top of page
GREG.jpg

The Process

I wanted to create a metric that can determine the success level an NFL wide receiver will achieve based on their athletic ability, collegiate statistics, and draft capital. In this case, NFL success will be measured on a players average PPR Fantasy Points per Game. I analyzed which metrics best correlated to prior NFL players success, and generated a score based on those criteria. This is my process of creating The Greg Jennings Score. 

Steps to success

Here you can read into the steps I took to build my model, or you skip past and play around with the simulation. Knock yourself out!

1 

Hypothesis Generation

I hypothesized that the most influential indicators were going to be:

  1. Draft Capital

  2. Age

  3. College fantasy points per game

  4. RAS score

  5. Level of competition

2

Research & Data Collection

I started my research with a favorite of mine; the Dynasty Football Flock Youtube channel. This gave me some basics to what criteria will be most important for creating my model, as he focuses on statistics such as draft capital, speed size, and breakout age. But even with an idea in mind, collecting data was a daunting task. Luckily, Peter Howard's PaHowdy database exists. It is a thorough breakdown of thousands of player's college and NFL stats. Almost all of my data was pulled from this database. Link below.

3

Data Cleansing

The original dataset was outliers galore. I needed to condense the player pool to a manageable amount and account for any obvious outliers, so I created the following criteria for a player to be included in my model:

  • Drafted between 2006-2022

  • Wide Receiver

  • Played at least eight games during their first season

  • If they played over two seasons, they must have played in at least 16 games. 

  • Seasons with zero games played are emitted

Next, I spent way too much time cleaning up formatting to allow automation in the analysis portion of this project.

4

Analysis

I started with over 30 different metrics that I theorized could be impactful in determining NFL success. After testing each metric to find which had the highest correlation, I was left with 12 metrics. (see the "Math for nerds" section for more information). These metrics were compared to three different categories to determine their validity:

  • Rookie PPG

  • 3 Year PPG

  • 7 Year PPG

And based on my model, the top indicators were:

  1. YAC/rec

  2. Draft Capital

  3.  40 time

  4. College PPG

  5. Age

  6. Rec YD/game

5

Crafting The Score

Once only 12 metrics remained, I analyzed which were the most impactful. Each metric received a score between zero and one to describe how high a player scored in the given category. The most impactful metrics were weighted heavier, then combined together to create a cohesive score. Each score was fed into the polynomial regression line of best fit to estimate how many PPR Fantasy Points Per Game each player is projected to score during their rookie season, first three seasons, and first seven seasons. 

6

Building The Model

This is where the fun comes in, at least for you. Everyone wants to draft the top-dog rookie wide receiver, but no one can accurately predict who that will be, until now. My model uses a players Greg Jennings Score to estimate his PPR PPG over three different time intervals. It provides the most likely finish for each player, offers a range of outcomes, and allows you to simulate who the highest scorers will be based on their Greg Jennings Score. You can also input criteria for a current college player, and see how they score in my model.

bottom of page