Like everyone else I am hoping that we have NFL football this fall, but also like everyone else I want that to be safe (at least as far as football can be) from virus related problems. I doubt that will happen but we’ll see.
When sitting in the stands, I’ve sometimes mentioned the concepts of expected points (EP) and expected points added (EPA) but I’ve never had a great feel for how many EP a given situation was worth.
In a previous post, I looked at some run vs pass splits using expected points added (EPA). Here I’m going to explore 20 years of data to look at some basic but I think interesting things related to EPA using the great newish package nflfastr.
Let’s start with a review of EPA. By using models, analysts can estimate the expected number of points (EP) that accompany a given field position, down, distance, and other factors.
I previously took a look at draft capital spent by team from 2010-2019 and draft capital by team by position over the same period. Here I will extend that analysis further with the question: Does draft capital spent have an impact on the subsequent year’s performance?
To do that, I’ll look at draft capital spent on offense and defense across the league and predict the following year’s DVOA. Draft capital will be determined using the Chase Stuart model.
Pro Football Focus has a new article looking at the positions that teams have spent the most draft capital on over the past few years. This is similar to what I did here three months ago.
Let me review what I like about their approach and add some comments:
They used their own PFF WAR metric to measure draft capital of each pick. Although I don’t fully understand their approach, I like trying to quantify value in this way.
Pat Kirwan’s book with David Seigerman Take Your Eye Off the Ball was published in 2010. There is a second edition, published in 2015, but I’m focusing on a series of predictions made in the original in Chapter 14 (State of the Art Football).
Here I’ll review his predictions and make some of my own for the next 10 years. No huddle for all. Here Kirwan predicted that the pre play huddle would be phased out, with teams going into the huddle for 10 plays per half.
Today’s question: What position groups have each team invested the most draft capital in over the past 10 NFL drafts (2010-2019)? I looked at overall leaguewide trends in draft capital from the last 10 years here.
This analysis might indicate that certain teams have a tendency to target certain position groups more than others. Let’s not overinterpret these results, though. Teams can bring in players through other means than the draft, such as resigning previous draftees, or signing free agents (veteran or undrafted).
Today’s reading about the draft combine and the theme is that it doesn’t do very well predicting NFL performance.
As I read these studies, what really stands out is how difficult it is to measure individual player performance. A predictor can’t possibly do well if the performance measure is unreliable or not valid, and it’s clear there is a lot of be desired from the performance measures that are commonly available.
