The Best WR Prospect: What Do They Look Like?
In my first piece for RotoExperts, I detailed the stats that matter at the wide receiver position. For over 50 statistics I was able to find key thresholds to help see how likely it would be for a prospect to succeed. WR prospect evaluation is a difficult task but I have approached it from a pure mathematical lens.
But what happens when we blend them all together?
Using statistical programming, we have the ability to not just look at individual statistics, but a combination of statistics in a single regression tree model. The program will determine the statistics of highest significance, and split our sample into various buckets. Statistics of lower significance are thrown out and unused.
I put together two different regression trees for WRs. One involves a scouting component, and one does not. In other words, we have a model that is agnostic of film in any way, a purely numbers-based look. The other will take the valuable scouting information into consideration. I like looking at both to see if there are any prospects that the scouts may be missing in their evaluations.
Keep in mind as you read that I am defining a successful WR prospect as one who hits 200 or more PPR points in at least one of his first three seasons.
Understanding the Output
Before diving into the trees, I want to make sure that the output makes sense for you non-numbers folks out there. Here is what a typical tree split will look like:
At the top, you can see a node that has three outputs. The number on the very top indicates the success probability for players falling into that node. The number on the bottom left tells us how many players fall into that bucket, and the number on the right tells us the percent of our sample that number represents.
Afterwards you will see a statistic, in this case “dsrank” (I’ll give a key on the statistics later), as well as a threshold. Player who meet that threshold will go to the left, and players who miss that threshold will go to the right. So in this case, players with a dsrank of seven or higher go to the left, while players with a dsrank less than seven go the right.
This process continues until we arrive at a termination node, or final bucket.
X85BA – Adjusted Age
reyptpa – Receiving Yards Per Team Pass Attempt (Final Season)
creyptpa – Receiving Yards Per Team Pass Attempt (Career)
retdptpa – Receiving Touchdowns Per Team Pass Attempt (Final Season)
cretdptpa – Receiving Touchdowns Per Team Pass Attempt (Career)
cMSCRIMTD – Market Share of Scrimmage Touchdowns (Career)
The biggest split in the tree was adjusted age. As was discussed in the key statistics piece, this is one of the most valuable statistics at the WR position. 38 percent of WRs with an Adjusted Age under 20 became hits.
We can increase that 38 percent number greatly if a WR was also able to account for at least 3.3 receiving yards per team pass attempt. This metric is excellent because it blends together usage and efficiency. There were 16 players in the sample who fit this as well as the Adjusted Age threshold, and 14 of them were hits.
Even if a WR misses the reyptpa threshold, they can get to a 50 percent success rate by owning a career yards per team pass attempt of at least 2.5
There was still hope for players who did not meet the age requirement. Each of the two successful nodes on the left side of the tree came as a result of being productive touchdown scorers. If your favorite wideout is a little older, but has a proclivity for finding the end-zone, he can still find success at the next level.
(New) Key Stats
dsrank – NFL Draft Scout Rank
MSRECYD – Market Share of Receiving Yards (Final Season)
creyd – Receiving Yards (Career)
WR Prospect Analysis
The moment we place the scout thoughts into the model, they become the most important part of the analysis. 42 percent of WRs to be ranked inside of the top six at the position went on to be successes for fantasy.
Beyond that, we can see that career receiving yards matter for the first time. This likely eliminates your market share stars from offenses that do not throw the ball very much. Players meeting the raw yards and yards per team pass attempt marks for their careers were successful 76 percent of the time.
On the left side of the tree, final season market share was the major deliminator between hits and misses. It is worth noting that about half of the players (15) to hit a 0.44 MSRECYD were also ranked inside of the top six by the scouts.