DraftKings Cash Games and GPPs: A 2019 Review
Taking a step back to review your DFS process and results is a necessary measure to becoming a profitable player, but many DFSers often walk away when the NFL season ends and don’t come back until Week 1. Others simply don’t play enough volume or a consistent enough game mix to properly review their results. This look back to the 2019 season on DraftKings will serve as a framework for what it took to be profitable last season. The following point thresholds outlined for various game types can be used as a reference point to examine if you were consistently building lineups that would be profitable over the long term in your game(s) of choice.
For this study, I focused on featured games from the main slate. Disciplined game selection could uncover games that may be beatable with lower point thresholds. For more on proper game selection, consider the Guide to Game Selection and this podcast theory segment.
DraftKings Cash Games
The featured cash games on DraftKings are the GIANT and MASSIVE Double Ups, where fields at the lower stake range from 10,000 to nearly 25,000 entries. While head-to-heads or 50/50s are usually a better long-term investment than Double Ups, playing in a huge field can help ensure that a given lineup falls where it should on the point curve, rather than being subject to variance in a small field. These large fields can be especially helpful for players only playing one or two games per week that are hoping to minimize variance, in terms of the sample size of their opponents.
All Double Ups are not created equally, however. While the stakes and size of the field are obvious variables to consider, the maximum number of entries allowed is an oft-overlooked aspect to large-field cash games. The difference between a single-entry Double Up and a multi-entry game can have a noticeable shift on the cash line of a game and, ultimately, one’s expected win rate. When profitable players can roll out one lineup 150 times in a single contest, the expected value for a less-experienced player declines sharply.
The following table highlights the average cash line (minimum score needed to cash) from Weeks 1–17 for featured Double Ups on DraftKings at various stakes.
|Buy-In||Multi-Entry Double Up||Single-Entry Double Up||Difference|
Click headers to sort.
The most striking difference shown above is the difference in the average cash lines for single-entries versus multi-entries at the low-mid stakes. From the $5 level to the $100 level, owners needed to average roughly three more points in multi-entry Double Ups than in single-entries and the difference was nearly six points at the $5 level, specifically. A 2–6 point difference can be a gap of hundreds, if not thousands of players, depending on the size of the contest. Except for the absolute biggest, most consistent DFS winners, seeking out single-entry cash games is a no-brainer.
Other than seeking out single-entry games, there are some stake-specific takeaways.
At the $2 level, the maximum number of entries is only 20, which limits how much sharks can dominate fields of nearly 25,000. There are likely enough casual players at these stakes that the average cash line is driven down, even in multi-entry games. The $250 Double Ups only allow 3–5 entries per person, explaining the convergence in the average cash lines between the two game types.
Single-entry Double Ups appear to have clear tier breaks in skill needed to win at different stakes, most notably between the $5 and $10 levels and $25 and $50 levels. Meanwhile, multi-entry contests have a relatively stable average cash line across stakes above the $2 level. DFS players who do choose to play these multi-entry games should be moving up aggressively in stakes in order to get the cheapest rake possible. Rake drops from 13% at the $2–$25 level to 12% at the $50 level. At $100 and $250, the rake is only 10%.
Games at the highest stakes sometimes don’t fill and there were multiple instances last season where those circumstances resulted in games with rake below 8% and as low as 1%. Opportunities like this can be bankroll savers. Because of the point jumps between tiers in single-entry contests, players should be cautious about moving up a tier (not necessarily a buy-in level) but those who are consistently producing scores that would beat the higher stakes should consider moving up for the rake considerations.
As was the case with cash games, this section will review average cash lines for featured tournaments on DraftKings but winning scores will also be discussed here. Additionally, a synopsis of the data presented in the weekly TJ’s #Taeks will show how winning players were constructing their lineups in 2019.
Average Cash Lines and Winning Scores
It’s nearly impossible to compare tournaments in an apples-to-apples manner since so many variables change from game to game, including field size, number of maximum entries allowed, payout structure and rake. Still, it’s valuable to look at scores from the biggest games on DraftKings in hopes of teasing out some actionable takeaways.
The following table breaks down average cash lines and average winning scores for the most popular (and maybe, more importantly, the most consistent) weekly featured GPPs on DK.
|Contest||Buy-In||Avg # Players||Max Entries||Avg Cash Line||Avg Winning Score||Min winning Score||Max Winning Score|
|Small Blind Side||$27||432||1||157.50||219.05||184.46||308.74|
Click headers to sort.
Probably the most valuable use of this table is for readers to find the games that are most similar to their usual contest(s) and see if they are putting up scores that can be profitable in these games, in turn, possibly finding leaks in game selection. Consider someone playing weekly at the $3 level who blindly fires a bullet at the Play-Action in hopes of bagging a six-digit prize. At the same buy-in level, the Pylon offers a significantly smaller field and an average winning score that is over 16 points lower than the Play-Action. While the prize in the Pylon is not nearly as big as in the Play-Action, a user playing less than $5 a week in GPPs that is looking for a quick bump to their bankroll might be better suited taking their money out of the large-field tournament in favor of the smaller contest.
In terms of general trends across all games, the strongest relationship between game variables and average scores was found by looking at the number of entries. The average number of entries in a contest showed correlation coefficients of -.67 and 0.64 with the average cash line and the average winning score, respectively. In other words, the more entries in a contest, the lower the average cash line and the higher the average winning score and vice versa. The explanation for this is likely (at least) two-fold.
An obvious narrative is that fewer entries mean fewer combinations of lineups which results in a lower likelihood of outliers on either end. A more theoretical explanation is that in smaller fields, players are likely taking fewer risks in lineup construction, opting for builds that are closer to optimal in a point projection system. Add in the fact that games with larger fields and bigger prize pools tend to attract more casual players and dead money, and it’s clear why smaller games tend to have higher cash thresholds and can be taken down with more modest scores than games at similar buy-in levels with far more players.
I asked 4for4’s resident data scientist, Kevin Zatloukal, to take a look at the data and he offered this:
“You can get a really good fit to the average winning score using the log of the number of players and max entries. The best fit...for DK is:
185 + 4.14 * LG(# players)
where "LG" is log base 2. That gives the average winning score with an R^2 of 0.97, so it's a very tight fit.
Doubling the number of players [in the field] adds about 4.14 more points to the score required to win.
One way of using that information is to decide between lineups based on their expected points and standard deviation. For example, you could decide between entering a lineup with a running back in the Flex or a wide receiver in the Flex. Usually, the RB-Flex lineup has a higher expected point total, which makes it almost certainly better for cash games, but the WR-Flex lineup usually (though not always) has a higher standard deviation. When that is the case, you can compare them to determine which one is more likely to hit the number of standard deviations above the mean that you need to win.
Use 4for4’s floor and ceiling projections to estimate a standard deviation on the projection. Get the estimate for the whole lineup by adding up the squares of those for each player and square rooting it.
Here are some examples:
In Week 2 of 2019, RB-Flex was projected 2.1 points above WR-Flex, but WR-Flex had a standard deviation higher by 0.17. In that case, the WR-Flex lineup is only more likely to win if you need 1.2+ standard deviations above the mean.
In Week 4, the RB-Flex lineup was projected 0.8 points above the WR-Flex lineup, while the latter had a standard deviation higher by 0.35. In that case, the WR lineup is more likely to win as long as you need 2.3+ standard deviations above to do so, which will happen for leagues with >1000 players.”
This isn’t to say that you should take bigger risks at every lineup spot just because the tournament has a large field—remember that larger-field tournaments generally have lower cash lines. The data here simply reiterates a notion that has been prevalent for years at 4for4: in order to be successful in tournaments be selectively contrarian around core value plays. A review of last year’s ownership numbers in DK’s biggest contests furthers that sentiment.
Roster Construction - Ownership
Throughout the season, I track winning lineups for the DraftKings Millionaire, logging data on how much winners spend on each roster spot and how popular every lineup is, as a whole and on the player level. The following table shows ownership percentages for every winning Millionaire lineup from 2019 (In Week 16, DK ran a $1,500 buy-in Main Event instead of the Millionaire).
|Week||QB||RB1||RB2||WR1||WR2||WR3||TE||Flex||RB or WR Flex?||DEF||Average|
Click headers to sort.
A zoomed-out look at ownership illustrates the fact that you don’t need to be contrarian across the board in GPPs to be successful, even in the largest fields. The 13% average ownership of winning lineups is enough to highlight that point but the idea is really driven home when you consider that only three winning Millionaire lineups last season had average ownership in the single digits.
Reviewing position-by-position ownership offers a more granular analysis of how winning lineups balanced chalk with contrarian plays.
Ownership, like salary and fantasy points, is generally quite flat at the quarterback position. Bottom- and mid-tier passers can, and do, match the weekly production of their elite counterparts on a regular basis. In fact, Millionaire winners had more success going against the grain than eating the chalk at the position—six winners used a sub-5% signal-caller while only one winning lineup used a quarterback rostered in at least 15% of lineups. Of course, there are far more quarterbacks drawing low ownership than ownership near 20%, but that’s kind of the point. Building an unpopular stack around a solid core of players is often a profitable plan.
Anyone who paid attention to the NFL last season shouldn’t be surprised that only six running backs used as an RB1 or RB2 in a winning Millionaire lineup showed up in fewer than 10% of lineups. Every-week studs were prevalent in winning lineups with Christian McCaffrey rostered six times, Derrick Henry four and Chris Carson, Dalvin Cook and Ezekiel Elliott appearing three times. Of the 25 highest-owned players in winning lineups last season, 17 were running backs.
Given the predictable, high volume of workhorse backs, it makes sense that it’s the position where we should be most willing to follow the crowd, even if it comes at a high price (more on that shortly). This isn’t to say that there isn’t room to be creative with your running backs, which will be shown when examining the Flex.
Contrary to running backs, fading the public—at least for the most popular plays—is often a wise strategy. Of the 32 players to show up in winning Millionaire lineups with at least 20% ownership, just eight were wide receivers and there were only two instances where a receiver on a winning roster was in over a quarter of lineups. The middle tier of ownership was the sweet spot. WR1s and WR2s in winning lineups both saw average ownership in the double digits and no WR1 was rostered in fewer than 10% of lineups.
As Kevin pointed out with his discussion on the Flex spot, selectively embracing volatility is a winning tactic in large-field contests and applying that logic to a team’s third wide receiver makes sense. This is a player that is rarely going to see huge target numbers or get predictable scoring opportunities so it represents a great spot to make a lineup unique.
Tight end has traditionally been a position in DFS where it makes sense to be contrarian—the position sees less volume than any of the other skill positions and there are usually only two or three really popular plays. Consequently, the chance of the chalk busting is usually quite high. In the last year or two, though, the gap between the top five or six tight ends and the rest of the field has expanded along with the pool of starting tight ends that are virtually unusable in fantasy. While a couple of mega-chalk tight ends hitting certainly drove up the average ownership seen above, there were still nine tight ends on winning teams that were on at least 10% of rosters. Unless a few more reliable options pop up at the position in 2020, continuing to roster a semi-popular tight end should be a viable strategy despite the inherent volatility at the position.
Before getting into this analysis, it should be noted that for this study, the Flex was defined as the lowest-owned RB3 or WR4 on a roster—defining a 30%-owned Christian McCaffrey as the Flex just because he was technically rostered there makes little sense.
Winning Millionaire lineups have historically seen a near-even split in running backs and wide receivers used in the Flex but 2019 slanted heavily towards running backs. This doesn’t necessarily refute Kevin’s discussion on embracing volatility, however. Most of the running backs used as a Flex in these lineups were low-owned plays, implying that they were likely highly volatile plays. Although Kevin compared a running back’s standard deviation to that of a wide receiver in his example, opting for the more volatile play when considering multiple players at the same position is a similar strategy that should be deployed in large-field tournaments.
The reasoning used for targeting low-owned WR3s extends to wide receivers in the Flex. It’s almost never a good play to roll out four relatively popular players at the position in these types of contests.
Needless to say, a two-tight end strategy should be used sparingly.
Defense, like quarterbacks, is a onesie position with generally flat scoring and ownership, which is why we see a relatively smooth ownership curve among the winners. Limiting your player pool to defenses that will be ahead, especially against teams that already throw a lot, is a fine strategy—not shying away from games that might be high scoring is a sneaky way to be contrarian. Big point totals often involve defense and special teams scores.
Roster Construction - Salary
Note that designations such as RB1 and WR1 in this section refer to the highest-priced player at the position on each roster. Consequently, the RB1 in terms of salary may not necessarily match up with the RB1 in terms of ownership above.
|Week||QB||RB1||RB2||WR1||WR2||WR3||TE||Flex||RB or WR Flex?||DEF||Total|
Click headers to sort.
Many of the takeaways in this section will parrot the points made with the ownership numbers, but there are some nuances worth exploring. On a macro level, leaving significant salary on the table doesn’t seem to be a winning strategy—only one winning Millionaire lineup left more than $200 unused and over half of the winners used all $50,000. Results could obviously be skewed if the overwhelming majority of users are using most of the salary by default—in other words, there may simply not be enough users experimenting with leaving substantial salary on the table to know if it can be a viable strategy.
The idea that quarterback scoring is replaceable shows up in virtually every type of fantasy contest and the salary distribution in these winning lineups restates that. No quarterback showed up more than twice in winning lineups in 2019 and there were twice as many quarterbacks used that were priced below $6,000 than those priced at or above $7,000. Using an expensive quarterback as part of your player pool can win but it clearly is not a must.
With the dominance of the aforementioned stud backs, it’s no surprise to see that the best lineups generally spent up for their RB1. The bigger takeaway here is that while the RB2 in winning lineups was often a popular play, it was generally a very affordable or cheap play. Again, regardless of talent or past performance, a running back that can be penciled in for an uptick in volume—whether it be as sudden a big favorite or because of injury to a starter—is a play that can usually be used with confidence. This is often the case with affordable but relatively popular plays.
Top-tier running backs have been so dominant in recent years that there has been a noticeable effect on high-end wide receivers in GPPs. Users have favored premier running backs, making it a challenge to roster expensive receivers and those running backs are scoring at such a high rate that the best receivers have rarely shown up in rosters atop the leaderboard, even when they are used. Only eight wide receivers priced $7,000 or higher were in winning Milly lineups and just two were priced above $8,000. This naturally meant that owners were also forced to save for their second or third receiver—only two winners spend $6,000 or more on three receivers and the average winner spent $17,350 at the position, excluding Flex considerations.
Reviewing the salaries of tight ends listed above might seem to refute the argument in the ownership section, but pricing doesn’t always catch up to public knowledge as quickly as we might think. Early in the season, there is an opportunity to find affordable tight ends who should have a significant role in their team’s passing attack. The pool of usable tight ends is still quite small, which is why ownership spikes quickly on these players. These cheap options do dry up, though—of the 10 tight ends in winning lineups priced at $5,000 or lower, only three came in the second half of the season.
With salary restrictions, it’s obvious that the last position player is the cheapest but there are some subtleties to notice when looking at Flex pricing. Winners spent more than $5,000 on their Flex spot just twice last year, suggesting that some version of a stars-and-scrubs approach trumps a balanced salary build in GPPs. Just three winning lineups in 2019 had no players priced above $7,000—owners don’t necessarily have to force an extreme stars-and-scrubs approach, though, as a Flex priced below $4,000 was used just three times.
As with ownership, defense saw a relatively smooth salary curve amongst winners, but as a whole, owners seemed more willing to pay up for defense in 2019 than in past years—the most expensive defenses were often the most popular last season. There is an explanation. From 2010–2018, there was an average of 28 games per season with a double-digit spread. In 2019, there were 42 such games. The result of more high-priced defenses that were massive favorites likely deterred owners from taking risks on cheaper defenses with a smaller spread. As the winning results show, though, there is still merit to saving money at the position. Having popular, pricey defenses in a GPP portfolio can work but it should be balanced with affordable options in favorable spots. If the number of large spreads regresses to the norm in 2020, there may be fewer defenses priced at the very top which could shift the average salary and average ownership numbers that we saw at the position this year.
Recap of Winning Trends
- DFS players should seek out single-entry cash games over multi-entry contests and look to move up to beat the rake.
- In general, the more players there are in a GPP, the lower the cash line but the higher the score needed to win. Players who cannot afford to max-enter multi-entry contests should often opt for smaller-field tournaments
- Quarterback scoring is flat and replaceable. Balance your GPP portfolio with high-priced, popular plays and cheap, contrarian plays. The same stands for defense, although a shift may be coming for the latter.
- Paying up for running backs, even if they are expected to carry high ownership, is a winning GPP strategy. It is the highest volume, most predictable position. If using a running back in the flex, going against the grain in a spot with a wide range of outcomes is a profitable way to make a unique lineup.
- The most expensive and the most popular wide receivers have rarely shown up in winning GPP lineups. This is a function of how dominant top-end running backs have been in recent years but also highlights the unpredictable nature of pass catchers on a weekly basis.
- Inherently, tight end is a volatile position, but there are so few usable players at the position that favoring somewhat popular plays has proven profitable. There are some value opportunities early in the season before pricing catches up to usage.