NFL DFS Strategy: Optimal Roster Construction for Tournaments
Being a successful NFL DFS player requires so much more than understanding the “best plays.” If it was that easy, everyone would be winning a tournament every week. Of course, that doesn’t happen, so how can we increase our chances of winning in GPPs? By utilizing optimal roster construction and understanding how to use roster percentages to gain leverage on the field. In this episode of the DFS Podcast, Kyle and I talked about these concepts in more detail. If you’re looking for more information on winning in GPPs, be sure to check it out!
Before we get into our favorite roster construction points, it’s crucial to understand that these concepts primarily apply only to tournaments or GPPs. If you’re playing in cash games like head-to-heads or double-ups, we’re trying to build the best lineup using median projections to give us a solid floor. This checklist is geared toward hitting a ceiling performance in order to achieve a high-end outcome to finish at the top of the leaderboard in tournaments. Most contests on DraftKings and FanDuel are top-heavy in terms of their payout structure. As a result, we should be chasing ceiling in every tournament lineup we build.
We looked at every winning DraftKings Milly Maker for Sunday’s Main Slate over the last two years to find the strongest trends:
- 89% of lineups had a Team Stack
- 83% of lineups had a Game Stack
- 89% of lineups stacked their QB with at least one pass catcher
- All others listed below occurred at a 60% rate or less
1. Team Stack
GPP lineups should include at least 2 players from the same team. The most simple way to think about this is a quote from our guy Kyle Borg – “If the offense does work, we all get to eat.” Think about it this way – if a team puts up 40 real-life points, there’s a very strong chance multiple players from the team have a great game.
2. Game Stack
Every week in the DFS Pass, I write up the Vegas Report to look at lines and totals to find the best game stacking environments. Kyle and I also discuss our favorite games for DFS scoring on our Tuesday episode every week during the regular season. In general, we want high total games where the two offenses playing in those games can interact and push each other to put up more points. This is all about correlation and minimizing the number of things we need to get right in our lineups. If we can correctly identify the top scoring environments and avoid the ones that bust, we’re likely to land on some high scores from the skill players in those lineups.
3. QB/Pass Catcher Stack
We’ve talked extensively about the benefits of stacking in best ball formats, and similar concepts apply here. If a QB throws for 300 yards and 3 TDs in a week, he’s obviously going to bring his pass catchers with him. By not stacking, we’re sacrificing the upside we need to win in DFS. Remember, 90% of winning lineups stack their QB. The data is conclusive that it works. Per FantasyLabs data going back to 2014, the following players have the best correlation with their starting QB:
- WR1 = 0.57
- WR2 = 0.51
- TE1 = 0.51
- RB1 = 0.43
- RB2 = 0.31
4. QB Stack + Bring Back
This point on our checklist is strongly related to #2 above, the game stack. More specifically, the data shows that the QB on one team is strongly correlated to a better performance not only from his own teammates (#3 above), but also his opponents. Again, I used FantasyLabs data going back to 2014 to see which player on the opposition correlates best:
- WR2 = 0.40
- WR1 = 0.39
- TE1 = 0.39
- RB1 = 0.38
- RB2 = 0.29
The take home from this data is that there isn’t a strong predictor of success across many of these players on the opposing team. In other words, the opposing WR1, WR2, TE1 and RB1 are all basically the same in terms of correlation. Basically, as long as you’re bringing it back with someone on the opposing team, you’re building correctly.
If this is new to you or you haven’t played in DFS tournaments before, let’s use an example to illustrate this point. In Week 1, the Detroit Lions play the Philadelphia Eagles. To implement a QB stack + a bring-back, one option would be to play Jalen Hurts stacked with DeVonta Smith, brought back with Amon-Ra St. Brown on the other side.
5. Max 1 RB Per Game*
There’s an asterisk here because this applies primarily to running backs who aren’t pass-catchers. In other words, it’s okay in certain game environments with very specific RB archetypes to play multiple RBs in the same game. While this is never a go-to move for us, it can work in the right scenario. Which archetypes fit this best? RBs who can see a high target share. However, the vast majority of the time, we want to limit our lineups to include only one RB from a game. Again, this point is most easily communicated with an example.
While Derrick Henry and Jonathan Taylor are capable of catching the football, generally we don’t think of them having their best games through the air. We think of Derrick Henry stiff-arming defenders into submission on his way to 150 rushing yards and 2 TD. Similarly, we think of Jonathan Taylor ripping off a 50-yard run for a score in his best games. This archetype of running back typically slows the game down, making it difficult for the opposing back to hit a ceiling performance.
Per FantasyLabs data, the RB1 and opposing RB1 only have a correlation of 0.29. When we control for the total of the game per Vegas odds to zoom in on the games we actually care about for DFS game stacks that are projected to be high scoring, the correlation is even weaker for RB-RB. When the game total was at least 48, the correlation was reduced to 0.25, and when the game total sample included only games that were 50+, the correlation was reduced to 0.23.
The take home? The higher the game total, the more we want to avoid two RBs from the same game.
6. Low Rostered Player(s)
This part of the checklist is absolutely crucial to NFL DFS success in tournaments. If we only include the “chalky” plays each week, we’re not gaining any leverage on the field and not moving up the leaderboard by separating ourselves. Let’s use data to see how exactly to go about doing this.
This largely depends on the type of contest you’re entering – a small field tournament or a large field tournament. For the data presented below, I looked at 2021 data for three different contests on DraftKings – the Milly Maker for large field data (20K+ entrants), and the Luxury Box (370 entrants) and 4th Down Conversion (250 entrants) for small field data.
Small Field Strategy
- On average, the winning lineup had 2.93 players per lineup that were <10% rostered
- Every lineup except one had at least 2 players that were 20+ % rostered
Large Field Strategy
- On average, the winning lineup had 5.1 players that were <10% rostered
- On average, the winning lineup had 1.31 players that were >20% rostered
- Every winning lineup except two had a player that was <3% rostered
The data is conclusive that regardless of which format you’re playing (whether it be a large field or small field), you should be trying to get unique with some low-rostered players. How many of those players depends on the size of the field. A general take home – the larger the tournament, the more willing we should be to target low-rostered players.
Another data point to drive home here is that there’s this idea from a lot of DFS players that you can’t play popular plays aka the “best plays” in tournaments. While we want to be cognizant of not going overboard here, the data shows that regardless of small vs. large field tourneys, winning lineups do roster popular guys most of the time.
7. Roster a WR in the Flex
Last August, I wrote a detailed article about How to Attack the WR Position in DFS. Rather than reinvent the wheel, I would recommend reading that article to get a thorough understanding of how valuable WRs are in the flex. Basically, we want the volatility that comes with WRs because that comes with massive spiked weeks. Additionally, if you’re playing on DraftKings where it’s full PPR scoring with the 100-yard bonus, WRs are generally more likely to hit those boom weeks to give us DFS winning performances. Does the data support this? You bet.
The first image is from 2020 Milly Maker data. These trends and images are from our friends at FantasyLabs. As you’ll see, the our opponents are playing RB in the flex more often than WR (blue ‘All Lineups’ bar). However, wide receivers tend to show up in the top 1% of lineups far more often than running backs (green bar).
What if we look at a different contest from 2021? The graph below is from the $500K Play-Action 20-max contest. Again, we see similar ‘mistakes’ from the field. The field doesn’t value the spike weeks WRs give us nearly enough. Wideouts are still showing up in the top 1% of lineups at a much higher rate than the field is willing to roster them.
Obviously, this concept is slate-dependent. Certainly, running backs are going to show up as the optimal flex play some weeks, but if we look at what is most likely to happen and what gives us an edge against our opponents, being willing to play a WR in your flex can give you a higher expected win rate and higher payouts in tournaments.
8. No Players vs. DST
The data is consistent that we should absolutely avoid rostering players against our own DST. Intuitively, this makes sense when you think about how defenses score fantasy points (sacks, fumbles, interceptions, etc.) If a defense is stifling an opposing offense, the players on the other side of the ball are unlikely to hit a ceiling outcome.
FantasyLabs data supports this, regardless of position. All positions (QB, RB1, RB2, WR1, WR2, TE) are negatively correlated with the opposing DST.
If you found this article helpful, we think you’ll love the DFS Pass, which includes in-season content like our top DraftKings and FanDuel picks as well as our Roster Percentage Report and so much more. Get it today as part of the 2022 Ultimate Draft Kit+.
Thank you for this, I’ve been doing this wrong