A mathematical look at the 2018 Minnesota Vikings

It’s been a long time since yours truly has taken a math class. Like, a looooooooooooong time. My last math class was so long ago that I had one of those Texas Instruments graphing calculators and thought the things that it could do were amazing.

Shut up, I know I’m not the only one.

But in the links to stories relating to the Minnesota Vikings that I get from Google every day, buried in there was this story from a man named William Butler. Butler took all of the play-by-play data from this past Vikings season and attempted to use the data in conjunction with “machine learning” techniques to see if he could figure out exactly where the purple fell short in 2018. According to Wikipedia, machine learning is

the scientific study of algorithms and statistical models that computer systems use to effectively perform a specific task without using explicit instructions, relying on models and inference instead. It is seen as a subset of artificial intelligence. Machine learning algorithms build a mathematical model of sample data, known as “training data”, in order to make predictions or decisions without being explicitly programmed to perform the task.

Now. . .I’m not going to pretend that I understand the majority of what Dr. Butler is talking about when he goes into the various algorithms and things of that nature. I’m sure that there are a number of you that actually do understand all of this stuff, and if you do feel free to explain to your heart’s content in the comments. But here are a few “plain language” takeaways from the article:

1. Whether it’s a function of where they lined up or the routes that they were running, when it comes to Adam Thielen and Stefon Diggs, they appear to prefer different sides of the field, particularly in the deep passing game. Thielen got most of his deep passing yards on the right side of the field, while Diggs got most of his deep passing yards to the left side.
2. The Vikings were the second-least penalized team in the NFL in 2018. They only had 88 penalties against them accepted (47 offensive penalties, 41 defensive penalties). Honestly, didn’t it seem like they had more penalties this year than they did in past years during the Zimmer era? I know it did to me, but apparently that wasn’t the case.
3. The lack of a real third option at wide receiver really hurt the Vikings quite a bit, compared to what Kirk Cousins dealt in his time in Washington. According to Butler’s article, in both 2016 and 2017 with the Washington Redskins, Cousins had five different receivers go over 500 yards in each of those two seasons. This year in Minnesota, he had three (Thielen, Diggs, and Kyle Rudolph). The drop-off from the two primary wide receivers to Rudolph was pretty significant, and the drop-off from Rudolph to the next tier was significant as well. Dalvin Cook wound up being fifth on the team in targets with 51, and he didn’t even play about half the season.

According to NFL Savant, which is the site that the statistics that Dr. Butler used are drawn from, Thielen, Diggs, and Rudolph accounted for over 65% of the targets the Vikings had in 2018. That’s only slightly higher than the 62% that those three accounted for in 2017, but the percentage for Diggs and Rudolph were slightly lower in 2017 than they were in 2018. In addition, Jerick McKinnon had nearly as many targets in 2017 as Rudolph did. . .a role that likely would have been filled by Cook had he not had his injury issues.

I don’t know if this makes WR3 a much higher priority for the Vikings this offseason than I originally thought, but it’s possible. I’m not sure if Aldrick Robinson is the answer, and I’m pretty sure that, at this point, nobody thinks Laquon Treadwell is. The Vikings could use someone to take the heat off of Thielen and Diggs, and while I think we can all agree that the offensive line is the team’s biggest need, maybe that third option should be higher on the list of priorities.

In any case, be sure to check out the rest of Dr. Butler’s article linked above. . .and, again, if anyone is more well-versed in mechanical learning than I am (which is not exactly a high bar to clear), feel free to share your knowledge with everyone else.