Finding 2020's League-Winning Fantasy Quarterbacks Through Season Simulations
Defining "upside" in fantasy football isn't easy, but we kind of know it when we see it.
A quarterback you snag in the 13th round who winds up being a weekly starter? That player showed upside. Your sixth-round pick ending up as the QB1? Yeah, I think that's upside.
But what if Lamar Jackson really did break fantasy football and repeats as the QB1 by a big margin? Even at his cost of 12th overall in NFC drafts and 25th in Bestball10 drafts, isn't that game-changing upside? For sure, it is.
That's why we get tempted by the occasional Jackson and Michael Vick pick in the first round. They can win us leagues. Heck, Patrick Mahomes is listed at picks 14 and 32, respectively, in NFC and BestBall10 drafts. That's still a steep price for a quarterback who is one season removed from a historic fantasy output in 2018.
They can pay off at those prices -- if they put up QB1-by-a-mile numbers. But just how likely is it that Jackson or Mahomes puts up that level of season?
Obviously, it's impossible to know with 100% certainty. We're getting one roll of the dice in 2020, and while I don't think it's right to say "anything can happen," a lot of things can happen. In this regard, median or "most likely" outcomes don't really matter over a one-season sample if our goal is to find true upside, but they actually can tell us a lot about what to expect.
So how can we begin to find the answer to which quarterbacks could give us league-winning upside in 2020?
Projections and Simulations
Honestly, when I build my fantasy football projections each year, I easily get bogged down in details. I can go back and forth over whether someone such as Ryan Tannehill will drop to a league-average yards per attempt rate or will instead stick up toward the top of the league after his 2019 surge. I will sweat expected touchdown rates for bottom-tier fantasy passers.
The longer I do that, the more I realize that doing something like that just doesn't account enough for the volatility in football.
Lamar Jackson threw a touchdown pass on 9.0% of his attempts last year, and you shouldn't project for that on a median basis. Cam Newton played two games last year. You don't project that as an expected outcome. But those things happen.
I just am starting to think that such minutia is less important than being flexible because we're not perfect at projecting. We never will be. We should get as close as we can, and we shouldn't project outlier seasons. From there, all we can do is hope we're close and that we don't get hit with a very unlikely outcome.
So, I have my projections for this year ready to go. The median projections. The "Here's What I Think Is Most Likely to Happen" type of projections.
They rely on numberFire's Net Expected Points (NEP) data, success rates, average target depth, and a few more advanced stats to project expected basic rate stats. For example, Passing NEP per drop back and Passing Success Rate tie strongly to yards per attempt, so I leverage those instead of copying and pasting last year's yards per attempt rates.
I'm constantly checking my projections against the consensus because A) I know enough to know that I don't know everything, and B) I like to see where I differ and then figure out why that is. My projections can be a tinge different because I ground them in an expected points model, which is stickier year-to-year than traditional stats, based on my research.
Okay, so enough already, right?
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.
It's important to keep in mind that these are really hard to falsify because even if we had fantasy football projections back to the start of the NFL and tested them against what actually happened, we'll have had only 100 seasons to deal with. We obviously have way fewer than that.
Simulation Results
I'm going to list out the relevant quarterbacks here in a table.
I have projected workload splits for three quarterback rooms right now (Nick Foles and Mitchell Trubisky, Ryan Fitzpatrick and Tua Tagovailoa, and Tyrod Taylor and Justin Herbert). They're not going to show up here. We're looking for upside. Potential timeshares don't really give us that at quarterback until one player takes over.
The results are sorted by the percentage of time a quarterback finished as a top-four performer. Why top-four? Well, having the QB1 is obvious upside, but with two star fantasy assets in the mix in Jackson and Mahomes, there aren't a whole lot of QB1 finishes to go around unless you really ramp up variance and randomness. That may be the correct play, but it's not how I'm starting this thing.
A top-four showing in a 12-team, single-quarterback league still indicates an upside performance, and we usually start to see things level off after the top four or so.
| Standard Variance | Average Finish |
QB1 Finish |
Top-4 Finish |
Worse Than QB12 |
|---|---|---|---|---|
| Patrick Mahomes | 5.5 | 26.0% | 61.5% | 12.6% |
| Lamar Jackson | 6.0 | 28.7% | 59.9% | 15.3% |
| Dak Prescott | 10.0 | 7.4% | 31.2% | 32.1% |
| Deshaun Watson | 10.3 | 6.6% | 28.3% | 33.9% |
| Kyler Murray | 11.2 | 6.3% | 26.8% | 38.8% |
| Russell Wilson | 10.9 | 5.9% | 26.1% | 37.2% |
| Matt Ryan | 11.3 | 4.0% | 22.5% | 39.1% |
| Drew Brees | 12.1 | 2.1% | 16.4% | 43.7% |
| Carson Wentz | 12.4 | 1.6% | 15.1% | 45.1% |
| Jared Goff | 12.5 | 1.7% | 14.3% | 45.9% |
| Aaron Rodgers | 12.4 | 1.6% | 14.2% | 45.5% |
| Josh Allen | 15.4 | 1.2% | 9.8% | 60.2% |
| Ryan Tannehill | 16.9 | 1.2% | 8.9% | 66.1% |
| Tom Brady | 14.3 | 0.9% | 8.7% | 56.4% |
| Matthew Stafford | 15.2 | 0.6% | 7.3% | 60.9% |
| Ben Roethlisberger | 15.3 | 0.6% | 7.3% | 61.9% |
| Daniel Jones | 16.9 | 0.9% | 7.2% | 67.0% |
| Jimmy Garoppolo | 16.1 | 0.4% | 5.7% | 66.3% |
| Baker Mayfield | 17.8 | 0.5% | 5.6% | 72.3% |
| Kirk Cousins | 16.3 | 0.4% | 4.6% | 67.3% |
| Philip Rivers | 16.8 | 0.4% | 4.2% | 70.4% |
| Gardner Minshew | 16.8 | 0.3% | 3.9% | 69.8% |
| Joe Burrow | 17.8 | 0.4% | 3.7% | 74.0% |
| Teddy Bridgewater | 18.8 | 0.1% | 2.2% | 78.9% |
| Drew Lock | 20.0 | 0.2% | 2.1% | 81.4% |
| Derek Carr | 19.9 | 0.1% | 1.4% | 83.9% |
| Jarrett Stidham | 21.5 | 0.0% | 0.7% | 89.2% |
| Sam Darnold | 22.5 | 0.0% | 0.4% | 92.2% |
| Dwayne Haskins | 22.6 | 0.0% | 0.2% | 92.7% |
You may think these numbers are way off. I don't care. Not that I don't want to be accurate. I do. I just don't care because in thinking through these results, you learn a lot about how you think about fantasy football, and that's just as valuable as these numbers -- in a way.
Do you think Lamar Jackson has only a 10% chance to repeat as the QB1? Do you think he gets hurt and misses multiple games 30% of the time? Do you think that someone like Jimmy Garoppolo is just as likely to get a top-four result as Dak Prescott? You probably view things as more random or volatile than I do and than a single standard deviation does.
All I know for sure is that these numbers are based in median projections that look pretty similar to consensus and that they are accounting for historical, backtested deviations from preseason projections in recent seasons.
What that means is this: a quarterback projected for X number of fantasy points usually falls within a certain range of that median projection -- even though there are huge outliers above and below that number. If you project Lamar Jackson for 350 fantasy points and Patrick Mahomes for 300 as your QB2, Jackson should be finishing at the top much more frequently than Mahomes.
Thought Experiments and Potential Ramblings
Okay, so my simulations have Lamar and Mahomes finishing as the QB1 a lot, and I'm fine with that assessment because I have them projected a tier higher than every other quarterback -- and I'm not alone in that. There will be dud years from these guys, but they shouldn't be super frequent if our sample is large enough, and if we're aiming to be as objective as possible, then we should be okay with that.
After all, an NFL season isn't a roll of the dice, and in no reality does Baker Mayfield finish as high as Lamar Jackson an equal number of times. Does that mean Mayfield will 100% not finish as 2020's QB1? No. He can. But with Mayfield, we're asking the QB15 or so to have an outlier season and outperform a dozen-plus quarterbacks in the process.
And that's actionable in its own right.
There's a cluster of really good quarterbacks after Jackson and Mahomes, and for Mayfield (or Daniel Jones, or Matthew Stafford, or Ryan Tannehill -- take your pick) to finish as a top-four passer, we need at least five of the top eight quarterbacks to deviate pretty significantly from their median, most likely outcomes.
Back to it: Dak Prescott, Deshaun Watson, Kyler Murray, Russell Wilson, and Matt Ryan all pop as having some pretty legitimate chances to finish as top-four fantasy passers this year. Coincidentally, those five are the other five in the top-seven in NFC average draft position so far this season. Does a top-four season from the QB7 count as "upside?" How many more times am I going to ask that question?
All of those players outside of Ryan, whom I'm high on due to elite passing efficiency data, have rushing potential, and rushing potential when combined with some strong passing numbers -- that's how you find a fantasy outlier.
Here's the thing, too: after the top few fantasy assets, you can replace these names however you want, really. Take a look at as many projections as you can and see how different things are between guys such as Stafford, Jones, Tannehill, and the rest. It's almost always 15 to 20 fantasy points over a full year.
You can still take your stands. I think that's another key takeaway here. Outlier performances happen, but just how easy are they to project is the bigger question.
I love Matthew Stafford, and I always will. He was the QB6 last year in points per game through Week 9. That was with a career-high 8.6 yards per attempt (besting his prior high of 7.9) and a career-high 6.5% touchdown rate (up from a career mark of 4.4% entering the season). We had an outlier start for Stafford last year, and he still wasn't a top-four quarterback. Take a step back and think about just how likely it is that Stafford is the QB1 with Jackson, Mahomes, and three to four other dual-threat options.
Does Stafford have "upside" at his QB14 cost? Yeah, he can outperform that easily. Is he actually likely to be the QB1 or a top-four performer? History says not exactly if we're projecting him in the QB10-14 range. He has to beat out 10-plus quarterbacks to do that.
A similar argument can be made for something like Gardner Minshew versus Drew Lock. Do the weapons and draft equity favor Lock? Yeah, sure. Does history suggest that Lock will outperform his preseason median projection by around 100 fantasy points to give us a top-four season? No, not at all. So this depends on what exactly you expect from a quarterback pick. There are realities in which Lock is a top-four performer, and if you draft him and 2020 turns out to be one of those realities, then congratulations. You beat the odds.
The most obvious, recent example of a quarterback shattering expectation really isn't even Lamar Jackson but rather Ryan Tannehill. Tannehill, once taking over as the Tennessee Titans' starter last year, ranked as the QB3 on a fantasy-point-per-game basis among qualified quarterbacks. His showing was a massive return on investment for fantasy players. It was also just nothing we could have predicted.
What's my point here? That a lot can happen over just one simulation real-life season and definitely not an iteration of the simulation that we're living in. But that's actually why the numbers in the chart shouldn't be taken for granted.
What matters is that we don't lie to ourselves and overestimate how likely it is that we can find a top-four, league-winning quarterback from nowhere. Plus, even when the 2020 version of Tannehill emerges, don't act like you'll be the only one trying to claim him.
So You're Asking for More Variance
If you think the numbers above are off base and a little too conservative, then that's fine. As much as I want there to be a right answer here, there never really can be.
However, we can run things with two standard deviations instead of one. Things get weirder, and here's how that looks.
| Added Variance | Average Finish |
QB1 Finish |
Top-4 Finish |
Worse Than QB12 |
|---|---|---|---|---|
| Patrick Mahomes | 9.5 | 15.7% | 40.3% | 29.9% |
| Lamar Jackson | 10.1 | 18.5% | 39.8% | 32.9% |
| Deshaun Watson | 12.8 | 6.7% | 24.1% | 45.5% |
| Dak Prescott | 12.7 | 6.8% | 23.8% | 44.3% |
| Kyler Murray | 13.3 | 7.0% | 23.4% | 46.6% |
| Russell Wilson | 13.0 | 6.1% | 22.7% | 46.7% |
| Matt Ryan | 13.1 | 5.2% | 20.7% | 47.1% |
| Drew Brees | 13.6 | 3.2% | 16.0% | 50.5% |
| Jared Goff | 13.7 | 2.6% | 14.9% | 51.2% |
| Carson Wentz | 13.9 | 2.8% | 14.9% | 51.8% |
| Aaron Rodgers | 13.9 | 2.8% | 14.4% | 51.9% |
| Josh Allen | 15.7 | 2.9% | 13.8% | 58.8% |
| Ryan Tannehill | 16.8 | 3.2% | 13.4% | 62.0% |
| Tom Brady | 14.8 | 1.8% | 12.0% | 57.0% |
| Daniel Jones | 16.7 | 2.3% | 11.8% | 62.9% |
| Matthew Stafford | 15.6 | 1.5% | 10.4% | 60.1% |
| Baker Mayfield | 17.3 | 2.1% | 9.9% | 65.9% |
| Ben Roethlisberger | 15.4 | 1.3% | 9.4% | 60.3% |
| Joe Burrow | 17.0 | 1.1% | 8.2% | 66.4% |
| Jimmy Garoppolo | 16.1 | 1.0% | 8.2% | 62.9% |
| Kirk Cousins | 16.0 | 1.0% | 8.1% | 63.5% |
| Philip Rivers | 16.3 | 0.9% | 7.1% | 64.7% |
| Gardner Minshew | 16.3 | 0.8% | 6.8% | 64.9% |
| Drew Lock | 18.4 | 0.8% | 6.5% | 71.6% |
| Teddy Bridgewater | 17.5 | 0.7% | 5.5% | 70.3% |
| Derek Carr | 18.1 | 0.5% | 4.5% | 73.2% |
| Jarrett Stidham | 19.3 | 0.4% | 3.8% | 77.3% |
| Sam Darnold | 19.8 | 0.4% | 3.1% | 79.3% |
| Dwayne Haskins | 20.0 | 0.2% | 2.8% | 81.0% |
Thanks, I hate it.
To me, this is a pretty good sign that we should hover around expectations quite a bit when projecting fantasy football quarterbacks because if you can look at me and say that Lamar Jackson and Mahomes finish outside the top 12 in fantasy scoring around 30% of the time, then I think we approach the game very differently.
What this higher-variance simulation shows us is that a lot of quarterbacks have top-four potential, and that's just not based on history or reality. This is much more of an outlier-friendly approach.
In this sense, you can make the stronger case for finding your cheaper quarterbacks in the draft. Maybe just wait for Daniel Jones instead of reaching for Tom Brady or Josh Allen. That kind of thing. Don't play it safe with Jimmy Garoppolo and instead shoot higher in case Ryan Tannehill really is a fantasy godsend all of a sudden.
Final Thoughts
Nothing presented here is going to help you nail the exact QB1 in 2020 -- because we don't know which version of the simulations are going to play out. Instead, this information should help you think about where QB1s historically come from from a projections standpoint.
Whether a 25% to 30% shot for the QB1 is enough for you to reach for Jackson or Mahomes -- that's on you. I don't think it is for me, given their asking price.
Instead, this opened my eyes to that second tier of quarterbacks. Someone such as Russell Wilson or Kyler has a pretty solid shot at a top-four season with QB1 upside, and there's value in knowing that. Typically, I let this tier pass me by. Usually, I wait for a Daniel Jones instead of snagging the final passer in that juicy second tier. I think the data suggests doing the opposite, at least based on how this 2020 crop of quarterbacks is shaping up.
The long shots, well, they can pan out. We know some will. Some have the variance and profile to do it more than others. But if you're looking for the passers who shatter their draft-day cost and have enough potential to be realistic top-four fantasy options, you may want to cling to the Murrays, Watsons, Prescotts, Ryans, and Wilsons instead of taking a stab on a true late-round flier.
As someone who always waits on quarterback as long as possible, I think I've changed my tune a bit if I'm more aggressively seeking league-winning upside without breaking the bank.
