There were a plethora of interesting coaching decisions this week, especially in the early games on Sunday. Avid readers cringe at these conservative calls so let's look at a few of them a little closer.

**Cleveland Browns: Punted on 4th-and-1 on the Indy 41-yard line down by four with a little over six minutes remaining.**

**
** On what was likely the worst decision of the day, the Browns cut their expected win probability almost in half by punting. On 4th-and-1 from the opponent's 41, you should almost always go for it. You are expected to score +1.58 points by going for it (74% league-wide conversion rate) versus -0.04 points by punting.

But, when you're losing in the 4th quarter, that decision becomes even more obvious. A punt means only a 24% chance of winning the game -- only 6% more than a failed conversion. This is because a punt from that spot nets you a maximum of 40 yards and more likely closer to 20. A successful conversion means a 45% chance of winning. The numbers say to go for it unless you think you can convert fewer than one in four.

The Browns would actually get the ball back, as Pat Shurmur had planned, but they would ultimately fall to the Colts. The fact that they got the ball back does

*not*mean it was the correct decision.

**Washington Redskins: Attempted a 44-yard field goal down seven on 4th-and-5 from the Giants' 27-yard line with about seven minutes remaining.**

**Not as absurd a decision, but a field goal converts at 68% and 4th-and-5 converts at 49%. Successful conversion means 31% chance to win the game while a made field goal only increases win probability from 19% to 21%. That means the break-even point is if you believe you can convert above 40% of the time. With the offensive weapons available for Kyle Shanahan and RGIII, the Skins should have gone for it.**

The Redskins would stop the Giants and score the go-ahead touchdown, only to have Eli complete a 77-yard bomb to Victor Cruz. Again, the fact that the Skins ended up ahead of the Giants (briefly), does not make kicking the field goal the correct decision. Given that we are using league-wide baselines, though, this was not nearly as egregious an error.

**Baltimore Ravens: Attempted a 54-yard field goal down 26 to the Texans on 4th-and-27 with about 13 minutes remaining.**

**Potentially a negligible decision, going for it, punting and kicking a field goal all result in ~1% chance of winning. A successful conversion, however, results in a 2% chance of winning. As an underdog, it is extremely important to play high-variance strategies. Attempting a 54-yard field goal to change the deficit from 26 to 23 is the opposite of high-variance. For the record, the 4th-and-27 converts about 10% of the time.**

**Dallas Cowboys: Ran a dive on 3rd-and-9 from the Panthers' 15-yard line down one with 3:28 left.**

**This situation gives the Cowboys about a 60% chance to win the game. They have a chip shot field goal to take the lead. But, in today's NFL with amazingly accurate kickers, leading by a field goal or less when your opponent has time left on the clock does not mean much. Rather than try to convert on 3rd-and-9 and score a go-ahead touchdown, the Cowboys were content with a two point lead -- even though Cam Newton would have over three minutes to lead a field-goal drive.**

If the Cowboys were able to convert, their win probability would jump to 68%. A touchdown means a 77-81% chance of winning depending on the two-point conversion. Why run the ball not hoping to convert, especially when there is a 90% chance to make the field goal as is?

These were the one's I happened to notice, I'm sure there were others too. I do give credit to Mike Shanahan for going for it on 4th-and-2 from the Giants' 48 and 4th-and-1 from the Giants' 30 during the game. It just gets tough to watch coaches continue to shoot themselves in the feet based on groupthink mentality.

*Some notes from the Jaguars/Raiders game:*

*Down seven to the Jaguars with 3:30 left in the game, the Raiders trotted out their field goal team on 4th-and-10 from the Jaguars' 26-yard line. The Raiders then burned a time out and decided to go for it instead. My initial reaction was, "Yes, finally!" -- a field goal is pretty much meaningless and the Raiders had just burned a timeout. The numbers actually say the field goal is the better option. 4th-and-10 converts at 35% which leaves expected win probability at 12% going for it versus 16% kicking the field goal. Palmer threw a ball to the end zone where the Jags were called for pass interference -- another plus of passing the ball.*

Then, with 1:50 left, the Jaguars punted on 4th-and-4 from the Raiders' 49-yard line in a tie game. This is a clear example of playing "not to lose" instead of playing to win. Especially for a team like the Jaguars that has very little to lose, going for it there should be the call. Yes, they had both a backup QB and running back in the game due to injuries and had only gained 7 yards on offense in the second half up to that point. The break-even point, though, is if you believe you can convert one out of every three times -- 20% less than the league-average 53% success rate on 4th-and-4.

*Keith Goldner is the creator of Drive-By Football, and Chief Analyst at numberFire.com - The leading fantasy sports analytics platform. Follow him on twitter @drivebyfootball or check out numberFire on Facebook.*

Mike McCarthy punted on fourth and one from midfield and fourth and one from the Ram's 41 and doesn't make the list?!? I don't understand how the same person frequently makes bold calls like surprise onside kicks and fake field goals on fourth and 26 and then punts on fourth and one from his opponent's 41.

It's OK. Jimmy Haslam looked pissed at Shurmur's decision. Not a great way to impress your new boss...

FG attempt from the 37 likely converts higher than 68% in 2012 NFL.

sorry..meant ... FG attempt from the *27*

Final play of third quarter, BUF scores to go up 33-28, kicks XP to go up 34-28. Naturally, they lose 35-34.

According to my numbers, it appears that the Browns have not been very good at converting 3d or 4th and one this season.

I show them converting 7/14 attempts, with one turnover, an interception that produced a 40 yard or so change in field position.

I lump 3d and 4th down together for two reasons. One, to increase the sample size, and two, because I view the situations much the same. It is assumed in the vast majority of situations that failure to convert 3d and 1 means relinquishing possession of the ball, so, I would think the success rates would be comparable.

In fact, it appears to me the Browns' opponents have succeed in 12/13 such situations against them.

If, in fact, we can only convert half the time, how does that change the 'WP'. Did you really mean to say that we only have to convert 25% of the time to make going for it preferable to punting? As I read your chart (first time reader) it appears that punting is better 25% of the time, and I know that not all 4th down conversions lead to points.

I apologize if I am confusing your terminology.

George, if you convert at 50%, going for it expected win probability is .5*.45 + .5*.17 = 0.31 > punt expected win probability of 0.24. And yes, 25% conversino is the break even point: .25*.45 + .75*.17 = 0.24

Also curious about the WPA for the Bills kicking extra points instead of going for the 2 pt conversion on both of their 2nd half touchdowns (first was to tie instead of trail by 1, second to take a 6 pt vs a 7 pt lead).

I realize the emphasis here is pure hard numbers in stats and probability, which for the record I think is a much better idea than strictly coaching according to established wisdom, however I don't think you're giving the human factor enough credit. All probability models tend to have outliers after all, sometimes not off the baseline enough to give serious consideration to, but sometimes they are really out there and in football teams have varying odds due to personnel and overall coaching strategy. The easiest example is which squad, offense or defense, is more efficient and by how much. Look at the Skins/Giants, the Skins secondary is not a strength, they are way too inconsistent to be counted on. On the other side, the offense has quickly turned into an offense that can make plays when they matter. In a vacuum it looks as if going for has an even bigger probability of being the right call, but you have to consider what kinds of pressure it would have put on the Skins defense, a defense than inexcusably let a WR with the big play ability of Victor Cruz get past the secondary playing with a slim lead with under two minutes left. As a long time Saints fan I can tell you that a weakness of Jim Haslett is getting sloppy undisciplined play from his players. It gets written off as being a by-product of an aggressive style, but it most certainly is not. Historically players under Haz play sloppy and it isn't hard to see things like missed assignments and pentalies that show a lack of concentration or are flat out avoidable boneheaded mistakes that shouldn't be seen on a consistent basis in the pros.

Regardless of any of that, I do believe numbers don't lie and successful teams are the ones who play the odds. I just think the human factor has enough of an empact on those odds that it's not always a one size fits all equation. Drew Brees and the Saints offense has a much greater shot at converting a 4th and long than an offense like Jacksonville or a QB like Mark Sanchez after all. And occasionally pulling out a totally batshit insane call like an onside kick with the second half kickoff in the Superbowl turns out to not be totally batshit if you have the element of surprise at your back.

I'd like to see a breakdown of the Jets decision to kick a FG down 16-10 from the 3 yard line on 4th and 2.

Great site! Thanks.

"Dallas Cowboys: Ran a Dive on 3rd and 9 from the Panthers 15-yard line down 1 with 3:28 left."

Uh, no they didn't. They ran power. They get a double-team on the frontside and they pull the backside guard. It's a base play in every running game.

How do you know what play Jason Garrett, or any other coach called? You don't. On most teams, especially those with a veteran quarterback, the decision at the line of scrimmage is based on what the defense is showing. Here, this was Tony Romo's call. He saw the defense in a 4-2-5 Nickel with the linebackers backing out before the snap. Romo's decision was correct. If the Panthers had shown blitz, he probably would've checked to the pass.

What matters are the variables in this particular game, not what other teams did on different days, in different stadiums, and with different players and coaches. In football, every situation is different. Similar, yes, but they're never the same.

JBill - If you run the numbers through the WP calculator, with 15 minutes to go, it's about a wash going for two versus kicking the XP (assuming that 2 point conversions succeed about 50% of the time).

I don't think you can fault the Bills on that one.

Keith, you forgot the worst blunder of the day. In the 4th quarter down by 7 with 4:15 left in the game, the Bengals had a 4th and 10 from their own 11, and they decided to punt the ball. Based on my math, that was a horrible decision because the breakeven at that point is like 20% while the success rate for 4th and 10 is like double. Do you have the numbers on that Keith?

The play calls the first and second down plays before the punt two running plays. Really looked like they were playing for the punt instead of taking shots at the first down.

ff -

I wrote this before that game! But you're right. I actually have the break-even point as 8% conversion rate while the actual conversion rate is 35%. Failing on 4th and 10 from your own 11 leaves a 4% chance of winning versus a 5% chance of winning if you punt.

Anon -

Jets FG decision was definitely the wrong won. It was in the 3rd quarter:

E[Points] - Go for it is +3.49, FG is +2.38 so breakeven point is 35% conversion rate (actual conversion rate is 55%)

E[Win Probability] - Go For it is 39%, FG is 35% so breakeven point is 38% conversion

Bill's decision to kick XP -

Kick XP, win probability is 79%

Successful 2 PT = 82%

Unsuccessful 2 PT = 74%

So, if we assume 50%, E[Win probability for 2 PT] = 78% win probability, basically break even, in fact the actual conversion rate is under 50% so the XP was slightly the better decision. Again, basically a toss up though as Mike mentioned.

Backing up the other Anon here, Romo was sent to the line with an option to either run or pass based on his pre-snap read of the defense. He correctly read the defense and called the run. If you watch the film, the 3rd string running back, Tanner, should have cut back, but failed to. If he had, or had Murray or Jones been in the game, the Cowboys may have converted.

The run was called in an attempt to get the conversion, not as a give-up play. Haven't there been other posts on this site that indicate teams don't run on 3rd down often enough?

P.S. As a Cowboys fan, I am not happy with Jason Garrett. But that run was not one of his many stupid decisions (e.g. opting to punt from Baltimore's 35).

Hey Keith, you forgot another decision. The Saints faced a similar situation that Belichick had before. They were up 7 points with 2 minutes left and had a 4th and 2 at their own 40. They went with the draw the opponent offsides then call a timeout move, then punted. I assume that is the wrong decision because Belichick went for it at the 30 while the Saints were at the 40 in the same situation (plus they were up 7 while the Patriots were up 6).

Yes, numbers say for Saints to go for it, but not dramatically. E[Win Prob] going for it is 91% and E[Win Prob] punting is 88%. Break even is if you can convert 42% of the time versus a league-average 60%. Given the Saints offensive prowess, the better decision was likely to go for it.

Thanks for the reply.

But I am still wrestling with something here.

Once the 1st down is gained, we still needed to score at least a TD to win the game. That isn't going to happen all the time, so how can we still win 25% of the time if we can only convert 25% of our 4th and one attempts?

To answer my own question, I would say that there was still 6:30 left in the game, enough time to stop them, get the ball back and score a TD.

So, my next question is:

Over the last 6:30 of a game, what are the probabilities of scoring at least 5 more points than your opponent scores during that time frame if they have 1st and 10 at their own 41 yard line.

Same question if we have the ball 1st and 10 at their 40 yard line?

"the fact that the Skins ended up ahead of the Giants (briefly), does not make kicking the field goal the correct decision."

um, yes it does. It really really does.

expectation values are irrelevant when talking about a single realization, he had as you say a 40% chance of success, and he got it. that is the correct decision, based on an absolute ton of local and relevant information. The NFL is not a perfect random number generator.

Anon -

No, it doesn't. You should not evaluate a decision based on the outcome, but based on the process. If you could flip a coin one time and if it lands heads you win $10 but if it lands tails you lose $2, you're obviously going to flip it E[Flip] = +$4. So if you flip it and it lands tails, does that mean it was the incorrect decision? No, the process is more important than the outcome.