Crazy thought of day: Cowboys should have missed extra point on purpose with 2:32 left so they'd be up 3, not 4

Malmo (and others),

I am not here to defend Rojo's faulty conclusion. Just wanted to point out that his frame for this discussion was a "crazy" thought for the day.

I put this discussion in the category as the statistical discussions of why coaches should always go for a two-point conversion in late game scenarios. The math works; the football doesn't.

But, when someone approaches it as "crazy," they are admitting as much.

Mundus

I tried the calculator under a variety of "close and late" scenarios and the conventional wisdom generally gave a higher win probability.

It seems problematic to me that the calculator uses score differential as the input rather than actual current score. I suspect that whether a game is low-scoring or high-scoring impacts the probability. Going ahead 3-0 with two minutes left is very different from going up 52-49 with two minutes left, but the model data treats both situations as equivalent.

I would like to see the odds of winning being up 3 points late vs. the combination of the winning odds of being up 4, 5, or 6 points late. I think that the core argument is that you are saying being up 3 is statistically advantageous because of the complacency, not going for 4th down, etc., which would hold true whether or not being down 4, 5, or 6 points, and that including more data with the 5 or 6 point leads would give a better reflection of how often teams really are able to drive down for the touchdown.

I suppose using 6 point lead might not be fair because they could drive down, get the TD, and miss the extra point and then go to overtime, which wouldn't be the case in a 4 point lead, but since extra points are made more than 98% of the time, I'm not sure that it makes much difference.

Mundus Vult wrote:

Malmo (and others),

I am not here to defend Rojo's faulty conclusion. Just wanted to point out that his frame for this discussion was a "crazy" thought for the day.

I put this discussion in the category as the statistical discussions of why coaches should always go for a two-point conversion in late game scenarios. The math works; the football doesn't.

But, when someone approaches it as "crazy," they are admitting as much.

Mundus

The subtleties you described above are missed by people on this message board as soon as rojo attaches his name to it.

I thought it was a fairly clever post to get a discussion going, sure to draw pro and con arguments and as always, the possibility of moving to a completely different tangent.

If rojo would have used an anonymous name, the responses would have been completely different.

Older guy wrote:

If rojo would have used an anonymous name, the responses would have been completely different.

The world wants to be deceived.

Good discussion. I think unless the probability difference is overwhelming, coaches will continue to take the point.

Not to go all Gregg Easterbrook on you, but I get tired of announcers calling every 4th down try "risky", even when it's clear that it has a better expected value than punting.

Teams will let other teams score a touchdown when tied or down one point with about a minute to go to keep the other team from running out the clock for a field goal and giving themselves a little time for another shot. The Packers did it in the superbowl against the broncos (lost anyway). So there are some precedents to what you're saying.

rojo wrote:

Why? Because down 4, the other team will go for it onn 4th down all the way to the end of the game and if they score (a TD) you lose. Up 3, the offense won't go for it on 4th down once in field goal range. So even if the offense gets a drive going, they often get conservative and kick a field goal where it's basically a 50-50% chance of winning for both teams.

the Cowboys had a 74% chance of winning the game up 4 with 2:32 left. I f they'd been up just 3, their win percentage would have been 82%.

So inversely:

An offense down by exactly three points with less than 2:32 remaining would improve their odds of winning by going for it on fourth down rather than kicking a game tying field goal?

The 74% -> 82% change is due to the crude TD prob / FG prob -> expected points calculation built into the calculator.

this guy wrote:

So inversely:

An offense down by exactly three points with less than 2:32 remaining would improve their odds of winning by going for it on fourth down rather than kicking a game tying field goal?

Yes, but if they go for it and fail, then it's the coach's fault for taking a risk. If they kick and miss or go to overtime and lose, then it's fault of the kicker or the players, or good play by the other team.

By kicking, your odds of winning may be lower, but your chance to avoid blame is higher.

Mundus Vult wrote:

Malmo (and others),

I am not here to defend Rojo's faulty conclusion. Just wanted to point out that his frame for this discussion was a "crazy" thought for the day.

s

DOINK!

elephino wrote:

Teams will let other teams score a touchdown when tied or down one point with about a minute to go to keep the other team from running out the clock for a field goal and giving themselves a little time for another shot. The Packers did it in the superbowl against the broncos (lost anyway). So there are some precedents to what you're saying.

But the Packers had nearly zero percent chance of winning had they allowed the Broncos to run out the clock and kick the field goal. It was a great move.

In yesterday's game the chance of wining was hugely favorable in Dallas favor by taking the extra point. Had they not taken the extra point and opted to retain their 3 point lead, their chance of wining was ..... also hugely favorable.

As I said before, "playing the percentages" this way is effective only if you get to repeat the game hundreds of times to realize the difference in the probable outcome. Otherwise, you are being "too clever by half." It's a good way for a bookie to make money, a bad way if your job depends on the results of today's game.

It amazes me that people can't think outside the box.

I provided stats which show I could greatly increase the Cowboys odds of winning and people dont want to hear it. My analysis could help NFL teams win games way more than all but a handful of players on the team actually can and yet I get ripped by the 30% of the population that has no brain.

Here is an article on a similar matter talking about the "dreaded six point lead."

Basically earlier this year, the ravens were up by 3 against san diego on the 13 yard line with 2:27left. It was 4th and 4. The ravens kicked a field goal. It would have been better for them to go for it.

http://www.advancedfootballanalytics.com/index.php/home/analysis/game-analysis/175-the-dreaded-six-point-lead

It's not quite the same thing I was talking about (purposely giviing up points) but it reflects the fact that being up 3 isn't a bad thing as you have a real good likelihood of goign to overtime if your defense falters.

If I ever get rich enough to own an NFL team, you can damn bet I'll have stat boys running the time outs, 4th down decisions.

rojo wrote:

If I ever get rich enough to own an NFL team, you can damn bet I'll have stat boys running the time outs, 4th down decisions.

That's already being done. Welcome to 1990.

rojo wrote:

I provided stats which show I could greatly increase the Cowboys odds of winning and people dont want to hear it. My analysis could help NFL teams win games way more than all but a handful of players on the team actually can and yet I get ripped by the 30% of the population that has no brain.

But the issue is that it doesn't "greatly increase" the Cowboys odds of winning. The odds of winning go from 74% (a very high percentage) to 82% (a similarly high percentage). It's not taking them from 30% to 90% or something crazy. In fact, given the specificity of the scenario, I wouldn't be surprised if there was a significant margin of error that could make those two percentages even closer.

Some intangible factors also come into play, which of course the stats guys hate. Going for the extra point shows the coach has faith in his defense to get a stop in the next series. Coaching isn't 100% stat based, not matter what some folks think.

I really like the new era of stats, but this particular instance isn't something you should be hanging your hat on rojo.

Rojo,

You don't think that every single NFL team has considered bringing in a stat guy at some point? With all the millions of $$$$$ on the line, you think you're the first person to think of this?

The smartest guys who ever walked on Wall Street focused on statistics and theoretical edge yet they lost bigtime because they didn't have enough common sense to pay attention to the risk of ruin.

I any game of probability the first and last rule is always make sure you live to play the game again tomorrow.

http://en.wikipedia.org/wiki/Long-Term_Capital_Management

excellent point. and that risk of ruin applies particularly to going for it on fourth down in particular situations, e.g. 4th and 10 on your own 35, where you are very likely to give up a touchdown or field goal if it doesn't work. but the risk of ruin also explains why teams would want to be up 3 rather than 4 near the end of a game, since the other team, down 4, is a virtual certainty to lose (ruin) if they do not go for it on fourth down every time it happens until they score the touchdown, since a field goal will not likely help them without a lucky onsides kick or fumble.

malmo wrote:

As I said before, "playing the percentages" this way is effective only if you get to repeat the game hundreds of times to realize the difference in the probable outcome. Otherwise, you are being "too clever by half." It's a good way for a bookie to make money, a bad way if your job depends on the results of today's game.

I'm trying to figure out your exact point. Are you saying that by doing something out of standard practice (not kicking the extra point) and losing threatens your job as coach more than doing the standard (kicking the extra point ) and losing?

My reading is that if rojo is correct about the winning percentages (which I don't necessarily believe) then you would go with the higher percentage of winning. Consider this situation: A psychopathic killer with bad hair tells you that you live if a coin he flips comes up heads and then gives you a choice between one that is fairly weighted or one that is weighted to come up heads 53% of the time. I'm going with the unfair coin even though I only get one flip.

You still get just one flip and your chances are 50/50