## Jim Zorn on 4th Down

Zorn is my hero today. On the Redskins’ final drive of their game against the Rams yesterday, head coach Jim Zorn went for it on 4th down not once, but twice. The network commentators were shocked, and the local media coverage has been decidedly critical. Were they good decisions?

The first decision “felt” right to me. Up by 2 points with 3:47 left in the 4th quarter, Washington faced a 4th and 1 at St. Louis’ 20 yd line. A FG attempt from there is an 84% proposition. A kickoff with a 5-point lead gives the Redskins a 0.76 Win Probability (WP). A missed FG attempt gives the ball to the Rams at the 27 and leaves the Redskins with a 0.56 WP. The net WP for the FG attempt is:

(0.84)(0.76) + (1-0.84)(0.56) = 0.73

For a 4th and 1 between the 10 and 20 yd lines, conversion attempts are successful 74% of the time. With a successful conversion the Redskins would have 0.96 WP. A failed conversion leaves them with a 0.62 WP, (not much different than a missed FG, but 7 yds further up field.) The net WP for the conversion attempt is:

(0.74) (0.96) + (1-0.74)(0.62) = 0.87

Although we’re using league baselines, 0.87 WP for the conversion attempt is much higher than the 0.73 WP for the FG attempt. The circumstances of the particular match-up would have to be overwhelmingly in favor of the FG for that to be the better decision. Zorn made the right call here.

The Second 4th Down

What about the second 4th down on the 2 yd line? I wasn't so sure about that one. Later in the drive, with just under 2 minutes remaining, Washington faced a 4th and 1. Close to the goal line, at the end of games is where WP is least confident, so I’m going to use drive scoring probabilities to back-up the WP model. Because it’s clear that no matter what happens on the 4th down, there is only enough time for St. Louis to make a single drive, I’m going to look at how likely they would be to score a FG or TD. Between the WP model and the drive model, I’ll take the number that is most optimistic for the FG option. This way, if the ‘go for it’ option comes out on top, it will be that much more convincing.

Short yardage conversion attempts near the goal line are less successful than elsewhere on the field. One-yard conversions are successful 68% of the time. A successful conversion attempt would realistically give the Redskins a 1.00 WP. At worst the ball would be at the 1 and Redskins could run out the clock to the point where the Rams could not possibly score. Still, let’s call it a 0.98 WP keeping in mind the Bettis fumble in the Pittsburgh-Indianapolis playoff game from the 2005 season. A failed conversion likely leaves the ball at the 2 and forces the Rams to go 63 yds to get into FG range. And even then, they’d need to hit a long FG. My model says Washington would have a 0.80 WP. The net WP for the second conversion attempt is:

(0.68)(0.98) + (1-0.68)(0.80) = 0.92

A FG attempt from that distance is virtually automatic—99%. A made FG gives Washington a 5-point lead and a 0.78 WP, as the Rams would have about a 1 in 5 chance of scoring a TD. A missed FG leaves the ball at the 20, giving the Rams a 0.28 WP and Washington a 0.72 WP. But this is an extremely unlikely possibility, so the exact number is inconsequential. In total, the net WP for the FG attempt is:

(0.99)(0.78) + (1-0.99)(0.72) = 0.78

This is a pretty healthy advantage for the conversion attempt—0.92 WP vs 0.78 WP. The game-specific particulars would have to be overwhelmingly in favor of the FG to change the final analysis. And remember, the more optimistic numbers were chosen for the FG attempt.

Zorn was right to go for it, even though they didn’t make the conversion.The Rams got the ball on their 5, and now they needed only a FG to win. However, what many people leave out of the analysis is that they were forced to begin their drive buried in their own territory.

Head coaches like Zorn can’t be expected to do all kinds of math to figure out what to do on 4th down. He has 15 seconds to make a decision and isn’t sitting in front of a spreadsheet on a computer. But Monday morning critics do have the luxury of time, so if they want to criticize they better do their homework. As reader Jeff Clarke points out, advanced analysis isn’t required.

The FG scenario requires the Rams to drive 70 yds for a TD. The ‘go for it’ option requires the Rams to drive 60 yds to get into FG range, and from there FGs are only good 50% of the time--And that’s only if the 4th and 1 conversion fails!

My final point is that outcome bias rules the day when it comes to these discussions. Had the Redskins converted, Zorn would be praised as some sort of genius who ‘sent a message to his guys that he was confident in them,’ or ‘was making a statement about what kind of team they are,’ or ‘understands the flow of the game.’ But because the attempt failed, he's absorbing lots of criticism. Imagine if the Redskins had actually lost.

Decisions shouldn't be judged based on the final outcome, when things that can't be controlled come into play. Any decision, whether in football or real life, is best judged based on the facts available at the time. Even if the Redskins had lost, Zorn made the right call.

### 33 Responses to “Jim Zorn on 4th Down”

1. John Candido says:

Great points. The general public is so caught up in binary outcomes that they don't realize that it is in the best interest of coaches to maximize expected logarithmic utility as the only way to consistently win championships and build dynasties. The arbitrage opportunity that currently exists in the coaching field is wide open.

2. Dave says:

I still think that the last 4th down call was questionable, even though I respect the opinion stated here. Zorn's bigger issue, though, is play calling and execution in the red zone. Maybe the numbers say that his play calls made sense based on league wide averages, but the Redskins experience during the Zorn era suggests he needs to take easy points on 4th down in the red zone.

3. Anonymous says:

Will you be doing your weekly win probability this year? I really loved looking at that last year for better insight.

Thanks,

Mike

4. nottom says:

Out of curiosity, how does Tomlin's decision to kick the FG (that was missed) on 4th and 2 from the 25 look versus going for it? Even if he makes it, 3 minutes seems like a lot of time for the Bears to get back down the field to tie or win. Obviously with those 2 teams on what seemed like a bit of a sloppy field, they were probably right to kick, but I think its likely closer than many would think and no one (in the mainstream sports-media) would ever question that decision even when it goes bad and would just blame that on the kicker.

5. Anonymous says:

This analysis of the Rams chances seems incomplete:

You seem to have assumed they can't score a TD.

You say, 'The ‘go for it’ option requires the Rams to drive 60 yds to get into FG range, and from there FGs are only good 50% of the time'

but they don't need to stop in FG range, see the end of the Raiders drive or Chargers drive or Broncos game last week or etc.

'A failed conversion likely leaves the ball at the 2 and forces the Rams to go 63 yds to get into FG range. And even then, they’d need to hit a long FG'

Pls describe more fully showing the odds of Rams scoring a TD, as they do not need to hit a long FG. Your analysis seems flawed, if you have incorporated the odds of a FG or a TD, pls show.

It seems obvs wrong that the Rams are 22% to score a TD after a FG, but 28% to score a TD or FG if Was misses the 4th down conversion. After all, one blown coverage or tipped ball or fallen defender or perfect fly pattern is a TD bomb from 65 or 90 yds.

As you say, the end of game stuff is not as precise as we'd like.

tks,

6. Anonymous says:

'Between the WP model and the drive model, I’ll take the number that is most optimistic for the FG option.'

Should be odds of FG + odds of TD, obvs.

7. Brian Burke says:

Sorry, I guess it seemed obvious that the Rams trying to score a TD on a 95 yd drive with 1:45 to go (failed conversion) would be more difficult than scoring a TD on a 70 yd drive (made FG). The WP model is based on real-life examples of real games, so the possibility of 'home-run' passes or runs is already incorporated.

The article says it's a 20% shot of scoring a FG or TD from the 2, and it's a 28% of scoring a FG or TD from the 20. It's a 22% from the 28 yd line (where the kickoff would be expected). Those seem to make sense to me. Or did I screw that up?

8. Brian Burke says:

"Should be odds of FG + odds of TD, obvs"

Agreed. That's exactly what I did.

9. Kulko says:

Well Headcoaches shouldnt be expected to have the math in their head, but the should be expected to have somebody with an PDA and a spreadsheet or an small application who can provide the numbers in 2 Secs. We are living in the 21st century after all. And if he guts say no he can still overrule it.

Anyway great to read, that at least some coaches understand how winning really works.

10. Jeff Clarke says:

Thanks for writing this.

I'm a lifelong Redskins fan that grew up in DC and moved to Seattle. When the Skins picked Zorn, all my new neighbors said how great he was. I was a little underwhelmed. I wasn't particularly impressed last year either. He seemed to have a habit of overconservatism. But between the fake fg and the two 4th down moves, he has proven he has the guts to be aggressive when he needs to be.

I'm a big Jim Zorn fan now. I just felt sorry for him that he was likely to be blamed if they lost.

Dave

"Zorn's bigger issue, though, is play calling and execution in the red zone...Redskins experience during the Zorn era suggests he needs to take easy points on 4th down in the red zone."

There have been several other studies done on this. Look at Brian's article: "Is red zone performance real?" for one example. The problem with red zone stats is that they are very small sample size. Historically, future red zone performance is more linked with past overall performance than it is with past red zone performance. If you hear that a team is particularly good or bad in the red zone, you should be prepared for significant regression to the mean.

Even if you believe the Skins would be below league average in converting here, they should still go for it. The benefits significantly outweigh the costs.

Bottom line is that the team that drives 60 yards to set up the winning FG would probably have driven 70 yards for the winning TD.

11. Anonymous says:

I take it as he's coaching as if he has nothing to lose; as if he knows big Dan will go after Cowher or Shanny regardless.

12. James says:

Don't know if you saw, but you got credited for your 4th down conversion work by TMQ on ESPN's Page 2.

13. Anonymous says:

this is a great way to look at the situation. But i feel your info was misleading and could be skewed, here is why, the rams had no tomouts, and the redskins were up by 2. In my opinion there is a less of a chance that a team with no timeouts can win with a td instead of a fg espesially since that FG would give them the lead. i dont care what study you do if you are down by two a field goal beats you!! and if you argue it is easier to score a TD; then a FG with no timeouts then your just plain worng obviously lol hence the 6 points for td and 3 for a FG. so all in all i think it was a bad deciocion still and even if he made the first down i would say we were lucky and it was was a dumb call. becasue we have a 100% chance to lose if they score a FG which would be easier to score then a TD with no timeouts if im wrong prove it!!

14. Anonymous says:

I just stumbled upon this site looking for a breakdown of penalties by type/ by player etc... Great site!

Can I find this info anywhere on the internet?

RW

15. RamsHerd says:

Great read. I had used your previous work on fourth down to analyze the opportunities that both the Rams and Redskins faced. I agreed with both of these decisions by Zorn during the game, and still lament Spagnuolo's decision to punt on 4th and 2 from the Washington 41, down only two points.

Your math makes clear that these decisions could be justified, but there is also a calculus of momentum and the matchup strengths that each team had. Pinning the Rams on the 4 with only a minute thirty left was as good as game over, as poor as this team is in the vertical game.

16. Anonymous says:

hi brian, 28% of scoring from your 20 and 22% from the 28 in your comments? Maybe that should be reversed.

Also, in the final two mins of the game the odds go up by almost 30% of getting a score by moving from your own 20 to your own 28? That looks very high to me. Yeah, 8 yds are worth something...but *that* important?

nice work overall,

17. Brian Burke says:

Sorry, I'm not being clear. I think I meant:

a. 20% shot of scoring a FG or TD from the 2 yl
b. 28% to score FG or TD from the 20 yl
c. 22% to score a TD from the 28 yl

(a.) is required if the conversion attempt fails
(b.) is required if the FG misses
(c.) is required if the FG is good

Does that make sense?

18. Anonymous says:

19. Jeff Clarke says:

"In my opinion there is a less of a chance that a team with no timeouts can win with a td instead of a fg espesially since that FG would give them the lead."

I think you are missing the point. Of course, a TD is more difficult, but the main reason why a TD is more difficult is that you have to go an extra 25-30 yards past FG range to get it.

But by pinning them deep, you are moving them farther an extra 25 yards away from FG range.

Don't think of it as the likelihood of a TD or FG. Think of it as the likelihood of moving the ball 65 or 70 yards.

Also, the timeout issue actually does have an impact and makes going an even better point. Assume the Rams have already moved the ball 40 yards. They are either 25 yards out of the end zone or 25 yards out of FG range with 15 seconds to play. If they are throwing to the end zone, they can use the whole end zone without worrying about the clock. If they are throwing to FG range, they can only throw to the sidelines.

20. Tarr says:

Good stuff. I was just trying to explain this to my fellow Redskins fans in fantasy league, and sure enough, you had already made the exact argument I was driving at.

Shorter version:

If you fail to score the TD, or kick a field goal, the Rams have to drive about the same distance either way.

If you score the TD, the game is over.

Therefore, try to score the TD.

21. Anonymous says:

Basically correct. Fail on 4th down, the Rams need to go about 60 yards to have a shot. Kick a FG and get an average return and they need to go 70 yards for a TD.

But a Skins first down or TD ends the game. Take the 50/50 chance to end the game. If you fail, you're not much different than if you succeeded with a FG.

Also of note, being backed up meant Bulger was throwing from his end zone -- and that could have resulted in a safety with either a sack, grounding or a holding call ... another way to win.

Lastly, kickoffs are highly variable plays, where lots of things can go right or wrong. A good return, an out of bounds kickoff or a facemask on the tackle could all have given the Rams the type of field position where only 25 yards of offense would have been needed.

22. Anonymous says:

All this probability and fancy math talk sounds real good. There's only one problem, this team shouldn't have been in this predicament late in the game. This game should have been over by the 3rd QTR by a wide margin. We have severe problems on this team and Jim Zorn's coaching is one of the main ones.

23. Anonymous says:

I agree with the statistical analysis about the decision to go for it on fourth down, but then you have to call the right play for that situation, and execute it.

24. Anonymous says:

Most of the critism that Zorn is getting in Washington is not because he went for it on the second 4th down, but because of the play he called. It was an obvious and slowly developing run, that hadn't worked all day. If you are going to go for it on 4th, you have to pick a play that actually has a chance of success.

Although, in this scenario their WP still says to go for it even if the play you pick is to just kneel down. By that point in the game they should have had a WP of 1.00, if the playing calling throughout wasn't so awful.

Whoever commented that they are a big Jim Zorn fan obviously hasn't watched his play calling. Oh well, maybe Snyder will fire him after they lose to Detroit this week.

25. Anonymous says:

"Head coaches like Zorn can’t be expected to do all kinds of math to figure out what to do on 4th down."

True, but they can have staff to do that fairly quickly, right? The Redskins employ a man named Chris Meidt who used a statistical approach as a college coach. There was a Washington Post article about him when Zorn hired him. I wouldn't be surprised if Zorn asked Meidt about it during the timeout.

26. Anonymous says:

Wow, I love this site. This is a great article, however, your mathematical calculations do not apply here. Your formulas and equations don't factor in the fact that you are dealing with the Redskins. The Redskins are a team that is known for constantly pulling a "Plexico" (or two) during every game jeopardizing their chances to win. You need a Redskins theorem, which factors in the amount of let downs, dropped balls, stupid penalties, bonehead mistakes, missed blocks, shoulda coulda's and just plain ole lack of effort. I guarantee your WPs will change a bit along with the coach's trust in his players to actually execute the plays correctly. (Hence Joe Gibbs' extra conservative play calling) I definitely respect the mathematical approach to the game, but I've been a Skins fan forever and the last 10 years have taught me that when dealing with the Skins it's probably around 85% accurate to expect the team to snatch defeat from the jaws of victory, especially when the game is riding on a 4th and 2.

BTW, I'm sure you'll have a 9 out a 10 chance of finding 10(miserable Dan Snyder)years worth of evidence to support my theory.

27. Anonymous says:

While the statistical analysis is great. I think you have to take into account real world situations. If you are playing the Saints or Colts for example, going for it is a no brainer. Against the Rams, who hadn't moved the ball well all day, I think the choice is to kick. I would consider the Rams horrible offense to be an "overwhelming" reason to kick. I would want to make them drive the field for a TD with no timeouts. Like you said, the end of games is where the stats can break down and situational analysis is much more important.

Blackwb

28. William T says:

I really think the Dolphins should have gone for it at the end of the second quarter on Monday night. They had 4th and 2, which was almost a lock given Indianapolis' rush defense. Instead, they kicked a FG and Manning led a drive with 47 seconds left to tie the game.

29. Jeff Clarke says:

Blackwb,

Why does it matter how good the other team's offense is?

Brian calculated that its 22% for the average team to score a td from their own 28. Its 20% for an average team to score a td or fg from their own 2 yl. A really good team might be at 35%/32%. A really bad team might be at 12%/10%.

The bottom line is that whatever a team's odds are for going 70 yards for a TD, their odds are roughly the same for going 65 yards and successfully kicking a fg.

That isn't going to change based on the strength of the opposing offense. Whether their odds are good or bad, you should still want to cut those odds by 2/3 if you can.

30. Anonymous says:

In an overtime scenario, at what point does it become attractive to attempt the FG rather than go for it?

EdBed.

31. Anonymous says:

was this a CBS televised game? This particular 'network' has LOUSY broadcasters during nfl games...I bet collinsworth would have not said anything negative...

32. denis says:

"A made FG gives Washington a 5-point lead and a 0.78 WP, as the Rams would have about a 1 in 5 chance of scoring a TD."

Over the previous season and week one of 2009,Washington allowed 12 of the 69 drives they faced from their own kickoff to end with an opponent scoring a TD.That's around a 17% chance of a TD.However,if we add in the time constraint we see that only 2 of the 12 TDs were scored inside 2 minutes and the average drive time for a TD was over 4 and a half minutes.

The average starting point was their opponents 26yard line,so if we add in drives that Washington faced that started around the 26 yard line but didn't originate from a kick off we now have just over 100 drives.The numbers are similar.Now 15% end up with a TD (16 from 105),but still only two TDs are conceded within 2 minutes.

Here are the numbers for SL's offense.

They scored TDs on just 9% of kickoffs.On plays starting from around their own 24 yard line(the average distance of their kickoff returns) they scored 10 TDs from 118 drives.But again only two TDs were scored within 2 minutes and the average time taken to hit the end zone on all touchdown plays was 3 and a half minutes.

If you put these numbers together,the quoted 1 in 5 chance of SL scoring a TD given the time left looks a tad optimistic,I'd put it closer to 1 in 20 at best.

Great analysis:-).

33. Anonymous says:

Just found this site (really great), and the more I think about it the more obviously correct Zorn's second decision was. To summarize everything thats been said. By going for it you essentially trade 10 yards of field position for a very high chance to win the game outright. Each decision has about the same drawbacks, but going for it gives you a huge benefit.

The only legitimate issue I see is wether having both the FG and TD options makes it the job more difficult for the WAS defense. Though this is certainly debatable, its the only viewpoint which legitimizes any criticism of the decision to go for it