A Deep Dive Into The NFL Draft: Part 1

With draft season in full swing, it is important to utilize a quantitative analysis to determine the best practices for success. I understand that data can’t qualify every aspect of a player’s game, but even as one who vehemently disagrees with today’s overemphasis on analytics, I believe there is a place for it, particularly when it comes to high-level decision-making.

For time purposes, I will be dividing this analysis into three parts: the best approach, player and positional trends according to pick value, and projections as well as how the Chicago Bears can use this analysis for success.

In this article, we will analyze the first of the three, the best approach. Generally-speaking, the typical NFL Draft approach spectrum is as follows.

Obviously, any General Manager can lie at any point within this approach spectrum due to the tradeoff between number of picks and how high the pick is. However, we will analyze three scenarios: both extremes as well as a 50-50 balance between picks and quantity, so without further ado, let’s dive right in.

The Methodology

For those interested in the statistical aspect, I figured I would provide a brief synopsis of the process behind the data.

First, I constructed a web scraping tool via Python to extract and manipulate historical data for rookies in the 2021 NFL Draft class as well as their draft chart value and team’s point output.

Among these data points was adjusted value, from which I derived a new stat: pure point value, shortened to PPV. PPV extracts all external factors from a player’s performance to determine their particular points-based impact on a game (i.e., points contributed/detracted from the offense, points detracted/contributed from the defense).

I then linked this data with Microsoft Excel and immediately had all the data necessary. After manipulating it a bit, I generated their respective derivations as well, after which I used regression mapping to identify some interesting trends.

Approach Analysis

I provided each of the three approaches with a corresponding term for reference. Average Pick Chart Value serves as a measurement of a team’s draft position, Average Pick Depth analyzes a team’s average draft position within each round, and # Picks - as the name suggests - lists the quantity of draft picks a team had.

We will begin with Average Pick Chart Value, which patterned out to the following graph.

In this graph, we see that there is a slight upward correlation between a team’s Average Pick Chart Value and their Net PPV (summation of all PPV’s from each of their draft picks), proving that having higher picks has somewhat of an impact on draft performance. Obviously, this makes sense seeing as better players are typically drafted higher.

However, despite this trend, the R² only has a value of 0.061, demonstrating a very weak correlation.

Overall, despite the slight benefit higher draft picks provide, it is not nearly as important a factor of drafting as many make it out to be.

Next comes Average Pick Depth, which has a graph that looks as follows.

According to the graph, there looks to be a slight downward correlation between Average Pick Depth and Net PPV, which - again - makes sense. Within each round, the further your draft pick is, the less likely you will be able to draft an impact player.

However, similar to Average Pick Chart Value, the R² value worked out to 0.0221, demonstrating an even weaker correlation than the Average Pick Chart Value.

Overall, much like getting the highest draft picks possible, striking a balance between the quantity of picks and the highest draft picks has little to no correlation with draft success.

Last but certainly not least is # Picks with a graph as follows.

As you can see, this was by far the best determinant of draft success. There was a clear upward trend between the two variables, and the marginal benefit of each additional draft pick (0.3919) worked out to be much greater than the marginal benefit of the other two factors adjusted for value.

Additionally, this correlation was the strongest of the three factors with an R² value of 0.1429. Yes, in the grand scheme of things, this is still a relatively weak correlation, which further supports the need for more draft picks because it proves that the draft truly is hit-or-miss.

Now, there is somewhat of a bias towards the number of picks because more picks will inherently mean more cumulative value. When analyzing these three approaches with Average PPV Per Pick, # Picks was least impactful. But how can this be? The answer lies in the weightage of each player’s productivity. The teams with high pick quantities generally draft at least one star-caliber player, with many of their other selections being of little to no value.

Of the eight teams with the highest Net Pure Point Value, seven had 9+ picks in last year’s draft, and of those seven, all but one had at least one player with a Net Pure Point Value that far surpassed their peers’. Interestingly, the only outlier was the Kansas City Chiefs, who had two players with phenomenal Pure Point Values in Trey Smith and Creed Humphrey.

Contrary to what many might believe, that is a good thing. Drafting 1-2 players with superstar-type impact is much more critical for team success than drafting a handful of players with average impact.

Conclusion

All in all, it is clear that consistent success in the NFL Draft is best achieved by trading back and stockpiling draft picks.

This trend is largely due to the fact that the draft is a 50-50 proposition. Statistically-speaking, over 50% of first-round picks bust within their first four years in the NFL. Still, teams appear intent on valuing a few high draft picks over many more lower ones in the hopes that they actually know what they’re doing.

This trend also explains why ex-Bears General Manager Ryan Pace routinely struggled with obtaining star players via the draft, as he would routinely trade up rather than trade down. For new General Manager Ryan Poles to have greater and more consistent success, it is imperative that he trades back and focuses on increasing the number of “lottery tickets” rather than the chart value of each pick.

In the next article of this series, we will get even more granular, as we will break down player and positional performance within each round to determine which positions are most consistent in their output.