2024 Quarterback Analysis

Hierarchal Clustering

Let’s start the analysis by building hierarchal clustering models to visualize quarterback groupings in 2022 and 2023. To measure the value of a quarterback, statistics relevant to the fantasy football scoring were selected. To account for potential injuries that could have happened during the season, each player was evaluated on a per game basis, with the minimum number of games played being 8. Finally, pass completion types were broken down into percentages to measure the efficiency of the quarterback. Below are tables describing the variables used as well as the typical scoring for a quarterback in fantasy football.

Hierarchal Clustering is an unsupervised machine learning algorithm that seeks to build a hierarchy of clusters. Typically, all the observation points start within their own cluster. Using a distance metric, the two closest data points then get grouped together and form a new cluster. This process repeats until all of the data points are grouped together under one big cluster. The results of hierarchal clustering can be visualized through a dendrogram, which is a tree like graph that shows the cluster groupings at various levels of cluster sizes. In this analysis, Euclidean distance was used. In terms of linkage (the method the distances are used to measure), average linkage was used.

2022 Analysis

The visualization above is a dendrogram showing the clusters each player was placed in. The red boxes show the clusters of interest for this particular analysis. In theory, anyone could choose any set number of clusters, but in this case, we want to view the big groups. We want to be able to generalize what makes each group work and see any correlations between production/styles and fantasy finish. Let’s visualize the dendrogram above from the very first split and categorize the top 12 players from the 2022 season. Here, an L indicates a player on the left side of the dendrogram, while an R indicated the right side.

As we can see here, 8 of the 12 quarterbacks were clustered from the right side of the dendrogram, while three came from the left side (more on Justin Fields later!). In fact, the only player from the right side that didn’t make it in the top 12 was due to an injury (Tua Tagovailoa, finished 15th). The main factor between the primary split on the dendrogram was the passing yards per game and the ratio variable. Every single quarterback with the R designation averaged at least 250 passing yards per game and had a touchdown-interception ratio of 2 or higher. This roughly translates to at least 4000 passing yards and 20 passing touchdowns per season, which is indicative of a good season by many analysts.

We also see on the dendrogram that Josh Allen and Jalen Hurts separate from the other seven quarterbacks in the R cluster. This is due to their rushing ability that they have to go with their passing prowess. They both exceeded 700 rushing yards in 2022. For comparison, the next highest player in the R cluster was Patrick Mahomes with 350, and everybody else had below 250. Hurts, Allen, and Mahomes are seen as the gold standard of fantasy football quarterbacks for their ability to run and pass effectively. Therefore, the quarterbacks that can both pass and run effectively should be the quarterbacks that go off the board first in fantasy drafts.

Now, let’s break down the left side even further. This time, the big difference between the two boxes is the rushing ability of the right side. Lamar Jackson, Daniel Jones, and Kyler Murray all had or were on pace for at least 650 rushing yards for the season, which is on par with Allen and Hurts in the top tier. None of these players were on par from a passing perspective as the players described above, but they are considered intriguing fantasy players when you consider that a rushing yard is worth 2.5x more than a passing yard. Jackson and Murray were both injured in 2022, which is why they did not finish in the top 12 from table 3.1.

Finally, we need to split the remaining players to see what the difference was between them. If you can’t naturally see the final split, we are going to compare the Jimmy Garoppolo-Taylor Heinicke cluster to the Geno Smith - Dak Prescott group. Simply put, the players in the Smith-Prescott group are given more opportunities to pass the ball than the players in the other group. They are not considered to be in an elite offense or a special talent at the quarterback position, but they were trusted enough to have opportunities to pass the ball. These are players that are capable of being rostered or drafted in fantasy football, but have no real upside or advantages at the position. As for the Garoppolo-Heinicke group, these are either backups or bottom of the barrel starters, and should not be considered for your fantasy team.

The Passing QB or the Rushing QB?

There is a growing popularity in fantasy to draft quarterbacks with “upside”. That means that the player has a chance to outperform their draft position and potentially finish the season as a top quarterback like Mahomes and Allen. These players are considered to be the athletic freaks that can run the ball super efficiently. This logic in theory makes sense since rushing yards in fantasy are worth 2.5x more than passing yards. But does this notion actually have any substance? Let’s take a look at Justin Fields.

In 2022, Justin Fields took the fantasy football world by storm when he rushed for over 1100 yards with 8 touchdowns. The vast majority of those yards coming in the second half of the season. As a result, he won many people their fantasy championships and finished as the 10th ranked quarterback. He was a very popular fantasy player to draft in 2023. However, he only averaged 150 pass yards per game, and had a touchdown-interception ratio of 1.9.

You may have noticed that he was omitted in the dendrograms from the hierarchal clustering analysis. Well, here is the dendrogram with Justin Fields.

He was in his own cluster, unable to be grouped with any other quarterback. He was first in rush yards per game among quarterbacks, but last in pass yards per game. But is this playstyle effective over multiple seasons? Let’s take a look at the 2023 dendrogram.

The dendrogram in 2023 has similarities and differences to the 2022 dendrogram, but the clusters types remain the same. If we start once again by looking to the right hand side of the dendrogram, we see that we still have the pure passer cluster and the pass-run cluster. This time, the rushing quarterback cluster made it over to the right hand side, but still retained its meaning. The left hand side is equivalent to the inferior passing cluster from the 2022 dendrogram, with two exceptions: Justin Fields and Joshua Dobbs.

That’s right. The Justin Fields 2022 season was an outlier when compared to his 2023 statistics. His passing yards per game actually improved by about 50 yards per game, but his rushing yards per game dropped by 25 yards. This resulted in a net loss of about 0.75 fantasy points per game, which is steep considering that quarterbacks generally score the most points on a week to week basis.

So what can we learn from this?

  1. The rush first quarterback has a higher ceiling, but a lower floor than a normal passing quarterback. Their play style has league winning upside since rushing yards are worth more than passing yards, but this strategy can just as easily backfire since the player is not as strong of a passer and has a higher probability of suffering an injury.

  2. The pass first quarterback has a higher floor, but a lower ceiling than a running quarterback. They are more likely to produce consistently week to week, but they might not be able to produce the output that can win championships unless your draft was fantastic.

  3. Draft the quarterback that is a passer first, then go for the rushing upside. We all know that the trio of Mahomes, Allen, and Hurts are exceptional at both, but guys like Kyler Murray and Lamar Jackson are also fantastic options. Neither have ever passed for over 4000 yards, but have come pretty close on several instances. They can make up for it with their rushing ability, which puts them on par with some of the top options.

  4. Never draft a guy that is unproven as a passer. Not even the greatest dual threat options relied solely on the run game. It may work every now and then, but it is not a reliable strategy that should be depended on every year.