How Well Did Heavily Rostered Players Perform for Daily Fantasy Football in 2021?

When you're staring down a popular play in daily fantasy football, two questions are sure to run through your mind:

1. Will this player flop?

2. Will this player make me regret not using him?

These questions are simply proxies for considering the player's range of outcomes. And they're good questions to ask as avoiding a popular play that flops gives you a leg up on the field.

I think we focus too much on that first question and not enough on the second.

In daily fantasy tournaments, floor gets more helium than it deserves. You're not trying to identify players who won't flop; you want players who will blow up and give you easier access to those big payouts.

When I've written about roster rates in the past, too much of the focus was on identifying whether a player would fall short when they were popular. That's well and good because there are upsides to pivoting from bad chalk. But if it takes focus away from investigating how good we are at pinpointing ceiling performances, then it's a relative waste of our time.

As such, today we're going to dig into how often popular plays in NFL DFS blew up. If a running back is more likely to go nuts when popular than the baseline, it increases our incentive to ride with the public; if not, we can fade them without too much fear.

Part of this will reference back to our offseason look at which kinds of players tend to wind up in perfect FanDuel lineups. Those are the guys who demonstrated their upside and helped DFS players take down tournaments. If certain conditions are more conducive to an upside performance than others, and the public isn't going hard at players in those conditions, we've found a weakness we can attack.

Let's lay out the criteria here before digging into what the data says.

Defining Popularity

There's never going to be a "right" answer on what constitutes a player's being "popular." We've just gotta set parameters and go from there.

For all of the single-player positions (quarterback, tight end, and defense and special teams), we'll look at the three players/teams on the most rosters in each week's FanDuel Sunday Million. That'll give us 54 popular players for each position across 18 weeks.

For running back and wide receiver, we'll go with the top seven. You're rostering at least two backs and three receivers, and you've got the flex spot to plug in more. The average roster rate for these running backs was 20.7% versus 20.0% for receivers, which is why the numbers are the same even though the average number of receivers rostered is higher (3.4 per lineup versus 2.5 for running backs).

So, whenever I refer to a player as "popular" or "heavily rostered," it means he was a top-three quarterback, tight end, or defense or a top-seven running back or wide receiver in roster rate for that week's FanDuel Sunday Million.

Finding the Baseline

If we want to compare popular players to the baselines at their position, we need a sub-set with which to compare them. And it's not apples-to-apples to compare them with all players at one position because you then get players with no roles or who are injured.

As such, the baseline will be based on all players who were on at least 1% of rosters in each week's FanDuel Sunday Million. That way, we're still getting a large sample of the less popular options without dabbling in unusable players.

The baselines will revolve around each player's "value," or how many FanDuel points they provided per $1,000 of salary. This is a hyper-flawed method for analyzing as it views 10 points from a player with a salary of $5,000 the same as 20 points from a player with a salary of $10,000. This system overvalues low-salaried players and undervalues the studs. But it's the best we've got to work with for now.

How Often the Chalk Hits

I promise to be brief with this, given we just dumped on the value of floor analysis earlier.

I defined a "hit" to be when a player exceeded the average value of players at his position among those on at least 1% of rosters in the FanDuel Sunday Million. It's a very low bar to set, to be sure. At running back, it was just 1.81-times value, which is about 14.5 points out of a back with a salary of $8,000. It's not doing a ton for you to hit this baseline. But it shows that coming up short of it is a massive detriment to your hopes.

The chart below shows how often popular players at each position hit this (very reachable) baseline. I promise you there are no typos.

It's a weird, dumb coincidence that I double-checked to make sure was accurate. Tight ends just gotta be different.

The translation of the chart is that 55.6% of popular running backs returned more than 1.81-times value this past year. That's not a bad number as it means they were better than the total population at the position.

It also means, though, that 44.4% of popular backs failed to hit even a low baseline, which should be troublesome. And these rates weren't even big deviations from past years. Here's the numbers from the same process across the past six years.

Last year, the chalk hit at an abnormally high rate. This year, things were closer to normal. So although NFL DFS is tough, our competition does still make mistakes.

One of those mistakes is point-chasing. It's still absolutely rampant.

Point-Chasing

As we've seen in past years, players typically score fewer points the week they're popular than they did the week before. We get excited about the box score and unnecessarily roster middling plays.

That didn't change this year.

The table below shows how many points players scored when they were popular, how many points they scored the game before (excluding Week 1 and players coming off of injury), and how often players scored more points when they were popular than they did the previous game. There were some massive falloffs here.

Popular quarterbacks saw their scoring decrease by 23.5% from what they did the previous week. Defense was the only position to score more points on average when popular.

We're going to talk throughout this piece about situations in which the public over-invests in players. There may not be a better example than when a player is coming off a huge game. If a big game is going to lead to increased roster rates (not to mention an increased salary), we've got incentive to be skeptical.

Identifying Upside

Now we get into the fun stuff. We're hunting for upside, baby.

Specifically, I want to see how often popular players hit various value thresholds versus the baseline at their position. If we typically nail down all the four-times value performances out of quarterbacks, then our willingness to ride with the public should increase.

That's not the case, at least at that position, though.

This table outlines how often popular players at each position hit various value thresholds and compares that with all players on 1% of rosters at that position. If the popular players are hitting high-upside thresholds at a much higher clip than the field, it means we're good at identifying upside at the position. If not, then the odds we get scorched for ignoring a popular play are lower.

In general, we were good at identifying high-floor plays. That's why the rate of hitting two-times value was higher among popular plays at each position than the baseline.

With upside, we sometimes struggled.

The big one was quarterback. Of the 17 quarterbacks on 1% of rosters who scored four-times value, only two were among the most popular options at the position for that week. Those were both by Kyler Murray in Weeks 1 and 2. Every other popular quarterback was at 3.93-times value or lower.

Quarterback is the position where that dynamic was most drastic. But each position had at least an element of "great at identifying floor, less stellar at predicting upside." It means -- in general -- we shouldn't be all that worried about missing out on an upside game from a popular play. Those options are just a little bit more likely to get there than the rest of the relevant field while the odds they fully disappoint are still decently high.

The two positions where we did the best job of identifying upside were likely running back and tight end.

At running back, 21.4% of popular backs hit three-times values versus just 13.9% of all relevant backs. The gap tightened at four-times and five-times value, but you would actually regret missing out on a 21-point game from a popular back whose salary was $7,000. Our process at running back is decently refined, so it's fine to have a higher FOMO factor there.

At tight end, only 18 relevant players hit three-times value for the entire season; five of them did so at high popularity. And it wasn't just the "use Mark Andrews" doctrine, either. Of the five popular tight ends who blew up, only George Kittle did so twice. We had a decent eye for upside at that position. The odds they flopped were high, but they did come with some ceiling juice.

Overall, the takeaway from the chart should be that the public is very good at identifying players with a high floor. So our first question -- will this player flop? -- is likely "no," relative to other players at the position. But with our second question -- will they burn me for not using them? -- it seems the answer also skews toward "no" outside of running back (and, to a lesser extent, tight end).

Quarterback

We've already seen we don't need to sweat the ceiling on popular quarterbacks all that much. The less popular players are just as likely to pop off as the popular ones. Are there conditions in which we actually should swallow the chalk or -- on the flip side -- be even more inclined to deviate?

We can get a read on this by comparing the popular quarterbacks to the ones in perfect lineups. We know where the public is going. We just have to see if that aligns with the conditions most likely to breed upside.

It doesn't.

The big one pertains to the spread of the game. As you'll recall from our perfect lineups discussion, a ton of quarterbacks this year blew up when their teams entered as slight underdogs. It was a competitive game where they had a bit extra incentive to throw, and it helped lead to explosions.

That didn't lead to players in those conditions being heavily rostered.

Here's a look at how often popular quarterbacks fell into each bucket. It's then compared to what percentage of quarterbacks in perfect lineups came from the same buckets. The "all teams" column shows what percentage of all teams fell into each bucket across the span of the 2022 season in order to give context behind each number.

We'll look at several of these charts today; this is the spot where the public deviated most from what was optimal.

Only five times all year was a quarterback popular on a team that was a slight underdog. There were actually six such quarterbacks in perfect lineups, even though there were just 18 total perfect quarterbacks versus 54 popular ones. Conversely, there were two heavily favored quarterbacks in perfect lineups versus 19 popular ones.

Quarterbacks on teams that are heavily favored have good floors. Thirteen of the 19 heavily favored popular quarterbacks hit the positional baseline for value, the highest rate of any bucket. But they hit three-times value just 31.6% of the time, lower than the 50% rate for underdogs and 34.3% rate for all quarterbacks in games with single-digit spreads.

This gives us a pretty clear edict for handling popular quarterbacks. If a team is in a game with a tight spread, and their quarterback projects to be popular, the FOMO factor should be pretty high. Those circumstances led to upside games often, and we know the public doesn't invest there as much as they should.

If a quarterback's team is heavily favored, they're unlikely to flop. But their ceiling is over-sold by their roster rate, which is the precise condition we should look for when deciding whether to pivot.

Outside of tight games, the other situation in which quarterbacks actually did show good upside was when they were in a game with a high total. They didn't hit baseline at a higher rate than the lower-total quarterbacks, but they did hit three-times value far more often.

Thus, if we were trying to craft a quarterback worthy of rostering at high popularity, we'd want them to be in a high-scoring game with a tight spread. Those to avoid would be the ones more heavily favored in a game with a lower total.

This year, 10 quarterbacks were popular while in games with a total of 50 or higher and a spread of less than five points. Two of them hit four-times value, a rate of 20%. The baseline for popular quarterbacks hitting that mark was 3.7%, and it was 6.2% for all quarterbacks on 1% of rosters. They still flopped at times, but that's the one scenario where we may actually want to roster the quarterback in hopes of capturing that upside.

On the flip side, 17 quarterbacks were heavily favored in a game with a total lower than 50. Only 3 (17.6%) hit even 3-times value, and none of them topped 3.35-times value. A quarterback in that setup is a pretty easy fade despite their high floor.

This may seem like we're setting narrow parameters for swallowing chalk at quarterback. But with how poor we are, broadly, at identifying upside at the position, that's the mindset we should have. We should go in assuming we'll want to pivot until we are forced to reconsider due to conditions conducive to upside.

Running Backs

Our mindset at running back should be different with much less aversion to the popular options. We were better at finding ceiling there, so there's higher odds we miss out on a needle-moving performance.

That was especially true with the low-salaried options.

Of the 126 popular running backs this year, 17 had a salary lower than $6,000. The rate they hit baseline value -- 70.6% -- is impressive but shouldn't matter much with how low the threshold is for them.

But the rate at which they hit three-times value was massive.

As a reminder, only 13.9% of all backs on at least 1% of rosters hit 3-times value last year. It was 21.4% for all popular backs but a whopping 47.1% for the value options.

This isn't merely a product of their low salaries (and, thus, low value thresholds), either. Of those 17, 8 scored 18-plus FanDuel points, and 3 more scored 14-plus. They were getting you raw points for a bargain.

In most instances, this was because of an injury to someone else on the depth chart, pushing those players into bigger roles. So we're not just targeting any low-salaried back. But if someone is gaining additional volume at a low salary, we should absolutely ride with them whether they're going to be popular or not.

As far as when to fade a popular back, the 2020 data pointed heavily toward doing so when the back was on a team that was heavily favored. This past year hints at that but is far less vehement.

Here's the same table as with the quarterbacks, comparing the spread buckets for popular backs versus those in perfect lineups. There's still a split, but it's not nearly as pronounced as it was here last year or as it was at quarterback.

That split likely isn't big enough to justify jumping ship on a popular back just because his team is heavily favored.

That changes a bit if the player carries a high salary. There were 27 popular backs on teams favored by double digits. About half -- 15 -- had salaries of $7,000 or higher. Those players were far less likely to return worthwhile value than their lower-salaried counterparts.

The lower-salaried backs on heavily favored teams had both a great floor and an elite ceiling. The higher-salaried options hit three-times value at about the same rate as all backs on 1% of rosters (13.9%).

Thus, if a higher-salaried back projects to be popular in a non-competitive game script, we have reason to deviate. Their hit rate is good, but they don't torch you for ignoring them often. However, if it's a value back in a situation like that, our incentive to pivot is almost non-existent.

Overall, popular backs are a better investment than the popular options at other positions. We do a good job of identifying ceiling here, especially with lower-salaried options. There are some spots we can safely ignore the chalk, but in general, roster rate should be lower on our list of concerns at this position.

Wide Receiver

Quarterback and wide receiver will always be somewhat correlated positions for obvious reasons. It shouldn't be a surprise, then, that the aversion to underdogs carries over here.

We were more willing to push underdog receivers up the popularity ladder than we were with quarterbacks. But there was still a chasm between the popular options and those in perfect lineups.

Because the rate of receivers in perfect lineups being underdogs is lower than the overall rate, we shouldn't actively seek out receivers in a projected negative script. But if the environment and volume are good enough, we should be very open to guys in that bucket.

Although we didn't chase underdogs often, they hit at a decently high rate when we did. This was true both for their floor and their ceiling.

All relevant receivers hit three-times value 12.1% of the time, and it was 17.5% among the popular options. But when those popular options were underdogs, they hit three-times value at a 23.3% clip.

The other thing we noted at quarterback is that ceilings increased when the total was higher. This was true at receiver, too, with the high-total options giving us a massive ceiling hit rate.

The receivers in games with low totals had both bad floors and bad ceilings. If someone projects to be popular in a low-scoring game, it's an easy pivot. When the total is high, the equation is much, much different.

This gives us two key conditions where our FOMO should increase at receiver: when they're underdogs and when the total is high. When they checked both boxes, the three-times value rate increased to 35.7% (5 of 14). There, our incentive to deviate is low.

The one situation that could get us to jump ship is if the wind is high.

This tends to be a touchy subject because there isn't a ton of research behind it. There's a perception out there that if wind speeds are elevated, everyone will swap off of those players, allowing us to scoop them at lower roster rates.

The data disputes this.

For the full season, 27.1% of all games had double-digit wind speeds. But 35.7% of popular receivers came from that sub-set of games.

We rostered receivers in high-wind games at a higher rate than the baseline, and we under-rostered receivers in low- or no-wind games. That's despite the fact that receivers in perfect lineups were aligned the way you'd think with the wind.

Not shockingly, the receivers in heavy wind came with lower floors and ceilings, as well. The three-times hit rate for popular receivers in wind was higher than the positional baseline, but it was much lower than the popular options in lower winds.

Next time someone tells you we over-value wind in DFS, feel free to ignore them and trust the process.

This gives us two key spots where we have big incentive to pivot from the chalk: when winds are high and when totals are low. Those two will often go hand-in-hand as totals decrease in higher winds, and it's for good reason.

There were 13 times this year where a popular receiver fell in both bad-chalk buckets with high winds and a low total. Those 13 players hit baseline value just 38.5% of the time and never exceeded 2.42-times value or 18.6 FanDuel points. If a receiver projects to be popular in these conditions, you can and should fade them with confidence.

The flip side is the 25 popular receivers in games with totals of 50 or higher and single-digit wind speeds. Those 25 hit baseline value 64.0% of the time and 3-times value at a 32.0% clip. That's a solid floor-ceiling combo that's worthy of rostering in tournaments.

Receiver seems to be the position with the most sturdy situations to pivot or swallow the chalk. If winds are high or the total is low, look elsewhere. If winds are low with a high total, we can feel comfortable following the crowd. And if the public is overlooking a receiver who checks key boxes while being a slight underdog, we might just find ourselves an under-rostered stud with an easy path to a ceiling.

Tight End

Take everything we just said about totals and spreads for receivers and paste it into your process at tight end.

This year, there were more underdog tight ends in perfect lineups than favored tight ends. We still didn't tend to roster them at a high rate, though.

If you have one takeaway from this, let it be that we should be more receptive to players on teams that are slight underdogs. As long as you can reasonably project them to be competitive and score points, they're good investments and equally likely to give you a good ceiling as those on teams with big spreads in their favor.

The best floor-ceiling combination at tight end was with those in games with high totals. Here, we'll look at how often they hit 2.5-times value as just 6.3% of relevant tight ends hit 3-times value this year. For reference, the 2.5-times value rate for all relevant tight ends was 11.2% this year.

There were two other tight ends in the high-total group who hit 2.43-times value, just missing the threshold. You can easily justify chalk tight ends if they're in a game projected to feature points.

The FOMO decreased significantly when a team was in a lower-scoring game with the spread more heavily in their favor. Eleven popular tight ends were on teams favored by five-plus points in a game with a total lower than 46. None of them hit even 2.35-times value, and only 5 hit the positional baseline value. A low hit rate with low ceiling odds gives us full freedom to fade.

Taken all together, it's clear we want pieces of passing games tied to high totals and low wind speeds, and we should be more receptive to players on teams that are slight underdogs. We saw the same threads pop up at quarterback, wide receiver, and tight end, giving us a solid blueprint to know when the chalk is good and when it should be avoided.

Defense and Special Teams

As you saw in the beginning, our process at defense and special teams is better than most positions. We don't tend to point chase, which means we're judging each team based on their merits for that given week.

Thus, we don't need to enter assuming we have to deviate from the consensus. There are some spots, though, where we could use some refinement.

Specifically, it ties to totals. This is what we focused on most in the perfect-lineup piece, so if you read that, you won't be surprised that we under-roster defenses in games with high totals.

Here's the thought process behind being wary of defenses in games with low totals. Low totals can be low due to bad offenses. But they can also be low due to slower-paced offenses or ones that run the ball a lot. That leads to fewer opportunities for your defense to accumulate sacks, interceptions, and fumbles, the key stats that lead to points but can also lead to upside via return touchdowns.

Once you think about it that way, it makes sense that defenses in games with lower totals would have both a lower hit rate and a lower ceiling rate, which is what you can see in the data.

There were just 16 popular defenses in games with a total of 46 or higher, but they were elite investments for DFS. The 11 in games with totals lower than 42 were far more disappointing.

Similar to quarterback, we also saw a big over-investment in heavily favored defenses. Nearly half of all popular defenses were double-digit favorites compared to just 16.7% of defenses in perfect lineups.

The heavily favored teams did hit their baseline and three-times value at a high rate, so they weren't bad targets by any means. It's important to note, though, that we can find upside elsewhere, too, and we are likely to get it at a lower roster rate.

The final note at defense pertains not to how popular defenses performed but which defenses wound up being popular in the first place. Namely, not many low-salaried options wound up being chalky.

Of the 54 popular defenses from this past year, only 13 came with salaries lower than $4,000. That's less than one per week on average. In general, the public was spending up here with an average salary of $4,380 for popular defenses.

Those lower-salaried defenses didn't have a great ceiling rate, so this isn't a situation where we should just follow the public whenever they flock to a lower-salaried option. It does, though, tell us that if we like a lower-salaried defenses, the odds we'll have to pay a roster-rate tax on them are fairly low.

The biggest takeaway at defense is the one we started with: we should be skeptical of popular options in games with low totals. If the total is low due to a lack of play volume and drop backs, it'll hurt both the floor and ceiling of the defenses involved. That gives us big incentive to look elsewhere. Combine that with an openness to rostering defenses that aren't heavily favored, and we should have a fairly sound process at the position going forward.

Overall Takeaways

Let's circle back to our two key questions about popular options in DFS:

1. Will this player flop?

2. Will this player make me regret not using him?

We've found some key and definitive answers to both questions which should guide our process going forward.

For pieces of a team's passing offense, the odds we get burned for ignoring a popular option increase as totals go up. Those players posted ceiling performances at a much higher rate than the baseline at their positions, so we should be receptive to the chalk there.

The flip side is being very willing to deviate as totals went down and wind speeds went up. They had both bad floors and bad ceilings there, a perfect recipe for a a fade.

At running back, we have almost no incentive to pivot away from lower-salaried options moving into bigger roles. They gave us not just good value but also impressive raw outputs. In those scenarios, the chalk was the chalk for a reason, and we should look for under-the-radar options elsewhere in our lineup.

At all positions, we should have a higher willingness to roster players on teams that aren't heavily favored. Although those bigger favorites come with great floors, their ceiling rate isn't all that high, sometimes being even lower than what you get out of players in more competitive scripts. We don't need to actively seek out these players, but if they shape up as good plays and aren't heavily favored, they're likely to be under-rostered.

In general, we're not great at identifying upside in daily fantasy, We overlook options that blow up regularly. That means we should be willing to deviate from the public fairly often. Using the key exceptions above -- while also actively pivoting in the obvious let-down spots -- should put us in a good position to benefit from bad chalk without passing up key ceiling performances.