2022 QB Analytical Review part 3

Author: Billy Jones

Introduction

Welcome back to the third installment of my blog series where I perform an analytical review for quarterback fantasy performance from 2022. In this blog, we start to “take a look under the hood” of fantasy football player production. We will be exploring the relationships between data points that make up fantasy scoring that simple logic would assume are closely linked, such as rushing yards and rushing touchdowns, passing yards and passing attempts, and more. We will be using visualization tools to help us determine if these relationships actually exist and if there are any outliers or anomalies in the data and will comment on what this might mean about their performance in 2023.

I'm particularly excited about this blog, as we start to get a little deeper into the realm of data science-y things. This blog touches on topics of correlation and anomalies, and how they can be used when evaluating player performance. These are easy concepts and tools which help towards a comprehensive understanding of what happened in 2022 and what could be in 2023. Let's get started!

Statistics 101

In today’s blog we will be adding two new data science terms into our study, correlation and outliers.

Correlation is a statistic that describes the relationship between two data points. It is expressed as a number between -1 and 1, with 1 being a perfect positive relationship, -1 being a perfect negative relationship, and 0 indicating no relationship between the data points. A positive correlation means that as one data point increases, the other data point tends to increase. A negative correlation is the opposite whereas one data point increases, the other tends to decrease.

It's important to note that just because there is a correlation between two variables, it does not necessarily mean that one variable causes the other. This is a common pitfall in statistical analysis, and one that we must be mindful of when analyzing the relationship between offensive market share and fantasy production. However, it's also important to remember that just because we can't pinpoint a clear direct cause-and-effect relationship between two variables, it doesn't mean there isn't something there. Correlation may still indicate that there is some underlying factor or factors that are causing the data to behave the way it does.

An outlier is an observation in a data set that lies far away from other values. Outliers can have a significant impact on the results of statistical analyses and can sometimes be indicative of measurement error or data entry errors (or fluky football results based upon a small-ish sample size).

Ground Rules

Before we jump back into the analytics, I would like to remind the readers of the ground rules we will be playing with. The most common scoring system is as follows so that’s what we will be going with:

1. Passing Touchdowns: 4 points for each touchdown pass.

1. Passing Yards: 1 point for every 25 passing yards.

1. Rushing Touchdowns: 6 points for each rushing touchdown.

1. Rushing Yards: 1 point for every 10 rushing yards.

1. Fumbles Lost: -2 points for each fumble lost.

1. Interceptions (INTs): -2 points for each interception thrown.

Additionally, I want to note we will be focusing on a pool of 32 QBs (shown below). The data used for this analysis was obtained from pro-football-reference.com, and all games where the QB had less than 10 passing attempts were removed to mitigate games where the player may have been injured or playing in garbage time. This will help ensure that the analytics aren’t skewed by anomalous game results and allow us to gain comfort in the conclusions we draw from the results.

Aaron Rodgers	Deshaun Watson	Josh Allen	Mac Jones	Taylor Heinicke
Andy Dalton	Geno Smith	Justin Fields	Marcus Mariota	Tom Brady
Brock Purdy	Jacoby Brissett	Justin Herbert	Matt Ryan	Trevor Lawerence
Dak Prescott	Jalen Hurts	Kenny Pickett	Matthew Stafford	Tua Tagovailoa
Daniel Jones	Jared Goff	Kirk Cousins	Patrick Mahomes
Davis Mills	Jimmy Garappolo	Kyler Murray	Russell Wilson
Derek Carr	Joe Burrow	Lamar Jackson	Ryan Tannehill

This analysis only encompasses 32 quarterbacks and therefore it's possible that some of your favorite QBs were excluded. The sole exception among QBs with more than 9 starts is Zach Wilson, who was not included due to widespread agreement that he is not good at football.

Visualizations and Analysis

With the ground rules established and the data sources defined, let’s get into analyzing some relationships.

In this analysis, we are examining two commonly assumed relationships in offensive production: the direct correlation between rushing attempts, yards and touchdowns, and the same for passing. Although we acknowledge that these are simplified views of offensive production, our scatterplots provide clear indications that these assumptions hold true.

* Visuals are colored based upon groupings identified in part 2. * 

Analysis: Upon examining the data, the relationship here appears strong for both rushing related correlations. Additionally, three points stood out as particularly noteworthy. We observed that Lamar Jackson and Justin Fields had rushing yards production that outpaced their rushing touchdown production, making them strong candidates for positive touchdown regression in the 2023 season. Conversely, Jalen Hurts had rushing touchdown production that outpaced his rushing yards & attempts production. Do we think these touchdown figures for Hurts are sustainable because his rushing profile is different than the rest as the Eagles short yardage runner or is he a prime negative touchdown regression candidate for the 2023 season?

Analysis: There is a clear strong correlation between passing attempts and yards, and a much weaker relationship for passing yards and touchdowns. Our analysis revealed that Patrick Mahomes II and Justin Herbert were notable players whose passing yards production outpaced their passing touchdown production. This suggests the possibility of even greater fantasy production from these players in the future. On the other hand, when it comes to the relationship between passing touchdowns and rushing touchdowns, our analysis showed that Justin Fields continues to be something of an outlier, with his passing production being imbalanced as well.

Analysis: This final visual shows how Brock Purdy had the most touchdown inflated fantasy profile out of all the players in our study. Additionally, while Justin Fields had some anomalies in his production relationships at the detailed level, a more zoomed out view showed his results were much more normal. This highlights the importance of taking a comprehensive approach to analysis, rather than simply reviewing a single statistic or visual in isolation.

General Point: Attempts vs. Yards for both rushing and passing were strongly correlated. Some players are more/less efficient but when looking for a quarterback that is going to put up points, look for a coach willing to make their quarterback the focal point of their offense. Yards/Attempts vs. touchdowns for both rushing and passing had much weaker correlation. While the relationship definitely still exists as players who produce more yards typically produce more touchdowns, the data is much more spread out with outliers on both sides. I hope to look into what are some of the key drivers of touchdown production in later blog series.

Conclusion

In conclusion, our analysis of the relationships between rushing yards and rushing touchdowns, as well as passing yards and passing touchdowns, has yielded some interesting insights into the offensive production of various players. Stay tuned for the last blog post in this mini study where we look into yardage production consistency.

Bonus visual: Average completions by average attempts (completion percentage).

* This blog post was enabled by ChatGPT. The text was generated by me, and the content is my own, but some sentences and wording were provided by the model. I take full responsibility for all information produced in this blog. More information about OpenAI and their technology can be found at https://openai.com. *