“Yards of Separation” is Fooling Us

Jun 27th 2021

Hayden Winks

I love NextGenStats. I don’t love how some of their stats are being used in the media. There’s so much context that’s involved with all football stats, so we have to be super careful about which ones sound awesome versus which ones are actually helpful in how we view players as fans and in how we rank players as fantasy footballers.

There’s not a single stat that’s getting on my nerves right now more than “yards of separation”.

In theory, it’s an awesome measure. In practice, it’s unhelpful. In fact, it’s actually hurting you if you’re trying to apply it to fantasy football.

“Yards of separation” had a -0.20 correlation to receiving yards among WRs last year, meaning there is a slightly negative relationship between the NextGenStats metric and how many yards a WR is gaining. Let me repeat that in another way. The fewer “yards of separation” a WR is being tracked on NextGenStats’ website, the more that WR is gaining yards on average. We are literally better off sorting by the “worst separators” when looking for fantasy clues than by sorting for the “best separators”. Amazing.

It’s not just weirdness with the data either. There’s on-field reasons why the data would show this, none bigger than a QB wanting to throw to his best WR regardless of whether his WR is “open” or “covered”. In other words, a QB is far less likely to throw to his №3 WR when he’s covered tightly than he is to his №1 WR when he is covered tightly. The QB is trusting his WR to make a play when he’s throwing into tight windows, thus creating an inverse relationship in that “yards of separation” data.

The other thing to note is that there’s a moderate negative relationship (-0.43 correlation) between “yards of separation” and average depth of target (aDOT), meaning that how deep downfield a WR is running his routes can explain where that WR ranks in “yards of separation”. The lower the aDOT, the higher “yards of separation”. Inversely, the higher the aDOT, the lower “yards of separation”.

There are reasons for this as well! First off, the data point is based on the catch point. There is simply more time for the defensive back to close the “yards of separation” on a downfield pass compared to a quick toss to the flats. Secondly, defenses are fine with allowing short completions like screens and 3-yard drags because they can rally to tackle before the offensive play is considered a success. A WR catching a screen on 3rd-and-14 will have a ton of “yards of separation”, even though he’s not “getting open” in the way we’d picture that metric would describe. Take this play for example:

Hopefully the next update to the NextGenStats website has a stat like “Yards of Separation Over Expected”, where they can account for how long the ball traveled in the air, the coverage type of the opposing defense, and the down-and-distance. Until that number is released, we should continue to expect “yards of separation” to have a negative relationship with fantasy points.