It is not even September and we have already developed a list of universally understood fantasy truths.
We know Zeke will eat. We know Bortles can’t repeat 2015. We know Kirk Cousins has an E-L-I-T-E matchup in Week 1. And we know we will likely be wrong. A lot.
But so will everyone else.
Throughout the offseason we develop expectations that drive much of the daily fantasy sports behavior in September and October. These expectations can inform early-season pricing, and pricing then informs our expectations. It’s a vicious cycle.
But pricing changes, and results change, and “we could all use a little chaaaaange.” [Editor’s Note: Pretty sure that’s the first time we’ve quoted a Smash Mouth song on this website. Congratulations.]
By November of last year we learned that Devonta Freeman was not going away and by December his price on DraftKings had nearly doubled — going from $4,300 in early September to $8,000 by Week 13. That’s unreal.
I looked at our Player Models for the first week of December last year and compared Freeman’s Fantasy Year Salary Change to the Salary Change of the rest of the running back field.
Of the 10 non-Devonta players on this list, we have our standard mix of injury relievers as well as three other running backs who likely won someone somewhere a load of cash.
I was not one of those people.
I actually drafted Tevin Coleman in my season-long home league and the guy who did draft Mr. Freeman was on auto-draft and certainly did not KNOW who he was drafting. The auto-drafter missed the draft due to travel complications and ultimately went on to win the league.
That is also unreal.
Fortunately, with DFS you don’t really need to know who is going to have a monster season or even a monster week. In general, understanding what others know and how results vary is just as valuable.
If You Don’t Know, Now You Know . . .
Looking back at 2015, I used our Trends tool to understand how running back Plus/Minus changed throughout the season.
Note: This is based on all running backs with DK salaries of at least $4,000. The intent here was to capture most players.
We came into the year with relatively accurate models of expected points but by October players began to outperform expectations. It was not until December that pricing started to sort itself out.
Getting a few extra points above expectation is great on average, but, as Shania Twain famously said, “That don’t impress me much.” [Editor’s Note: That’s another first.] When I’m playing in a tournament or even large-field cash game I want players who are capable of drastically outperforming their salary-based expectations — of hitting their Upside (2x expectations).
So that’s what I looked for next.
Timing Is Everything
Using another series of trends, I looked at how frequently players hit the 2x threshold, taking into account both fantasy month and salary ranges. I’ll call this the ‘Hit Rate.’
Full disclosure: The salary ranges below are a bit arbitrary. ‘Cheap’ running backs are no more than $5,500. ‘Expensive’ running backs are no less than $5,600.
This is where it gets intriguing.
• Cheap running backs are much more likely to exceed greatly their expected output than expensive running backs. Cheap running backs have a much higher Hit Rate.
• The Hit Rate for cheap running backs is significantly higher in October and November compared to September.
• Because the first point felt obvious — it’s easier to exceed greatly a low projection — I added a “Weighted Points Above Expectation” metric to the graph because I really wanted to show off. This measure takes into account both the absolute point differential above expectation as well the frequency.
• And, finally, something I’ll expand on at another time: The Hit Rate does not appear to correct itself nearly as much as the projection.
The Nine-Yard Run
There is a lot to take away from here that I think is useful for both tournaments and cash games.
• Regardless of format and timing, cheaper running backs consistently provide more value than expensive players. In cash games they should be targeted almost as a matter of course. In guaranteed prize pools, they should be rostered when their ownership is not likely to exceed their odds of providing value.
• Because running back models are relatively accurate early in the season — actual points and expected points are often comparable — cash games might be especially attractive at that time. In cash games, we desire certainty. Early in the season, we have as much certainty at the running back position as we’re going to get. Conventional wisdom suggests that our limited information in September should produce more volatility, but at the running back position at least the salary a player has early in the season tells us a lot about what we’re likely to get out of him.
• October and November may be the prime months for playing cheap running backs in GPPs. The Hit Rates and Plus/Minus values for these players skyrocket. It seems that DFS platforms may be slow to adjust their pricing as cheap running backs emerge. In comparison to expensive running backs, these cheap runners can provide outsized value, which affords GPP participants the opportunity to invest more salary space in other positions.
The final point is that all data within this article came from the incredibly powerful Trends tool. In addition to providing guidance on rostering decisions, Trends can also help us backtest theories and discover performance and pricing anomalies, like those discussed in this piece.
