Hey everyone! With NFL preseason ongoing and regular season just around the corner, I wanted to write a comprehensive guide to NFL Daily Fantasy Sports covering topics and tips that I’ve picked up since 2009 when I jumped into the scene 14 years ago. If you are completely new to DFS, have dabbled and want to increase your play this season, or are a full-on DFS grinder, hopefully this has a little bit for everyone.
This guide will only go over GPP (tournament) play, the contests that have top-heavy payouts. This is also mainly geared towards multi-game slates and not single-game showdown slates, although I’m thinking about writing up a showdown guide soon. Cash games (50/50s, double ups, head-to-heads), on the other hand, do not employ any of this strategy and need a totally different mindset when approaching them.
If you are reading through the writeup and come across a term that is foreign to you, be sure to reference the Term Definitions that I have listed below and it will probably be defined there.
DFS: Daily Fantasy Sports
GPP contests: “Guaranteed prize pool”. This is the term used for tournament-style contests on DFS sites. Top-heavy payouts and a guaranteed prize pool even if only 1 person joins the tournament.
Cash contests: Contests where the top 40-50% of finishers approximately double their buy-in. These contests are usually called double-ups, 50/50s, and head-to-heads, although there are some other variants such as triple-ups where the top third of participants approximately triple their money.
Min cash: Finishing in the lowest payout bracket in a tournament, typically 1.5x-2x the buy-in.
pOwn: Projected ownership of a player across all lineups within a contest.
Rake (Margin): What percent of the buy-ins the DFS website is taking for itself and not paying back to players.
Overlay: When a contest does not fill and there is either reduced rake, no rake, or even more prize money than money entered by buy-ins.
Fade: To do less of or even fully avoid.
Chalk: A player that is carrying a considerable amount of projected ownership relative to players priced around him.
EV: Expected Value, or the value you expect to take out of a situation if you were able to partake in this identical situation many times over. +EV means you will take out more value than you put in, while -EV means you will take out less value than you put in.
SE: Single entry, or a contest where you are only allowed to enter one lineup.
3-max: A contest where you are allowed to enter 3 lineups.
20-max: A contest where you are allowed to enter 20 lineups.
150-max: A contest where you are allowed to enter 150 lineups.
Field: The other lineups in your contest that are not yours.
Steamed: Generally used in terms of ownership, when a player’s ownership gets artificially inflated past where they should logically be due to word of mouth and sheep mentality.
Tout: A person paid to do internet content that names specific players that people should be putting in their lineups.
Slate: A playable set of games on a DFS website. For NFL, Main (Sunday noon and afternoon games), Early (Sunday only noon games), Afternoon (Sunday only afternoon games), etc.
Dupe (Duplicate): When your lineup is identical to one or more lineups in a contest, it is considered duped.
Showdown: A contest that consists of one sporting event rather than a series of games. In these contests, duplicate lineups are very common as player choices are very minimal.
Since this write-up will only cover GPPs, I won’t hit on cash game strategy. All I will say on the topic of cash games is you need to truly identify an edge you have over the field to grind a profit. Over the years, cash games have gotten so sharp as more and more people have access to pretty accurate point projections and ownership projections. To accomplish any profit at all in cash games, a player has to win over 56% of the time. In the current cash game atmosphere that is very hard to do, and pulling any meaningful ROI over that is truly a feat and takes a lot of experience, research, and intuition on the field. The best cash game players in the world are grinding out 4-6% ROI long-term, and to make that meaningful they are having to play a ton of volume. If you do decide to venture into the cash game streets, I recommend playing the largest contests that are the lowest stakes possible. One last tip for cash games: if you are posting head-to-heads for other people to enter, be sure to cancel the ones that are still open before the slate locks or you will be auto-paired with the other players that have open contests. And guess what! Those are generally the highly skilled players that everyone was avoiding.
Now onto the topic of the day: GPPs. Obviously, to be profitable at GPPs, you need to identify that you have an edge over the field (or at least a majority of the field). In my opinion, there are more avenues to an edge in GPPs than in cash games. When it comes to identifying good contests to join, it is best to look at the buy-in and payout structure. High-volume players are locked out of contests with less than $3 buy-in (less than $5 if the payout is less than $25,000), so if you are able to join those contests then I highly recommend hammering them and solely focusing on them as you can get a lot of volume down on them weekly without having to stray too far out of that buy-in range. Contests vary widely with respect to how many lineups you are allowed to enter into them. There are contests ranging from single entry, 3-max, 20-max, 150-max, and the occasional contest somewhere between those. I recommend not focusing on just one type, but playing them all. Unlike other sports that play nightly, there are less than 20 weeks of playable NFL DFS, so to realize your true edge you will need to play as many slates as possible. Don’t just limit yourself to the main slate every week, also play the early slate, afternoon slate, thurs-mon slate, etc. As for payout structure, I look at how the top 10 finishers are paid, the rake (margin) of the contest, and how much of the field is paid.
Percent to first place: Generally I like when contests have 10-20% of the prize pool going to first place. If a contest reaches 30%+ to first place, I tend to avoid that contest as that tends to really push players to hit first more often over the long run to realize a decent ROI.
Percent to second place: I like to see second place receive around half of first place’s prize. Sometimes you will see second place take home 10-20% of first place’s prize (this generally tends to happen when first place takes home 30%+ of the prize pool), and I tend to avoid those contests.
Percent to tenth place: One-tenth of first place’s prize is where I like to see tenth place’s payout to be.
Rake (Margin): Draftkings takes home around 15-16% of the prize pool for an ordinary GPP contest. Obviously, players would love for this to be smaller, but 15-16% is where we have kind of accepted the industry standard. Anything higher and I tend to fade the contest. On the other hand, a GPP contest can sometimes not fill which leads to reduced rake, no rake, or even more money in the prize pool than money entered by buy-ins. In this situation, it is highly recommended to partake in these contests as this situation greatly increases your EV.
Percent of field paid: Draftkings tends to pay 20-30% of the field for cookie-cutter GPPs. Anything outside of this likely happens for odd one-off type contests.
Min cash: This refers to how much the lowest winning person will take home. Draftkings generally keeps this between 1.5x and 2x buy-in.
To see these stats overlaid on the Draftkings contest page at a glance as seen in the screenshots below, you can download the Rotogrinders Draftkings Chrome Extension here: https://chrome.google.com/webstore/detail/rotogrinders-draftkings-t/lokmacldfjfgajcebibmmfohacnikhhd
One of the biggest edges you can find in DFS is predicting what the field (other players submitting lineups) is going to do and proactively adjusting to your benefit. A majority of the field is taking some projection source, inputting them into an optimizer, clicking the ‘Create lineups’ button, then submitting those lineups into the contests. These are the players that we can easily bank on beating as they are just throwing fodder in there that generally matches everyone else in the contest.
There are a ton of ownership projection sources that can give you a rough estimate of how much each NFL player will be used in a given contest. This information is immensely valuable and should be evaluated and used to your advantage. Every week there will be situations all over the place like a wide receiver priced at 8,700 and projected for 14 fantasy points who will be 28% owned, while a wide receiver priced at 8,900 and projected for 13.5 fantasy points who will be 4% owned. Optimizers are insanely ‘dumb’ in their thinking and will often slam the first player because they have a slight bit of value since they are $200 cheaper and project for .5 fantasy points more. I think you know who the better DFS play is in this vacuum, though.
Another ownership trend to keep an eye out for are players whose ownership gets ‘steamed’, or artificially inflated due to touts, word of mouth, and sheep mentality. I generally advise not to intake much content each week, especially tout content where they all just throw the exact same names out. Still, it can be beneficial to take a glance to see which players are getting steamed by these touts in order to potentially fade these players in your lineups as they will likely be over-owned. The fades that are usually easy to identify are high-salary studs that everyone is talking about that week.
Salary usage is a spot where people think they are getting different by leaving some on the table, but that is largely a huge misconception and leaving those fantasy points on the table is detrimental to your lineups. 97-98% of the field uses $49,500-50,000 salary (keep in mind this is Draftkings), and it is just a place where you do not want to be different. On any slate with 3+ NFL games, I usually set a salary floor of $49,500. Note the screenshot below to see the distribution of salary used last season.
Another aspect of the field to keep in mind is how everyone is stacking. Last season, roughly 88% of the field did some sort of stack with their QB and 35-50% of the field used a player on the opposing team of their stack (will get more into that later). Roughly 75% of the field had a WR included in the stack, 25% had a TE included, and 20% had a RB included.
Player and Lineup Projected Ownership:
Almost equally as important as projected points, projected ownership (pOwn) is critical to NFL DFS success. Knowing how much the field will be using each player is a vital piece of information needed to correctly implement any kind of game theory strategy. Much like projected points, it is wise to have access to multiple sources of projected ownership in order to aggregate them for a more realistic sense of what the field will actually be doing.
Evaluating the ownership of a player is both an art and a science:
A good contrarian play consists of some combination of underpriced, under-owned, and has an opportunity advantage over players that are similarly priced.
A bad contrarian play doesn’t have equal or more avenues of success compared to the players around their salary and their primary contrarian value is just that they are low-owned.
A good chalk play can be compared to similarly priced players and have a conclusive opportunity advantage.
A bad chalk play has an ownership gap compared to similarly priced players that is far too large. The chalk player may have better opportunity advantage, but not enough to make the ownership gap size valid. One can also evaluate if the player has a large number of paths to failure or if they are playing a highly variant position.
Pairing good chalk with good contrarian plays is the recipe to set yourself up for a lineup that can separate itself and make its way to the top.
Any projected ownerships you get from a projection website will be modeled for the flagship contest, like the Milly Maker on Draftkings. Ownership, however, is very different between different types of contests. Ownership gets more concentrated as players are able to enter fewer lineups into a contest. As mentioned, projected ownerships are modeled for the 150-max entry flagship contest, so generally the highest projected owned players will be higher owned in 20 maxes, and even higher owned in single entry contests. Buy-in price is also a factor in ownership. As the buy-in price goes up, ownership tends to get more concentrated at the top.
Projected ownership is also useful in evaluating a lineup as a whole. Below I will outline three methods for devising an ownership metric for a lineup. There may be more and better ways, but these are the ones I am familiar with and have used or currently use.
Ownership sum: A basic metric that can be used to compare lineups based on the sum of each player’s projected ownership. This is the simplest method but should be avoided due to what is outlined below in the ownership product section.
Ownership product: A better iteration over ownership sum is ownership product takes the product of all of the projected ownerships for the players in a lineup. As shown below, this method gives a much better idea of a lineup’s ownership relative to other lineups.
You have 2 lineups, each consisting of 4 players:
Lineup A’s ownerships are 40%, 10%, 40%, 10%
Lineup B’s ownerships are 20%, 30%, 20%, 30%
Sum of Lineup A’s ownerships is 100%
Sum of Lineup B’s ownerships is 100%
Product of Lineup A’s ownerships is .0016
Product of Lineup B’s ownerships is .0036
You can clearly see that ownership product shows a clearer picture. If using ownership sum, one would think the lineups are equal with respect to ownership. If using ownership product, you are able to tell that Lineup A is using lower-owned players (10% owned) making it more of a contrarian lineup.
Geometric Mean (Geomean): Geomean is one step further and in my opinion an excellent metric to use. To find the geomean of a lineup, basically take the product ownership of the lineup and then take the Nth root of it where N is the number of players in the lineup.
(product ownership)^(1/number of players in the lineup)
Geomean of Lineup A: (.0016)^(1/4) = .2 or 20%
Geomean of Lineup B: (.0036)^(1/4) = .245 or 24.5%
To take geomean one step further, you can use it to estimate how many lineups you should expect to be duplicated with in a contest. This is more helpful in showdown (single game) contests where dupes are more common.
Say we have a 20,000 entry showdown tournament and we don’t want our lineups to have more than 5 dupes. We need to find the geomean cutoff for 5 dupes in this contest by taking the Nth root of the number of desired max dupes divided by the number of total entries in the contest, where N is the number of players in the lineup. Shown below:
(number of desired max dupes/number of entries in the contest)^(1/number of players in the lineup)
(5/20,000)^(1/4) = .126 or 12.6% or lower desired geomean
With Lineup A’s geomean of 20% and Lineup B’s geomean of 24.5%, one can see that these lineups are expected to have more than 5 dupes so they would not be desirable to us. Lineup A is expected to have around 32 dupes and Lineup B is expected to have around 72 dupes.
We are playing in tournaments to win, not min cash. Tournaments typically pay out 15-20% of the field, with about the top 10% doubling their money and the top 10 spots paying 20-50% of the total prize pool. Your winnings are evaluated on how many points you score relative to the others in the tournament, not solely on how many fantasy points you score. There are often high-scoring weeks where people question how they didn’t win when their lineups end up with a score that would have won them any of the other NFL weeks. Players shouldn’t be evaluated just on their point projection, but also on their price and ownership. Even further, instead of median point projections (the projections you receive from your favorite projection website), players should be evaluated on their upper percentile, outlier projections. To win one of these tournaments, especially one with a large field, you will generally need all of your players to hit their 75th+ percentile outcome.
When integrating pivots and contrarian plays into your lineup, you will likely have to give up some median projection points along the way. This comes with the territory and the player needs to evaluate just how far from optimal they would like to get in order to gain some sense of ‘differentness’ with respect to projected ownership.
The first thing that should be focused on when making lineups is ensuring that all lineups are correlated appropriately with stacks. About 12% of the field doesn’t stack each week. Part of that is people who don’t know how to play NFL DFS and the other part is people who are trying to play a mobile quarterback thinking that the combination of rushing and passing will provide enough upside, although it has been proven that the outcome needed for that to happen is far rarer than people think and the necessary upside in that lineup construction just isn’t there.
If you think about your lineup as a series of parlay bets, you’ll realize that taking correlated parlay bets greatly increases your chances of hitting all bets in the parlay. You can go too far into the correlation direction as well, though. Stacking your QB with 3-4 pass-catchers means that QB has to have an unreal performance if all 3-4 of those pass-catchers are to have the 75th+ percentile performance needed to win a tournament contest. The usual sweet spot is pairing your QB with 1-2 pass-catchers, and as a general rule of thumb the more expensive the QB the more he needs to produce to make up for his price tag, thus more production to go around which could support the 2 pass-catchers over the singular pass-catcher.
An additional correlated puzzle piece that is very valuable to a high-upside lineup is a player from your QB’s opposing team. There are many names to this part of the stack such as Opp, Bring Back, Run it Back, Game-stack, etc. Think through the game script: your QB and his pass-catchers that you have stacked are having a day and running up the score on the opposing team. The logical next step for the opposing team is to air the ball out to try to catch up as quickly as possible. This is a point of correlation that you can easily take advantage of. Only a third to half the field utilizes the Opp stack with their QB, so there is room to be over the field in that aspect.
The final correlation piece you can use in your lineups to help your set of parlayed bets be that much easier to hit is the mini game-stack. Aiming for games that you think could be high-scoring shootouts and choosing a piece from both sides to throw into your lineup can be just that piece. You are looking for teams that can go back and forth all game long, and in particular choosing some of the best options from each team.
So to sum up correlation, I generally pair my QB with 1-2 pass-catchers from his team, a pass-catcher from the opposing team, and 1-2 mini game-stacks from other games. Runningbacks can fit in the equation here in places, that just takes some personal preference.
Now to the point of actual methods to create your lineups. For years, using a simple optimizer integrated with some rules to force stacks was the way to go for a majority of DFS players unless you created your own smarter optimizer/sim that incorporated correlation for you. Nowadays, the best course of action is to create your lineups using a play-by-play simulator that is able to sim the slate thousands of times over and give you lineups directly based on that. This allows correlation to be naturally incorporated and your lineups will emerge with stacks without any need for forcing rules. I use SaberSim for my lineup creation because they use play-by-play simulations.
An added level of simulation you can add to your process is creating an additional pool of lineups using projected player ownership and modeled after the field you are expected to play against with respect to salary usage, stack usage, etc. Then simulate the slate thousands of times over while pitting your lineups against this additional field to find the projected ROI, win percent, etc of all of your lineups.
My final note on lineup construction is to not be afraid to take stands on players, but don’t feel like you need to own a player at 5x their projected ownership or completely fade a certain player. There is merit to taking stands but there is also room to spread your risk a bit, especially when playing 150+ lineups. Make sure you have a good mix.
One aspect of DFS that is rarely used by the casual player is late swap. Sure, the majority of people use it when one of their players is scratched, but this is only scratching the surface of its potential. As a slate progresses, you gain more and more knowledge that you can use to further iterate your lineups to further increase their EV.
There can be other use cases, but the two main late swap situations are as follows:
The noon NFL games are wrapping up and you played a very low-owned player that went bananas. You should consider swapping any other low-owned pieces in those lineups to higher-owned chalk.
Alternatively, if you played a chalk player who failed, you should consider swapping other chalk pieces in those lineups to lower-owned players in order to salvage some cash equity. If you keep those additional chalk players in the lineups, they may do well but all that does for you is moves you up the leaderboard with everyone else except you are still behind due to the original chalk player failing. If you substitute in low-owned players, you have a clearer path to the top if other chalk fails.
Manage Your Expectations:
These contests are very top-heavy. You have to finish very high to make any meaningful money. Focus on your process and your lineups rather than your weekly results. Use a website like https://rototracker.com/ to track your results through the season or even over your lifetime. Focus on having more than 1% of your lineups in the top 1% bucket in contests. If you do this, you are likely a winning player and may need to take some time to realize your EV. Everyone has downswings, some lasting a long time. If you know you have an edge, stick with it and focus on continuing to make +EV lineups.