The Monday Morning Agent

Ask five coaches what "establish the run" means.

You get five answers. They are different claims, and they need different data. A quarter of a million plays, 18 columns each, can test two of them.

1 Run early, even when it is not working, and it pays off late. What it would take: first-half runs, second-half results, and games where neither team was already ahead at the half. The plays have all three.
These plays can test itThe one we rule on
2 Wear the defense down. What it would take: who was on the field for the defense, play after play, and how tired they got. Not in the plays.
Not with these plays
3 Set up play-action. What it would take: which passes used a run fake, and how each one went. Not in the 18 columns.
Not with these plays
4 Keep the other offense off the field. What it would take: the clock each drive used, and what the other offense did with less of it. Not what this tool counts.
Not with these plays
5 Teams that run a lot win. The one on the bar napkin. What it would take: each team's share of runs and whether it won, whole game. The plays have both.
These plays can test it
Space next Backspace back R start over

Where the research stands

On the first meaning: a Bates College study (Shea and Quenzel) of 429 games tied at halftime from 1994 to 2012 found no late payoff from first-half running. On play-action: Ben Baldwin (Football Outsiders, 2018) and Josh Hermsmeyer (FiveThirtyEight, 2019) found that the fake works whatever a team's run volume or run success before it. On the last meaning: winning causes running, because teams that are ahead run to use the clock, and the episode page shows how far the whole-game number moves once the halftime score is held close.

The agent rules on the first meaning. Its policy asks for the comparison that could prove it wrong: the first half only, in games within 3 points at halftime. Change the filters yourself and watch the answer move.