Extra Points
News and commentary from around the Web

Risky Business: Week 5

David Montgomery
Photo: USA Today Sports Images

Thursday night's Tampa Bay-Chicago game featured a number of questionable coaching decisions in the final minutes as the two teams traded Game-Winning Chance back and forth like a hot potato. The most interesting might be Tampa's decision to punt on fourth-and-12 from their own 14. It doesn't sound like a situation where Tampa should go for it, but Chicago's Game-Winning Chance was almost the same whether the Bucs punted or turned it over on downs! More analysis at the link below.

View Full Article


12 comments, Last at 13 Oct 2020, 3:33pm

1 4. (2:21) Bucs punt on 4th…

4. (2:21) Bucs punt on 4th and 12 from their own 14-yard line, LEADING 19-17.

  • Punting costs the Bucs 12.1% GWC! In case you thought the Bears’ punt moments earlier was a surprising blunder, how about this one?

The needed conversion rate math isn't the surprising part here. It's that failure to convert only reduces down to 37.9%. That's the number that needs to be defended.

3 I can see why this works in…

I can see why this works in the case when you have a less than a FG lead, and especially with a 2 point lead (as that almost eliminates overtime from the equation).

You opponent cannot get a more than one score lead against you, and if you turn the ball over on downs inside your own 20 they are most likely going to score fairly quickly, giving you a second chance for a game winning drive (maybe the current drive, if it continues, is more properly called a game icing drive?). If you punt from inside your own 20 most NFL teams under most weather conditions would probably fancy their chances of at least getting a shot at a FG, and they might take a lot of the remaining clock doing it.

It is an interesting bit of theory, and it does make sense mathematically, but it would take a really brave coach to go all in under such circumstances. It is a similar approach to the letting your opponent score in to have more time for your last drive, except you have a chance of putting the game away with your hands on the ball.

4 I disagree with the math in…

I disagree with the math in this case, due to the way the game had played out, which was that both offenses were struggling, with the exception of New England's run game.

If Tampa goes for it, they'll be trying to convert a 4th and 12 against a pass defense that was pretty good against them. If they fail, they're practically handing Chicago 3 points and the lead, then having to drive half the field without getting a chance to run the ball. And even if Tampa succeeds on 4th down, Chicago still had two timeouts and the two minute warning, which means that if they can prevent Tampa from getting another first down, Chicago will get the ball back with around 2 minutes left.

On the other hand, if they punt it away, Chicago will need about 20 yards to have good FG chances, and their offense had been struggling even more.

If both teams had been good on offense, going for it would have made more sense.

(I had written New England instead of Tampa in my first draft of this. Still not used to Brady being on a different team.)

10 Agree 100%, going for it did…

Agree 100%, going for it did not seem like the right call at all. Interesting, though, how little failing on 4th down would cost them in GWC, but Tim's explanation above makes a lot of sense.

Still, this is the problem, with GWC we're talking averages, not specific game performance, teams, injuries, etc.

5 This is an interesting…

This is an interesting aspect of time management.  Basically, the Bucs wanted to do one of four things:

1. run out the clock themselves;

2. score again;

3. prevent the Bears from scoring again; or,

4. make sure they have time to respond if the Bears score again.

The numbers here say go for it (using historical #'s, and ignoring the observation about how well each O was or wasn't doing in this particular game), because the punt eliminates #1 and #2 as possibilities, doesn't significantly change #3, and makes #4 less likely.  Going for it and failing also eliminates #1 and #2, pretty much eliminates #3, but makes #4 more likely.  Going for it and succeeding doesn't guarantee anything, either, but all of #1, #2, and #3 become a lot more likely (while #4 becomes a lot less likely).

The way the remaining time left in the game weights the value of each of these outcomes compared to the same choices earlier in the game moves this out of normal coaching decisions and into a specialized category.  Or to put it in other terms, all football is situational.


6 Also, do those historical…

Also, do those historical numbers take timeouts into consideration? I'm pretty sure they don't, especially since old box scores don't even record timeouts. So the fact that both teams had two timeouts left (TB had all 3) will definitely change the probabilites.

2 (1:28) On the Bears’ last…

(1:28) On the Bears’ last offensive drive, they chose to pass on 2nd and 9 from the Bucs’ 24-yard line while already well into field goal territory. An incompletion stopped the clock, and the Bears successfully kicked what was to be the winning field goal moments later. However, this clock management snafu gifted Tom Brady a solid 1:13 to attempt a game-winning drive.

No, this is the wrong analysis. The decision tree to evaluate is:

1. Run on 2nd-9.
2. Completed pass on 2nd-9 for new set of downs.
3. Incomplete pass.

The advantage of the outcome was never even under debate. What was debated was the decision, but this writeup is begging the question instead of providing insight into that debate. This is just clickbait.

9 It does seem like the…

It does seem like the EdjSports article looked at that decision solely based on its outcome (an incomplete pass) and not on its potential outcomes, like everywhere else in the article.  Ignoring the Bears turning the ball over and missing the FG (which is never a safe assumption for the Bears, but those events don't really change the calculus for the decision), we can assume if they completed the pass for the first down, they could kneel three times and kick the field goal with 0:00 on the clock.  So looking at the three states:

A. Tampa Bay begins final drive with 1:13 remaining from their own 25 (incomplete pass then unsuccessful run on 3rd down): 42.3%
B. Tampa Bay begins final drive with 0:30 remaining from their own 25 (two unsuccessful runs): 13.7%.

C. Tampa Bay does not get a final drive: 0%.

A completion would gain 13.7% WP at the cost of 28.6% (42.3-13.7).  So for that trade-off to be worth it, the Bears would need a 68% chance of completing the pass for a first down (28.6 / 42.3) for the pass to be the right decision.

It would actually be a little less, since the FG is also more likely to be made from 9 yards closer, so let's be generous and say 60%.  And the pass play itself isn't a binary thing - an unsuccessful completion has the same effect as an unsuccessful run, and there is a higher likelihood of a turnover on a pass play as well.  You can maybe assume away a sack or turnover based on playcalling, so I think the pass is certainly defensible even though I think it's not accurate to say the Bears were 60% likely to gain a first down via the pass play.

8 Bucs risk (51.9 – 43.8) = 8…

Bucs risk (51.9 – 43.8) = 8.1% GWC to gain (67.1 – 51.9) = 15.2% GWC, for a required success rate of:


This math is just incorrect; 8.1/(15.2+8.1) = 34.7%, not 29.7%.

12 comment

So, does this mean the analytics were taking into effect that later in the game a QB was going to lose track of downs?