## End Game Decisions in the Jets-Chargers Game

With just over 2 minutes to go and only 1 timeout, Chargers head coach Norv Turner faced a tough call. After scoring a touchdown to cut the deficit to 3 points, he could attempt on onside kick, or he could kick away. Turner elected to go for the onside kick. Was it a good decision? What about Rex Ryan's decision to go for it on 4th down to seal the win?

Onside kicks, when expected, are successful just 20% of the time. A successful kick would have meant a first down at the Chargers' own 40 or so. This equates to a win probability (WP) of 0.26. A failed onside attempt gives the Jets a first down at the same field position, giving the Chargers a 0.07 WP. On net, an onside attempt is worth:

(0.20 * 0.26) + (0.80 * 0.07) = 0.11 WP

A conventional deep kick likely gives the Jets a first down at about their own 30 (28 is median, 33 is the average). This equates to a 0.14 WP for the Chargers. The deep kick seems to be the percentage call, especially if you have faith in your defense to stop the run. It's close, so either decision is defensible if there is a clear reason to prefer one over the other.

It may have worked out for the Turner, except that Jets head coach Rex Ryan played a wildcard. Following the unsuccessful onside kick, on 4th and 1 from the Chargers 29, Ryan went for the 1st down. 4th and 1s are converted over 70% of the time, but I suspect that the Chargers were going to sell out on the run, and the probability of conversion is closer to 65%, the probability for a 4th and goal at the 1.

A successful conversion means certain victory, a 1.00 WP. A failed attempt gives the Chargers' a first down at their own 29 with just over a minute to play. This equates to a 0.22 WP for the Chargers and a 0.78 WP for the Jets. On net, the conversion attempt is worth:

(0.65 *  1) + (0.35 * 0.78) = 0.92 WP

A punt would have likely put the Chargers somewhere inside their own 20. There are very few examples of such a situation, but my best numbers suggest the Chargers would have a 0.07 WP with such a long field and so little time remaining. That's a 0.93 WP for the Jets. It's a wash, particularly considering how uncertain the numbers are for the punt.

The other option Ryan had was a FG. A successful kick would have meant a 6-point lead, worth 0.88 WP for the Jets. We don't need to analyze it any further, because even an automatically good FG is worth less than the either the punt or going for the 1st down.

Sadly, I think most coaches would have tried the FG here. I could be wrong, but my guess is that coaches overestimate the benefit of being up by 6 points. All that does is force the opponent to go for a TD, which would win the game outright. Teams down by 3 nearly always play for the tie, and giving up a FG to tie the game still gives you a 50% shot at winning. Congratulations to Rex Ryan for making the right call.

The bottom line is that both decisions were very close calls, and both decisions are defensible. We can expect nearly all the criticism directed at Turner, since his team lost in the end. But could you imagine if the Jets had failed to convert, and the Chargers went on to tie or win?

### 37 Responses to “End Game Decisions in the Jets-Chargers Game”

1. Anonymous says:

With the best rushing attack in the league, it seemed to me at the time, the gamble was the best call.

2. Shawn says:

Not to mention the Jets were obviously expecting an onside kick so a conventional kick could easily result in the Jets having significantly worse field position than the league average.

Also, as far as nitpicks go, "conversion" is misspelled at the end of the 4th paragraph and "Chargers" has an unnecessary comma in the 5th paragraph.

3. Zach says:

I know the numbers don't take timeouts into account, but does the 2 minute warning factor in for the onside kick analysis?

4. Brian Burke says:

Shawn-Thanks. (I get paid by the comma.)

Zach-Yes 2-min warning is factored in.

5. Shawn says:

Whoops. I actually meant apostrophe, so meh.

6. Anonymous says:

Nice post. The numbers also rely on historical data, it seems, and don't take into account how the teams are playing on this day. In this case, the Chargers began to unravel early and played poorly through most of the game. Not sure how that can be quantified, though this factor may have influenced each coach's decision.

7. Brett says:

The 4th down decision would have been moot if the Chargers had kicked deep. With a 4th and 1 at their own 30ish yard line, the Jets definitely would have punted, and the Chargers would have had 50-60 seconds to go 35 yards to get into FG range.

I know hindsight is 20/20, but even at the time, right after the Chargers scored the TD, I considered the options, asked myself, "What would Brian Burke do?", and decided that the best option would be to kick it deep because all you have to do is stop 3 definite run plays and you get the ball back with a minute to play. That Phil Simms agreed with me caused considerable self-doubt, but I still would have done it that way.

Or perhaps Norville had so little confidence in his kicker at that point that he decided they were going to go for a TD or nothing at all. In that case, the onside kick is the better option.

8. Zach says:

I'll admit that at the time I thought the onside kick was the correct call--in either case, you have to make the Jets go 3 and out, so why not try to get the ball on the onside and skip the 3-and-out part?

9. Anonymous says:

I just started watching American football and I am confused about something. I was watching the highlights on NFL.com and with a minute to go the commentator says that Jets just need a first down to win. I dont get it. Surely it only takes a few seconds to win a game? Seems to me that the game ended a minute early. It wont have been the first time a team has grabbed the ball in the last minute to win a game.What am I missing?

10. Anonymous says:

Between plays, you're given 40 seconds to snap the ball. In that time, the clock will run (as long as the last play is not a touchdown, out of bounds, incomplete pass, etc.). Without any timeouts for SD, a minute left in the game, and a fresh set of 4 downs for the jets, the jets simply kneel after each snap leaving SD helpless to stop the clock.

As to what you mentioned, SD CAN grab the ball during these kneel downs, but this is very unlikely since they'd have to force a fumble on a team that is looking to avoid that and only that to win the game.

11. Robbie says:

Does anyone know if a team has ever fumbled in the victory formation? It's possible, but so unlikely that I wouldn't be surprised if it has never happened.

12. Anonymous says:

the media has been highly critical of the call this morning. on footballoutsiders they quoted a 10% success rate for onside kicks. on kjr am 950 Seattle they again quoted 10%. does anybody have an idea of where this 10% is coming from?

13. Billingham says:

What about the time out called after the two minute warning rather than before?

14. EdBed says:

Going back to the kick deep option: I know it is a string of "what ifs" but what are the WPs of going for it or punting if the Jets had a 4th & 1 after an average SD punt?

15. EdBed says:

Excuse me, average SD kickoff, not punt.

16. Anonymous says:

In the FG analysis, did you factor in a missed field goal and possession on the NYJ 38? Wo

17. Joe G says:

Interesting, I thought the same as Zach re:onsides kick, you need to stop jets either way, so why not give yourself the 20% chance to recover.

I wonder if there is really enough data built into the WP AFTER the jets recovered to account for the fact that in that part of the field, the leading team is much more likey to go for it on 4th and 1, thus making kicking deep even a little more desirable for the trailing team.

18. Brian Burke says:

I originally did the analysis based on an assumption SD would force a 3-and-out. It's absolutely required, so why even worry about the alternative? It makes the math simpler, but I thought I'd get a lot of cranky comments-- "How can you possibly ASSUME a 3-and-out?!!!"

Plus, as it turns out, the chances of forcing a 3-and-out were not equal between the deep kick and the onside kick. It's hard to imagine Ryan going for it on 4th down and 1 on his own side of the field.

And Joe is right, there is very little data. In fact, no team has ever lost up by 3 with a 1st down past midfield and 2 minutes to go, but the real answer is not a zero probability. I had to dig into things like conversion probabilities and then look at how often a team down by 3 will score from various field positions and time remaining, and interpolate. The WP model that runs automatically to make the graph doesn't do all that on its own.

In the end, there is a lot of uncertainty, even with the luxury of time and data. I really can't fault Norv.

Also, I've been hearing a couple of citations for 10% recovery rates for expected onside kicks. Not sure why I get a different rate. The rules are different lately, so it's tough to tell. (No do overs, no overloading to one side of the kicker.) If it really is 10%, that would change the analysis solidly in favor of the deep kick.

19. Brett says:

Something else we haven't factored into the 4th down decision is the increased likelihood of a blocked punt in that situation. Surely the Chargers would have brought the house if the Jets tried to punt there, and they had already come very close to blocking several punts earlier in the game. In fact, I believe Jacob Hester got a piece of the last one.

20. Anonymous says:

The analysis for Onsides Kick fails to take one thing into account:
There is no way that the Jets go for it on 4-1 from their own 40. So, the onsides kick gave the Jets the ability to close the game out by going for it on 4-1. If you kick deep, you almost certainly get the ball back, because the Jets are going to be conservative, so all you need to do is stop three running plays. If you onsides kick, you need to stop four running plays.

21. Ian Simcox says:

"But could you imagine if the Jets had failed to convert, and the Chargers went on to tie or win?"

Yep, and it would be the same cast of know-it-all commentators who said Belichick was wrong to go for it against the Colts.

Good thing they won. Shame it doesn't get the same amount of coverage as the failed attempt though.

22. Jeff Clarke says:

I have to admit that I thought that an onside kick was clearly the best option. I guess I was wrong. I'm curious as to exactly what the odds are on getting a first down in an obvious 3 run situation.

The odds of a 3-out are considerably less than 50% in a normal situation. The obvious play calls do increase the odds, but I would think the odds would still be less than 50% even in this situation. I think this is especially true with a power running team like the Jets.

I'd also point out that teams don't have to run straight up the middle in this situation. The Chargers are definitely going to be overloading the box. A play action bootleg or screen pass are both high percentage plays that could be used effectively.

Bottom line is that the Chargers had very low WP before the kickoff. A deep kick might have worked. An onside kick might have worked. Neither was particularly likely to win the game.

Its easy to second guess, but I can see quite a few Chargers fans being annoyed if they kicked deep and the Jets got a quick first down.

Secondary issue:

Did you notice how little analysis is being given to Ryan's fourth down call?

There is no doubt in my mind that if the Chargers stopped them and won the game, everybody would be talking about Ryan and nobody would be talking about Turner.

The thing with "gambles" is that nobody seems to remember the ones you win nearly as vividly as the ones you lose. Your mind does play tricks on you. If you are working solely off of memory of previous games (as I believe most announcers and even a few coaches are), selective memory is going to screw up your analysis.

23. Buzz says:

I too wonder if the % of recovering onside kicks has changed (no second chances, number of players allowed on each side, etc) enough that the more recent data, despite the smaller sample size is a better data point. I would imagine that in the past there were several times where a team would kick multiple times to get another chance and it wouldn't surprise me at all if all those multiple chances really averaged approximately one per possesion.

what type of percent recovery is the average for this year?

24. Anonymous says:

Did you include the probability of a receiving
team fumbling and losing a deep punt?

25. Brian Burke says:

Buzz-Good question. Out of all 49 onside attempts this year, 23% were recovered. Of the 8 unexpected attempts (defined as WP>.20), only 1 was recovered. The sample is just too small to tell.

Ron-Regrettably, no. Blocked punts aren't factored in either. Both possibilities would have marginal, but countervailing effects.

26. Anonymous says:

The onsides kick is even worse for another reason. On third and long, there was at least a possibility that the Jets would throw a pass if back on their own 20. An incomplete there would essentially give SD another time out.

The simple analysis is that they needed to hold the Jets to win. If they held after a regular kickoff, they were likely to get the punt around their own 40 with 1:10 on the clock. They traded 1 minute of clock for a small chance of getting the ball.

What is the WP difference for 2:10 left vs. 1:10 left from your own 40, down by 3?

27. Shawn says:

Assuming 1st and 10, going by the wp calculator the difference is 0.06 (.28 to .22)

28. Jeff Clarke says:

Brian,

Can you give a little more detail on how you got to the various WPs given?

In particular, I'm not really sure about the 0.26 for if the onside kick is successful. That seems really low to me. I'm guessing that you are saying that there is a decent chance that SD gets into FG range, makes the kick and the Jets drive themselves to another FG before OT. Is that correct? I think the probability is greater that they score a TD on that drive and win outright. What do you have as the component probabilities for the various events? No score, SD fg + OT, SD fg + Jets last sec FG, SD TD?

Also, if they kick away, they have a 14% prob of winning the game. If they force a 3-out and get the ball at roughly the same field position, how is the prob affected by the fact that SD now has considerably less time (and the Jets have less time if they do get a FG).

I would think that SD's WP from their own 40 would be considerably less with 1:00 and 0 timeouts than it was with 2:13 and a timeout. If its 26% with plenty of time, wouldn't it be max of about 21% with less time. But if it was 21% that they win after they get the ball back then that implies that there is a 2/3 probability that they will force a 3-out. 21% * .67 = 14% overall WP.

I know that the Jets predictable playcalling would increase the possibility of forcing a 3-out, but only 33% of NFL drives result in 3-outs. The Jets are a power rushing team. It seems wrong to use a 67% assumption on a forced punt in this situation.

But if we don't use the 67% 3-out number, either the 14% or the 26% numbers must be wrong.

I know you are overworked and I don't mean to be demanding, but can you give a little more detail on the component probabilities you used.

29. Zach says:

Jeff, SD was down by three when they attempted the onside kick. Brian's WP calculator gives them a 22% chance to score a TD if they recover the onside kick and a 14% chance to kick a FG. If they have a 50-50 shot in OT, then their total WP is equal to 22% + 1/2 * 14% = 29%, which is about what Brian's calculations give them of winning.

30. Jeff Clarke says:

I went to the WP calculator and typed in SD has the ball on their own 40 yard line with a minute to go (where they would be following a successful deep kick...3-out...punt). I got 20%.

So in order for us to get to 14% overall deep kick probability, we need to have a 70% chance of a 3-out.

0.2 x = 0.14
x = .7

I suppose you can throw in a couple of very rare ways of winning outside of forcing a 3-out (a fumble on the kickoff, a Bettis situation, etc).

But even if you use 12% as the overall WP number, you still need to have a 60% shot at forcing a 3-out.

I guess this just brings up a stat that I've meant to look up but never bothered to.

How often do teams force a 3-out when they absolutely need one?

The overall NFL 3-out percentage is only 33%. Yes, it simplifies things for the defense that 3 straight rushes are expected. You still have the #1 rushing offense going up against the #20 rushing defense.

Its going to take some convincing to get me to believe 60% is the appropriate number for the 3-out.

The larger point is that San Diego was screwed here no matter which choice they picked. The odds were the onside wasn't going to work. The odds were that the forcing a 3-out wasn't going to work.

Turner was faced with making a pick, watching it fail and then hearing everybody say "It was obvious that was going to fail...you should have went the other way".

31. Brian Burke says:

Good question, Jeff. When a team is certain not to risk an int or sack/fumble, how likely is the 3-and-out?

Typical conversion rate for a standard 1st and 10 is 67%, so I'd guess it's close to or below 50%. The 60% probability of a 3-and-out makes some sense.

The WP Calculator still uses my older model. I haven't upgraded it to use the new end-game model. I'm only halfway done, but I did have -3 points finished.

Yeah, Norv was in a pickle. I would like to hear him explain his decision though. Not out of 'gotcha' second-guessing, but out of genuine curiosity how these coaches really make decisions. Usually we just get the "I thought we could recover the kick," or "I thought we'd be able to get the yard," explanation.

32. Anonymous says:

Why would SD have only a 50% chance of winning in overtime? Does home field advantage completely disappear? I would also have to think that their odds of even attempting a field goal were a lot worse than normal.

33. Brian Burke says:

Yes. HFA disappears in OT.

34. Jeff Clarke says:

Thanks Brian.

I'd like to hear their explanations as well. This problem exists in politics as well. The decision maker is aware of the fact that he is talking to reporters and that reporters want complex decisions summed up in sound bites. The problem is that some logic can't be summed up in a sound bite. So they say something simplistic even when they know it doesn't tell the whole story. Other people then take those simplistic statements at face value and extrapolate them in the wrong situations.

I get the impression that some coaches really believe the cliches and others just say them because they don't want reporters to misquote a longer explanation. The problem is that it is kind of difficult to know which ones are which.

35. Anonymous says:

I have ABSOLUTELY no doubt that Rex Ryan would have punted if it was 4th and 1 at the Jet's 29. If you factor in the probability of a blocked punt or punt return for a TD the WP has to be higher. It was a terrible decision not to kick the ball out of the back of the endzone. Plus you have have the "letdown" factor when you don't recover the onside kick. Momentum was CLEARLY on SD's side after the touchdown and the defense proved they could stop 3 straight running plays. They lost the crowd and all that momentum when the onside kick was lost. Terrible call Norm and I really hope it costs him his job.

36. Anonymous says:

Does anyone have numbers on onside kick success rate this year vs. before the rule changes? Seems like it's SO much harder to recover now.

37. Brian Burke says:

Sure (See my 3rd comment above.) "Out of all 49 onside attempts this year, 23% were recovered. Of the 8 unexpected attempts (defined as WP>.20), only 1 was recovered. The sample is just too small to tell."

So 23% is not out of line with recent history, which was 26% overall. We'll need a couple more years of data to tell if there is a real difference.