It's a quantifiable edge in the game that no team exploits more often than the Ravens.
"What a good job they do of creating conflict right before the snap, changing your fits," Rams coach Sean McVay said in his post-game press conference after the 45-6 blowout, referring to Baltimore's pre-snap motion that altered defenders' run responsibilities.
This season, for the first time, ESPN Stats & Information is tracking pre-snap motion on every play in every NFL game and whether the player in motion came set at the snap or remained in motion. Of the 21,928 offensive plays run this season (excluding spikes and kneels), 11% of them included motion at the snap.
The Ravens have taken it to another level, though, and it's fueling their success. They have had a man in motion at the snap on 34% of their offensive plays this season, by far the most in the league. On Monday night, it was a whopping 47%. And it's working. Here's why the rest of the NFL should follow Baltimore's lead.
How effective is pre-snap motion?
Let's start by laying out exactly what we're talking about with pre-snap motion. From the NFL's rulebook, "When the ball is snapped, one player who is lined up in the backfield may be in motion, provided that he is moving parallel to or away from the line of scrimmage. No player is permitted to be moving toward the line of scrimmage when the ball is snapped. All other players must be stationary in their positions."
ESPN counts motion as a player running or jogging to a new spot (so running backs that just slide a step or a tight end who backs up off the line of scrimmage do not count). If a motion man gets to his new spot and faces upfield before the snap, that is considered "set" even if he isn't completely still.
But motion men still moving parallel to the line of scrimmage at the time of the snap is what the Ravens do more than anyone else and what we're focusing on today. It's also where offensive value is added.
Across the league, pass plays -- including sacks and scrambles -- with a man in motion at the snap have earned 0.08 expected points added per play (EPA/P) more than when there is no pre-snap motion. That's roughly the equivalent of the difference between the Chiefs' offense and the Raiders' offense this season. However, plays in which a player was put in motion but then was set before the snap showed only a minimal advantage over non-motion plays.
On designed-run plays, the edge was even stronger, with a 0.11 EPA/P difference in favor of motion at the snap over plays where there was no motion. And again, the edge for designed run plays in which a player was put in motion but came set before the snap was minimal.
On a per-play basis, the Ravens' gains on motion at the snap relative to plays where every player is set (non-motion and motion to set) slightly outpace the league average. It is worth noting, however, that Baltimore's rush plays with motion at the snap only maintain their (absurdly high) level of production compared with only non-motion plays.
The 49ers, who the Ravens face this weekend, put a man in motion at the snap at the fourth-highest rate in the league. San Francisco, however, has only produced similar results when rushing with a man in motion at the snap relative to when it does not, and it actually has been worse passing the ball.
So why is it so effective?
"[Defensive backs] are responsible for gaps or contain in the run game as well, and that motion can get them [out of the] correct gap," ESPN analyst Dan Orlovsky told me, reinforcing McVay's comment on the effectiveness of pre-snap motion. Analyst Matt Bowen added, "With pre-snap motion or movement vs. man-coverage defenses, defensive backs have to process with speed to communicate vs. bunch or stack sets, realign to avoid pick/rub concepts and avoid busting calls in the secondary."
The data also backs up what Bowen mentioned: Motion at the snap is a greater offensive advantage on pass plays against man coverage than against zone. Against man coverage, the EPA/P advantage for motion over either no motion or motion and set has been 0.12 -- we're talking the difference between the Chiefs' offense and the Bucs' now -- while against zone it has been just 0.04.
Through some more sophisticated approaches (read below for those), we found that the advantage provided by motion at the snap is slightly smaller than the basic observations above. But critically, those approaches also confirmed our beliefs and allow us to confidently conclude that motion at the snap appears to provide a distinct benefit for offenses, beyond just the fact that the Ravens are using it frequently.
The rest of the NFL can't copy exactly what Baltimore is doing, because there is no other Lamar Jackson. But this part of the blueprint is available for the taking.
Are the results biased by team strengths and tendencies?
At this point, we have to wonder. Because the Ravens are particularly good at pre-snap motion and run it more often they any other team, are they skewing the results? Or more likely, is it possible that just in general, the teams that have more success running motion at the snap simply run it more?
Those are plausible explanations. But there are a few ways we can tease this out, accounting for individual teams. The most basic approach is simply looking at the difference between each team's run and pass play EPA/P on a team-by-team basis with and without motion at the snap, in order to find out if the motion effect is confined to just a few good running teams or if it is common to most offenses.
The majority of teams improved their EPA/P by running motion at the snap, and the average of the difference favored motion at the snap on both run and pass plays, thus giving us more confidence in our hypothesis.
But this isn't quite right either. In our initial approach, we overweighted the teams that ran more motion. In this second approach, we're overweighting snaps by teams that don't run motion much.
My colleague Brian Burke applied a more advanced technique -- a hierarchical regression -- in order to estimate the signal attributed purely to the motion independent of team strengths and tendencies. That analysis produced results that suggested motion at the snap is worth .04 EPA/P on pass plays and .08 on rush plays (again, the difference between no motion and motion and set was negligible). This represents our best guess as to the true value of motion at the snap.
To be clear, our sample here is not gigantic. It's less than one season, but that amounts to over 2,300 motion at the snap plays. It's worth monitoring going forward. But ultimately, four different approaches -- baseline split, by-team split, an ordinary least squares regression (Burke tried that, too) and a hierarchical regression -- all pointed in the same direction.
The rest of the league would be wise to listen to that signal.
Brian Burke contributed to this article.