'd be ignoring a lot of historical precedent by thinking that a Tier 3 running back (projected for 60 or 70 points off the mark of Tier 1) has just as good a shot as finishing as an RB1.
Every time I dig into running backs, I get a little frustrated because it's sort of a damned if you do, damned if you don't type of situation. You want someone with Christian McCaffrey, Saquon Barkley, Ezekiel Elliott upside? You have to burn a top pick on a player who can easily underperform significantly.
You want to get a high chance at a top-12 season with someone such as Derrick Henry or Kenyan Drake before the tier drops off a cliff? Well, that's the risk you have to take. There's no sugar coating it, and the data just doesn't support punting running back entirely in your draft.
Naturally, applying this to a draft will require thoughts about all other positions, but to give you a call-to-action or a legitimate takeaway if you aren't going to build out your own projections: make tiers.
Tier out your running back list. See where the significant gaps are in projected points or your expectations for these players, and then plug them into this chart wherever they fall overall. Make sure that you don't undervalue touchdown potential (which is why someone such as Le'Veon Bell doesn't tempt me like he used to).
Additionally, the middle rounds of the draft -- based on the sims and history -- are a dead zone, and you should draft accordingly.
The search for the league-winning running back is never going to be solved because there's too much uncertainty about which players actually stay healthy. The search of where we can find them and just how likely it is that we can get a top-six performance is a bit clearer.