Daily Fantasy Sports vs. Season-Long

DFS Blog Header

Season-long fantasy sports have been played for decades.

Players started by keeping track of stats by hand-copying figures out of the morning paper box scores. From this humble beginning, season-long fantasy sports have grown to a massive American hallmark. According to the Fantasy Sports Trade Association (FSTA), in 2014, more than 41 million Americans engaged in some form of fantasy sport. (See more demographics at http://tinyurl.com/qguvtke.)

Throughout this massive boom, the game has fundamentally remained the same — friends or co-workers get together on draft day, they have a traditional-style draft to pick teams, these teams go head to head against each other repeatedly throughout the year leading up to the fantasy playoffs, which usually consist of some sort of bracket structure. The winner of the championship takes home all of the pride, glory and often, cash prizes. The losers lick their wounds and prepare for next season.

There are plenty of variations on the season-long theme. Different-sized leagues, different fantasy scoring systems, and different roster setups all add nuance to the strategy and fun of the season-long format. Dynasty leagues let players keep some or all of their teams heading into the next season, creating a whole new level of complexity and forcing owners to pay attention — even after it’s become clear that they have no chance of winning the league that season. Late-season trades and waiver wire pickups can set up these losing teams to dominate in the years to come.

Daily fantasy sports (DFS) is the newest evolution of the fantasy sports idea.

In the daily variant, users have a salary cap and are presented a set of salaries for each athlete against which they must construct a roster. Rosters play daily or weekly (hence the name), and you can redraft as frequently as the real-world games themselves take place. Currently roughly 1 percent of season-long fantasy sports players participate in DFS, but that number is growing quickly. DFS is only in its infancy — but the appeal of rapid turnaround games, the non-committal ‘only play when I feel like it’ format, and the potential of turning your every day DFS fanatic’s knowledge of their favorite game into (not kidding) millions of dollars all make it a pretty easy sell to newcomers.

I’ve been playing both formats for years, and I love them both in their own way. For season-long formats I enjoy the social aspect: staying in touch with friends, talking smack, etc. I also like strategic elements like picking guys up off the waiver wire or draft strategy. In DFS I like to strategize what sort of roster to play in which contests to give myself the best odds of winning.

As the CIO of DFS-newcomer StarsDraft.com, I obviously enjoy the more analytical nature of DFS, which much of this article is about. In season-long leagues I find myself a bit checked out on game day — seasons are long and no single fantasy performance will make or break you. Contrast this to the DFS format where every game takes on critical importance — it makes NFL Sundays even more magical, a feat I thought was impossible.

One of the biggest hurdles facing DFS is that not everyone understands the fun of it at first glance. Traditional DFS setups are essentially photocopied versions of season-long leagues, compressed into a single day. Everything from the roster setup to the scoring and even the player pricing seems to fit in with the season-long mentality. Many DFS sites implicitly suggest that DFS is just season-long’s “little, hyperactive brother” because of the way they present the game to their users. But the foundation — DFS itself — remains undifferentiated in many important ways. DFS has the potential to become not just a different variant of season-long fantasy sports, but something altogether more unique, more engaging, and ultimately more fun.

At StarsDraft.com we’ve fielded plenty of questions from long-standing DFS users who are used to their more season-long geared sites. For example, in NHL we ran analytics on goalie scoring as it pertains to the awarding of a win bonus and found that the win bonus was making the outcomes of daily games hinge far too dramatically on whether or not you drafted a goalie from a winning team. The incumbent system of goalie scoring lessened the skill and challenge related to putting together a roster for a single night. Consider that the wins stat – while a relatively important and telling measure of goalie performance when looked at over the course of the season – is relatively random and unimportant when the sample size is a single night. So we changed it — in our daily NHL product, goalies get points roughly in proportion to the quality of their goal tending for the night, win, lose, or draw.

In the NBA we’ve added an industry first with our substitute position. Any NBA DFS user knows the woes of having started a lineup where one or more of your starters just decide not to play on a give night. At StarsDraft.com, if your substitute outscores any of your starters, the substitute’s score will be used instead. This addresses an industry hole for the NBA (known as late-swap) by allowing the user more flexibility during roster creation.

In the NFL our player pricing is the subject of much controversy. We price our players agnostic to position — that is, our player salaries are meant to be as predictive of player performance as possible across all positions. Behind the scenes, we’ve employed a lot of high-powered analytics to ensure this is the case; accurate player pricing offers the fairest, most challenging experience to our users and limits player-pricing arbitrage opportunities that can damage the new-user experience. One result of this innovative approach is that prices react much more quickly to recent events or performance, and present a more interesting challenge to the fantasy NFL enthusiast. As a result of this unique approach, the QB position – the most consistent and prodigious producer of fantasy points by far – is consistently priced much higher than the other skill positions. TEs, on the other hand, are consistently priced lower at StarsDraft.com.

This is the subject of this post, the first in what I hope to be a series of posts breaking down the differences between season-long fantasy sports and DFS. How should a DFS user value the different NFL positions, by the numbers?

By The Numbers

Single-season performance by position

Let’s start with the data from 2011 through 2014 to give us a couple years in which the rules in the NFL were generally comparable, and a solid sample size across all positions. First, let’s look at the distribution of single-season fantasy point production by position (using StarsDraft’s scoring, which awards 4 points per passing touchdown and .5 points per reception, and is similar to standard scoring in other regards — for details see our FAQ).

Season Long Fantasy Sports

Putting it into words — Over the last four years, 68 of the single-season fantasy leaders were QBs, 20 were RBs, 10 were WRs and only 2 were TEs.

Implications for season-long formats — No one in a season-long format would posture that QBs get outscored by other positions on average. Everyone is aware that over the course of a season QBs will put up the most points on average. So why are the first players off the draft board RBs and, occasionally, a few WRs? A huge part of the strategy of season long is based on notions of exclusivity and replacement value. In a 10-team season-long league every team will be able to have a decent fantasy QB — and the difference between decent fantasy QBs and the elite fantasy QBs isn’t that large. Running backs are a different story, or so the established draft logic goes. At the RB position there are really only a handful of excellent options that you can expect to rely on week in and week out. These RBs might not score as many points as a mid-to-high level QB, but they are so much better than the other RB options that the value of you having them and your opponents not having them is huge.

Implications for DFS — Who cares about season-long totals? DFS is played weekly for the NFL, sometimes on even shorter time frames than that. What matters in DFS is per-game production. On this more relevant metric, the case for valuing QBs is even more compelling: over the last four years, 73 of the top 100 per-game season averages are QBs, 21 are RBs, 5 are WRs and only 1 is a TE. QBs don’t just score more points than the other positions per game, they score *a lot* more points, and it’s not close. The only TE to make that top 100 highest per-game average list is Rob Gronkowski — in his monster 2011 season he averaged 15 fantasy points per game. Even this herculean effort only lands Gronk at the 73rd best single-season per game  average. When you look at the numbers you can see that Gronk was out produced by very mediocre QBs like Mark Sanchez and Ryan Fitzpatrick in that same year.

In a daily format the objective is to roster a team that puts out the most points, period, end of story. Everyone is working with the same salary cap, and you cannot block your opponents by rostering certain players early or paying more for them. The best arguments for pricing Gronk at the same level as even a mid-tier QB include:

His value above replacement is much higher than the other positions.

Since you have to roster at least one TE, rostering the one that will actually score points is exceptionally valuable.

His ceiling is so much higher than other players.

If you thought you always rostered Gronk because he actually put up more points than that elite QB, WR or RB on average, you may want to review the data. I think the first two bullet points above have some validity and we’ll get to them below, but first lets look at the argument that Gronk and the elite TEs have a higher ceiling than do other positions.

Single game performances by position

To get a handle on the ceiling question, let’s look at the top 100 single game fantasy performances :

One-day Fantasy Sports

Putting it into words — Over the last four years, 60 of the best single-game fantasy performances were QBs, 17 were RBs, 22 were WRs and only 1 was a TE.

Implications for season-long formats — There aren’t many. If you need to stretch to beat an opponent, you might favor WRs slightly over RBs (and the effect would be even greater in a full PPR league). But other than that, there’s nothing new. QBs have the highest upside, but fortunately there are so many to choose from you’d be foolish to spend a high draft pick on a QB. RBs and WRs are the really rare commodities.

Implications for DFS — If you’re inclined to like the ceiling argument for pricing TEs as high as QBs or top WRs, you’re just wrong. The only TE to make the top 100 highest fantasy points in a single game list is the 37.4 points Gronk put up week 8 of 2014.

Looking at value above replacement

Ok, so on average QBs dramatically outscore all other positions and TEs typically underperform. It’s pretty clear that TEs also have a lower ceiling — including the elite TEs like Gronk, Jimmy Graham or Julius Thomas — than most other positions when looking at the highest single-game outputs. But there is still an argument to be made that the elite TEs have a higher replacement value at their position than do QBs. According to this line of thinking, the issue with not paying up for Gronk in a week is that it leaves you with some other TE. And if Gronk’s average13 point production seem paltry, when compared to the production of the TEs you are forced to use in his place 13 points looks like a pretty sweet deal.

To get a handle on this perspective I looked into the ‘average value above replacement’ by position – taking the points scored by each player in each week and comparing it to the average points scored by all players at that position in that week. The difference represents how much better a player produced than some hypothetical ‘average’ player at that same position:

Value Over Replacement

Here in lies the redemption for the elite TEs in DFS, and this is probably the source of most of the mystique surrounding Gronk and Graham. Respectively they outscore other tight ends by 8.6 and 6.6 points on average – ranking them the 2nd and 8th most differentiated players at their position in the league. Everyone has to draft at least one TE, so deciding to past on Gronk and Graham leaves you settling for the Julius Thomas’ or Antonio Gate’s of the world, giving up almost 6 points in the process.

While the value above replacement figures are encouraging for the elite TEs, they don’t do enough in my mind to dispel the misconception that Gronk and Graham are on a par with elite WRs or RBs. Value above replacement is just one consideration when creating a roster, and in all likelihood the money you save by avoiding Gronk and Graham on sites where they are priced as top WRs and RBs could be better spent on … well – top WRs and RBs! Notice as well that the elite QBs here still hold their water with Brees, Manning, Rogers and Luck all coming in the top 20 in terms of their value over replacement. And that value over replacement comes with a much higher floor (see their average points) and a much higher ceiling (see the single-game statistics).

It’s this kind of analysis that supports our position-agnostic analytics driven salary algorithms, and undoubtedly drives the traditional DFS holdovers nuts when Graham, priced appropriately low, puts up 0 points as he did last week! That performance isn’t included in this analysis – but it only serves to strengthen the case. Of course, mistakes happen more often than we’d like (see the oft cited Kelvin Benjamin debacle) as a result in human failure for manual processes that we use to verify and tweak algorithmic projections. But the goal is always to put out the most accurate set of player prices in the industry to enhance the fun and challenge of roster creation. DFS is a unique game that’s only beginning to come into it’s own. StarsDraft’s advanced player pricing, and unique perspective on all things DFS, aims to catalyze that differentiation process.