Have you ever watched your team call a screen pass or an inside run in a long-yardage third down? Has it left you confused and frustrated? The commentators may have remarked that it was the “smart” or “safe” play, leading you to believe that you must be wrong because the football experts must know what they’re talking about. You’re not alone. I find myself in this situation every Sunday, so I wanted to confirm my theory that the best decision on long-yardage third downs is to … well, try and gain a lot of yards.
To begin, I scraped every long-yardage third down play from the past 10 years.1 Our dataset includes exactly 5,984 different plays. Let’s see if teams usually elected to pass or run the ball in these situations.
NFL teams unsurprisingly tend to pass the ball in long-yardage third down situations. Let’s ignore the run plays for now and just analyze the passing downs. We’ll return to the run game later.
We want to see how aggression correlates with success when teams pass the ball on third-and-long. I used air yards to quantify aggression in this exercise. Air yards are how far the ball traveled relative to the line of scrimmage — anything that happens after the ball reaches the receiver does not impact air yards.
For each play, I divided air yards by the yards to go for a first down because a 20-yard pass on 3rd & 30 is not equivalent to a 20-yard pass on 3rd & 15. Then, I plotted the first down rate and the turnover rate along the x-axis (air yards) to see if there’s a relationship between these variables. Here’s the resulting line plot:
I marked the line of scrimmage with a blue line and the first down marker with an orange line. You can immediately tell that the average first down rate skyrockets when the ball is thrown near the first down marker. Shocking, right? The difference in the chance of success is huge, while the increase in turnover rate is incredibly small.
You may have noticed that the first down rate is greater when the ball is thrown behind the line of scrimmage than when it travels 20-40% of the way to the first down marker. It makes sense when you think about how screen passes work — you’re putting your best playmaker in the open field with a few blockers. While I’ve already shown that throwing the ball near the first down marker is the best decision on third-and-longs, I wondered if a screen pass might not be a bad decision on its own.
We can check this by looking at Expected Points Added and Win Probability Added. Both metrics should give us similar results, but we might as well use both because they’re at our disposal. Is a team’s EPA and WPA positive or negative on screen passes in third-and-long situations? It’s obviously going to be the highest when the ball is thrown near the first down marker, but maybe screen passes are also positive plays. Let’s see.
Nope. Executing a screen pass on a long-yardage third down play still decreases a team’s win probability. In most cases, the only good decision is to throw the ball near the first down marker. One surprising thing about this graph to me was the fact that the EPA and WPA for passes between the chains are roughly equal to the EPA and WPA for screen passes. The first down rates in the previous graph would lead you to believe that the EPA and WPA for screen passes would be higher, but this is not the case. I’m also not sure why there’s a huge drop in EPA and WPA at the line of scrimmage. My only theory is that plays are usually not designed for a player to catch the ball right at the line of scrimmage, so such passes would be indicative of a broken play.
Anyway, we’ve established that it’s essentially a no-brainer to throw the ball near the first down marker on third-and-long. Have teams realized this? My experiences as a football fan make me think that the answer is no, but maybe that’s just confirmation bias. Let’s see how often teams actually do make the right decision in long-yardage third down situations.
This trend is basically the opposite of what NFL teams should be doing based on our previous evidence. The best pass that you can throw on third-and-long is rarely the throw that teams actually attempt.
What about running the ball, though? We previously found that teams run the ball on approximately 17% of third-and-long plays. In 2018, the average rush attempt gained 4.4 yards, so you might guess that running the ball is generally unsuccessful on third-and-long. You would be correct. The first down rate on long-yardage third down rushes is just 5.50% — worse than a screen pass in the same situation.
Of course, coaches aren’t calling these run plays because they think they have a good shot at gaining the first down. They simply believe that it’s not worth the risk to go for the first down. However, the average EPA and WPA for third-and-long run plays is still negative. Even after factoring in the fact that a team is less likely to commit a turnover when running the ball, it is still a negative play on average. Granted, there are certain situations where the average doesn’t matter. If you have the ball and just need to waste another forty seconds to ice the game, there’s no reward for you to take the slightly higher risk of passing the ball near the first-down marker. More often than not, though, NFL teams don’t have an excuse for their poor decision-making.
Next time your team throws a screen or runs the ball on 3rd & 15, remember that you have a very good reason to yell at the TV. Don’t listen to those pesky commentators praising the offensive coordinator for making the smart play. Deliberately not calling a play that is intended to gain 15 yards when you need to gain 15 yards is not smart.