The process for coming up with what to explore is sometimes fairly easy. There might be a hot take on the radio or a sport discussion I have been a part of and want to explore it more. But it is always on something I find interesting. This post is dedicated again to the NFL draft, a topic I have touched on before (Success of a QB, NFL talent from Utah’s schools pt. 1 and 2, and why less film is better). This is also topical given we are almost in draft season and my question to answer was if the NFL draft trends have changed over time. My hypothesis was that teams have changed the way they draft over the last decade for three reasons: (1) Teams have evolved with the change in rules, (2) the passing game has never been more important, and (3) teams have become more sophisticated.
The scope of data collection quickly outgrew the time I had to do some in depth analysis. As I gathered 12 plus years of data I realized that I would need more data to suss this question out and that a draft decision was so specific to each franchise’s state per year. I couldn’t just simply gloss over the summary of data and extract gems to wow myself with. Another major factor would be to consider was the draft class per year and how that affected a teams decision. All of this is an attempt to say, what I hoped to find wasn’t easily found. However, there are some interesting patterns to be shared and some team specific findings.
Other important factors as I looked through the data to also keep in mind was:
A. The stability of the team’s ownership, coaches, GMs. I think this is one of the most important factors in a team’s success. Effectively the leadership of the franchise determines the success of the franchise. There are obvious examples to make, like the Patriots or the Steelers, who both have the longest tenured coaches and dedicated owners in the league. But also consider expansion teams like the Texans or teams in a constant state of rebuilding and trying to find their footing like the Jets or the Browns.
B. Part of the previous point is also how the leadership has embraced evolution of the game. Look at the Rams. A new owner comes on in 2010, they hire the youngest coach ever in Sean McVay, build a massive stadium, buy out stars this past year, and BOOM they win a superbowl. Or the Ravens, who with John Harbaugh at the helm, have fully embraced analytics. How do they think about prospects and positions.
C. Finally, the other factor that was less nuanced and easier to see was the performance of the team in the year before the draft. This became a way for me to segment teams and see how wins and losses affected the overall thinking for teams.
Again, to drive the point home, a teams situation is nuanced in every year. There are so many variables that it is hard to pick up on any single trend or pattern. That being said, here are some interesting finds. This data is from 2000 to the 2021 draft. I used the following websites: drafthistory.com, pro-football-reference.com, and 247sports.com.
The QB: Good v Bad Teams
The Patriots have drafted 12 QBs in the last 21 years, despite having an average +.714 win percentage in those years when they drafted a QB. Now, the summarized data hides the underlying truth that the Patriots were dying for a QB when Tom Brady left. On the heels of Brady’s departure and a dismal season of .438 in 2020, Mac Jones was taken 15th overall in the 2021. Jones was the earliest QB they took since Jimmy Garoppolo in 2014, who was taken in the second round at the 62nd pick (who was earlier than the previous draftee Ryan Mallett in the third round at pick 74). This illustrates the truth that in order to win in the NFL you have to find a franchise QB. The Patriots were lucky enough to find Tom Brady in 2000 in the sixth round. But they had been trying to find someone that could replace Brady. It wasn’t that they didn’t draft QBs, they just couldn’t find someone like Tom.
The Patriots might seem like cherry picking, but they have been the model franchise. Stable leadership and investments made into the franchise. They were willing to find talent, like Randy Moss, misfit toys, and even got more out of players than previously thought. That being said there are plenty more teams to choose from.
Now, consider the other side of the spectrum and once rivals to the Patriots, the Jets. The Jets have undergone a lot of changes in the last 20 years. A new owner in 2000 and a new stadium in 2010, and have had six different head coaches. No team has drafted more QBs in this time frame than the Jets with 13. The below charts illustrate their fall from grace since the days of Chad Pennington.
If anything can be gleaned from this analysis is that teams believe that QBs are the first step to the promised land. Teams with a less than .500 record (excluding the Texans) have a negative correlation of -.16 to QBs drafted, meaning that as the record declines more draft picks are used on QBs. Alternatively, teams above .500 (excluding the Pats) have a correlation of just .03. The Jets are a prime example of this belief.
Two other examples. The 49ers and the Browns and a word on each.
The 49ers are interesting as they were stuck with paying Alex Smith for several season, but he became serviceable under Jim Harbaugh. My Miners would then draft Colin Kaepernick who would take Smith’s job. However, under young head coach Kyle Shanahan, they have shown a lack of patience for Jimmy Garoppolo and drafted Trey Lance despite a good year.
The Browns tell a similar story to the Jets. A new owner comes on board in 2012 and they want to change their stars. They search and search for a QB.
Before we continue down this path and it is perceived that QBs change everything, I am one that is in the camp with Mina Kimes in that wins are not a QB stat. They have a real and substantial impact on the game, but wins and losses are shared on the entire team. And yes, I am going to cite myself again for you to read more about it here.
To find the association of categorical variables, like draft round and the position taken, you build a contingency table. This is just a fancy way of making a table that shows the frequency of when the two variables appear together. For example, count how many times a QB, DE, etc. shows up in round one, two, three and so on. From here you come up with ways to measure if the variables are independent or not (In the below plots I used a called a Chi-squared to find Pearson residuals). In the below charts this shows a measure of how much a position showed up in the table versus the expected given the distribution or make up of the column and row. The size and color of the bubbles show how strong these associations are.
This illustration was used to further enhance the difference in where teams spend their draft capital. Across the data set we can see that NFL teams looks to find QBs in the first round. What also can be seen clearly is that good teams look to improve their teams in other ways; see the below figure of teams with records above .500.
And finally, in the below plot, it becomes even clearer that bad teams hunt for the QB that may, but mostly likely NOT, change the franchise.
As a final takeaway, it was found that draft distribution of players taken are eerily similar to how an NFL franchise is composed, which makes total and complete and obvious sense. Why wouldn’t a team ultimately draft to the needs of filling a roster? But maybe that isn’t so obvious. Here is Lovie Smith on the Pat McAfee show explaining how a team’s foundation is built through the draft.
Consider that a typical team is comprised as follows vs what has been the proportion of players drafted in the below table.
|Position||Pct Typical Roster v |
Proportion of Drafted
|QB||4 — 5%|
|OL||17 — 17%|
|RB||8 — 9%|
|WR||11 — 13%|
|TE||6 — 6%|
|DL||17 — 18%|
|LB||13 — 13%|
|DB||19 — 20%|
So yeah, go figure, teams draft to their needs and most of the time the QB is the missing link.