Trestman's 4th and Inches Call

I received a few requests to analyze Marc Trestman's decision to go for it on 4th and inches from his own 32, up by 4 with 7:50 to play in the 4th quarter. So here goes:

Punting would hand GB the ball at or around their own 20 yard line, worth 0.71 WP for CHI.

A successful conversion means a 1st and 10 at CHI's own 33, which would give CHI a Win Probability (WP) of 0.79. And a failed conversion attempt gives GB the ball at the CHI 32, worth 0.51 WP for CHI. [That's a relatively high-leverage situation--a potential swing of 0.28 WP.]

The break-even conversion probability (x) required to make it worthwhile to go for the conversion can be found by setting the value of the punt equal to the total value of the conversion attempt:

0.71 = x * 0.79 + (1-x) * 0.51
x = 71%

Fourth and 1 situations in that region of the field can typically be expected to be converted 74% of the time. So the total value of the go for it option was 0.72 WP, and going for it was defensible as the percentage play, but only by a hair.

The Bears went on to score a FG, killing 7 more minutes of game clock and leaving the Packers with only 50 seconds and no timeouts to try to tie the game with a desperation TD drive.

A few notes: I would always prefer to tell a coach 'Here is your break-even conversion probability...and by the way, the league average conversion rate is this,' rather than simply saying the model says go or the model says punt. This approach arms the coach with information rather than trying to replace his judgment.

The only criticism I have of Trestman is that CHI burned a timeout deciding what to do. In a one-score game at that point in the 4th quarter, timeouts can be worth a percentage point or two of WP. So by burning the timeout, CHI forfeited the slight advantage going for it offered.

Whenever possible, coaches should be prepared to make the 4th down decision before it arises. They can delegate the job to someone in the booth. For example, for each series a team has possession, someone in the booth can have a 'go for it on x or less yards to go' ready for the head coach and OC. This can be overruled of course, but it saves time and more importantly, timeouts.

One last observation--Note how muted the analysis was of the decision. Imagine the uproar if CHI failed to convert and went on to lose.

10 Responses to “Trestman's 4th and Inches Call”

1. Anonymous says:

i'd love to be an analyst for a team, but i can imagine getting fired when they ask me if they should punt or go for it, and i tell them they can do either.

2. Anonymous says:

As a Bears fan, I was screaming: PUNT! It felt like a probable win with a punt, a probable win with a conversion and a near-automatic loss with a failure the way the Bears couldn't stop the run. And there were too many Bears runs for no-gain leading up to that play.

But...
As an ANS fan, I was thinking: so glad we have an aggressive coach like Marc Trestman!

3. Jeff Clarke says:

I think this is kind of a landmark moment. A coach goes for it at a very high leverage point. He makes it and it plays a major role in winning the game. The fact that it was a coach most people haven't heard of makes it even better. Belichick ultimately knew he had the fan capital to survive the backlash. Did Trestman?

I didn't really care but was sort of cheering for the Packers. I changed my allegiance at that point.

I've been curious about the CFL and how this math would apply there. It strikes me that with only 3 downs, its even more important to not autopunt on 3rd down. During the few CFL games I've seen, coaches tend to be just as conservative.

4. Robert Burns says:

I thought that Chicago should run on 4th fourth and inches for two reasons
1. it was inches, not a full yard,
2. the Bear's offensive lune was outplaying Green Bay.

I think there were a only few Chicago running plays up to that point that did not gain yardage, and the pass protection was very good. Having the offensive line playing well makes the call a lot easier.

I am also not sure the time out was wasted, as the time out let the Bears decide on what play to call and get the right people on the field.

5. shawndgoldman says:

The other thing they could/should have done is challenge the prior play. The spot was at least questionable enough to throw the flag if one was going to burn the timeout anyways. That late in the game, there is very little cost to throwing the challenge flag if you're going to call the timeout anyways.

6. Dan Hart says:

As a Bears fan I like that our new coach makes "unconventional" calls like this. However, does GB having their backup QB playing effect the outcome of your analysis? If Aaron Rodgers is out there I think it's a great call, with Wallace, I don't know if he can drive them 80 yards for a TD.

7. Dan Hart says:

That seems to be the one thing no one is talking about while doing these types of analysis. Given that we are talking about a slight edge towards going for it there, I felt Wallace being QB quite likely neutralizes that, though looking at his overall career he is a surprisingly average NFL QB. So maybe it doesn't matter.

8. Xeifrank says:

I agree that you can use the league averages as a baseline, but for serious analysis you must take into consideration who the actual players are in the game.

Same thing in baseball. Fangraphs will start a game with win probability at 50% for both teams even though it is Clayton Kershaw pitching at home against a crappy pitcher from the Colorado Rockies.

Context is King.

9. Steveo says:

Jeff Clark,

The propensity to go for it on 3rd down in the CFL is significantly higher than in the NFL for 3rd and less than a yard. One reason is that with 3 downs, you just need to go for it more often, but the other reason is much more significant. In the NFL, the D-line lines up on the line of scrimmage, but in the CFL, the D-line has to line up a yard off the ball, so if the O-line gets a draw, thats half a yard.

I have never looked up the stats, but as a CFL fan, once a team reaches near midfield, going for it on 3rd and less than a yard is usually automatic. Also, QB sneaks are much more common. You will often see a team pull their starting QB out and put in their back up for 3rd down QB sneaks to avoid the hits on the starters.

Anyways, as a Packers fan, and a Canadian, I have to cheer for Trestman. He was a great coach in the CFL, and he is a class act all around. I hope he has a lot of success, and the Bears finish 2nd in the NFC North.

10. Anonymous says:

Yes, because the CFL defense gives up a yard, 3rd-and-1 situations can be converted most of the time (my guess would be close to 90%). Third and less than one are almost automatic. Not only must the CFL defense give up a yard, but the CFL offensive line can line up on the front tip of the ball, unlike the NFL O-line that lines up on the back tip of the ball.

OTOH, in 3-down football, the "two-and-out" is far more common than the NFL "3-and-out". So by punting on 3rd and short from midfield isn't giving up as much as in the NFL.

Also keep in mind that CFL timing rules are very different from the NFL, so clock-management tactics are not at all the same.

For starters the play clock is only 20 seconds, and within 3 minutes of the half, the clock stops on all plays until the ball is set. So 40-45 seconds is all you can run off with two kneel-downs. And if the clock hits 0:00 while the ball is not in play, there is still one last play.

So until you're into the last minute, one doesn't really play "kill the clock". Thus gambling on 3rd late in the game to get a first down only wins the game if you're at 40 seconds or less and the other team has used its time out. When the team that's behind is on offense, they don't have to get out of bounds or spike the ball, since the clock stops until the ball gets set - in a hurryup offense, the offense will usually be ready for the snap.

It's not unusual to see 4 or 5 possession changes in the last 3 minutes. This greatly changes late-game decision making.