On Thursday night, Sam Darnold had an ugly performance against the Browns in a 21-17 loss. Darnold completed 15 of 31 passes for 169 yards, with no touchdowns and two interceptions (he also took two sacks). Now, if you didn’t watch the game, you might think Darnold’s two interceptions were the worst part of his game, or the main reason the Jets lost. But that’s not true: Darnold two picks came on his final three passes of the game. With 90 seconds left, the Jets trailed 21-17 with the ball at their own 37-yard line. Darnold was 14 of 28 for 144 yards with no interceptions and two sacks — a still terrible stat line that translated to an average of just 4.5 Net Yards/Attempt.

Yes, Darnold’s interception on the next play cost the Jets a chance to complete a comeback (and his second interception was close to meaningless from a win probability perspective), but the main reason the Jets lost was the ugly play over the first 58:30, not the last 1:30.

Not all interceptions are created equally, and Browns fans know this well. In the season opener, Cleveland had the ball at the Steelers 43 with 23 seconds left in the game. It was tied 21-21, and the Browns had 1st-and-10. The Browns had a great chance of winning, but Tyrod Taylor threw an interception on the next pass, ending Cleveland’s chances of winning the game in regulation.

The next week, Cleveland led New Orleans 12-10 with 5 minutes left in the 4th quarter. The Browns had somewhere between a 3-in-5 and 2-in-3 chance of winning the game at this point, facing 2nd-and-4 from the Cleveland 30-yard line. A moment later, Taylor threw an interception, and now the Saints had a 60-66% chance of winning the game. One could argue that Taylor has thrown the two most impactful interceptions of the season thus far.

The best way to analyze the impact of interceptions is through win probability. But that’s not always easy to understand or readily available, so let’s use some quick workarounds. If nothing else, the analysis here should help provide some context around interceptions.

I looked at all passes that were intercepted over the prior three seasons.

Time

The most common time for an interception to occur is at the end of either half. The graph below shows the number of interceptions in each minute of all games. There is a huge spike at the end of the first half, and also in minutes 59 and 60. None of this should be surprising, but it’s still interesting to examine nonetheless:

Scoring Differential

The most common points differential when an interception is thrown is… zero. That’s mostly a function of there being a lot of passes thrown in tie games, of course. But as you can see, teams are much more likely to throw interceptions when trailing than when leading (of course, some survivorship bias is included here, as the best quarterbacks are less likely to be trailing than the worst quarterbacks).

Down

What about interceptions by down? The graph is flatter than you might think, with a similar number of interceptions being thrown on first, second, or third down:

Tomorrow, we’ll take a deeper look at these numbers.