For those keeping score at home, this look at wide receiver completes our initial look at reliability and variance of all the standard fantasy football positions. But before we delve into the receiver correlation data, some shortcomings of what we've done so far should be discussed.
For what it's worth, reevaluating methods is a very common element of research. Sometimes things appear to be more thinking out loud than a presentation or actionable data, but the best way to come up with innovations is to throw stuff against the wall and see what sticks.
As I've been thinking this through, there are two faults with the presentations to this point. The first is every QB, K and Def each week has a realistic chance of being used in DFS - you absolutely know they'll be playing. Last week we looked at a running back inventory equal to 2 players per scheduled team in any given week. Later we'll examine a pool of three WR or every team playing. Realistically, not every one of these players is a practical choice. The problem is there's no real manner to absolutely identify the pool to examine each week. I have some ideas which will be presented later this week.
The second issue is the parsing of the pool into thirds and running the correlations for each tier. One issue was discussed last time. A high correlation doesn't necessarily mean the tier did as expected. It could mean they all performed worse or better but in accordance to their ranking. The larger problem is the top tier has nowhere to go but down while the bottom tier has nowhere to go but up. This is the main reason the middle tier always displayed the most variance since those players could go up or down. I am thinking through a fix to this problem.
In the interest of closure, here's the wide receiver correlation factor. For those new, the salary of the player is compared to the points as scored on three popular DFS sites. A high correlation indicates better predictability, which is a trait desired in cash games. A low correlation could be a target for GPP where you're looking for a lesser player to go nuts.
| Draft Kings | FanDuel | Draft Day | |
|---|---|---|---|
| WEEK 1 | 0.53 | 0.46 | 0.56 |
| WEEK 2 | 0.46 | 0.50 | 0.48 |
| WEEK 3 | 0.42 | 0.39 | 0.46 |
| WEEK 4 | 0.37 | 0.38 | 0.45 |
| WEEK 5 | 0.41 | 0.37 | 0.40 |
| WEEK 6 | 0.31 | 0.26 | 0.34 |
What's really interesting is the sites are getting worse at predicting wide receiver production which is probably due to emerging receivers as opposed to misreading talent. Injuries to some top wide-outs is also an issue. Calvin Johnson and A.J. Green come to mind as there salaries are high but there usual numbers are not.
Let's take a look at the corresponding running back data from last week with Week 6 added.
| Draft Kings | FanDuel | Draft Day | |
|---|---|---|---|
| WEEK 1 | 0.47 | 0.43 | 0.44 |
| WEEK 2 | 0.40 | 0.31 | 0.33 |
| WEEK 3 | 0.49 | 0.39 | 0.44 |
| WEEK 4 | 0.48 | 0.44 | 0.47 |
| WEEK 5 | 0.38 | 0.30 | 0.31 |
| WEEK 6 | 0.56 | 0.46 | 0.50 |
The numbers are close though it can be argued RB are a little more reliable than WR (which makes sense). More has to go right for a receiver to contribute.
Ultimately, the goal is to decide which spot to spend on and which to go bargain hunting. We don't have the answer yet but rest assured, I'm going to keep ruminating until something clicks. Trust me, it will.