“OK I see your side of the argument. I still disagree though when you consider that over the course of a ten year coaching career you are talking about 16 points. The likelihood of those points all falling in games that actually matter is extremely low. However, it is likely that you could cause your team to lose in games where you were not able to convert 2pcs”

It is the Super Bowl. There are 0 seconds remaining on the clock, you are down by one point. Your odds of successfully converting a 2PC are known to be above 50%. What is it you do?

It is the first quarter of the first game of the season. You’ve just scored in 2 minutes. You are up by 6 points presented with the decision to kick an XP, which you know has a success rate no higher than 100%, or attempt a 2PC which is known to succeed 52% of the time. What is it you do?

If 16 points are inconsequential, then it shouldn’t matter what coaches do, right? So why do they all do the same thing? We can either discuss the consequences of a 2PC strategy or we can write off consequences all together, but we can’t do both. It can’t be consequential and inconsequential at the same time.

Also, isn’t it strange how the issue is framed at this point? Stating: “It’s only 16 points…” raises the question, why is the 2PC the strategy that needs to justify itself? X or X+16? If you’re really granting that it scores more points, why does it matter if it is 16 or 1 or .0000001? The better strategy doesn’t have to explain itself to the worse one, afterall. Is Blackjack worse than Roullette because the odds are only slightly better? How does that make sense?

“In my eyes an NFL coach is the sixty year old man. He doesn’t have a lot of chances to make either two points or one point after a touchdown and the effect of a bad string of 2pcs could be disasterous for a season.”

Why is it that ONLY a 2PC strategy has “disasterous” results? Disastrous results are LOSSES. You keep wanting to frame the issue in terms of the success rate of the XP, but who cares how successful the XP is? Do coaches win games because their kicker successfully kicks extra points? As it turns out, no, in fact they’d win MORE games if they didn’t send the kicker out at all.

This is not RATIONAL RISK AVERSION. No one doubts that XPs have a higher success rate than 2PC. That is not the “risk” part of the proposition. The “risk” coaches take is the percentage chance that they will win or lose a football game. If that is the case — I don’t understand how it couldn’t be the case — than the RISKIER STRATEGY is kicking extra points. If coaches understood the statistics, a risk averse move would be to… go for 2!

The analogy to investment can be dismissed easily enough. Risky high % return investments have an increased likelihood of completely divesting the investor of all his moneys. At which point the game stops, they lose, fooey. Depending on the level of risk, the % return, etc. reasonable minds can avoid the risk.

Football does not stop when a coach loses a game. Another game happens. In so far as we’re talking about XP and 2PC strategy competing in the entire universe of Football, across the rest of the history of the NFL, there is no risk. The game will not stop if teams go for 2. Statistically speaking, no coach will lose every game they play because they go for 2. Indeed, when placed against a 2PC strategy, coaches who go for XPs face a higher likelihood of losing football games, which is the ultimate risk coaches want to avoid.

Unless, of course, it is as everyone knows it to be, that coaches aren’t trying to maximize their % chance of winning by their post-touchdown strategy, but rather are just trying to play it safe. This might be a classic example of Herding. In any event, that wouldn’t justify the decision if we all agree that a coach should do what maximizes winning rather than maximizing job safety. (And I would argue that winning has a much more positive impact on whether one keeps a job.)

“Frankly, sometimes you are playing not to lose just as much as you are playing to win.”

If you are playing not to lose, the best strategy is to go for 2PC.

“Past results do not exactly replicate real probabilities. Meaning that just because 52% of the past however many 2pcs have been made, doesn’t mean there is a liklihood of 52% that 2pcs will be made in the future.”

This, I think, is the strongest criticism of the 2PC strategy. While you are correct, I can’t assume that going for it will remain above 50%, I certainly think it will. All I’m saying, my only conclusion really, is that if an NFL coach thinks he has a better than 50% chance of converting a 2PC, then he’s stupid if he kicks an XP. And in so far as the data suggests that the rate is something like 52% in the 2008 NFL game, 32 NFL coaches cannot reasonably conclude that their rate is well below 50%. If you grant me nothing else, at least acknowledge that the best offense in the league plaing against the worst defense in the league converts over 50% of its 2PC attempts. If that’s the case, we aren’t seeing enough of the 2PC.

“There have been significant portions ofthe NFL’s history where 2pcs were under 50% effective so this recent turn of good fortune for those that tried for two doesn’t necessarily mean that is indicative of the true probability.”

My response: Are 1997 statistics more or less relevant to the 2008 season than are 2007 statistics?

In any event, even as recently as 4 years ago or so, the historical average (since ‘94) of the 2PC was around 48%. It is approaching 50%. Assuming rates stay around where they are now above 52%, it won’t be long before the historic rate is above 50%. Will you, then, approve of a change in strategy?

JeffV you’ve been a great sport and I really enjoyed it too. Have a good one, thanks to Mr. I for letting us rant on this thing.

Another Point

In investing a lot of what your doing isn’t trying to maximize return as much as it is trying to manage a clients goals.

For example: a young man of 24 who is investing money that he doesn’t have an every day need for and is planning to leave untouched for 40 years is wise to choose a high risk, high return portfolio.

On the other hand, A sixty year old man who plans on retiring in 5 years is better served by a low risk, low return portfolio that doesn’t stand to bankrupt his retirement in a bad year but also doesn’t stand to make him much more rich.

In my eyes an NFL coach is the sixty year old man. He doesn’t have a lot of chances to make either two points or one point after a touchdown and the effect of a bad string of 2pcs could be disasterous for a season.

I think the point you made about professor such and such’s paper with risk tolerance and how everyone should maximize their return is interesting but not applicable to every investor (or in this case coach).

Frankly, sometimes you are playing not to lose just as much as you are playing to win. The worst case scenario of going for two is much worse than kicking.

Another new thought

Past results do not exactly replicate real probabilities. Meaning that just because 52% of the past however many 2pcs have been made, doesn’t mean there is a liklihood of 52% that 2pcs will be made in the future.

There have been significant portions ofthe NFL’s history where 2pcs were under 50% effective so this recent turn of good fortune for those that tried for two doesn’t necessarily mean that is indicative of the true probability.

Anyway. I’m going home and turning my brain off now. I really enjoyed commenting on this thread with you all, especially Hogs.

Good night and Go ‘Skins.

“This paper examines a single, narrow decision–the choice on fourth down in the National Football League between kicking and trying for a first down– as a case study of the standard view that competition in the goods, capital, and labor markets leads frims to make maximizing choices. Play-by-play data and dynamic programming are used to estimate the average payoffs to kicking and trying for a first down under different circumstances. Examination of teams’ actual decisions shows [ED emphasis added] SYSTEMATIC, CLEAR-CUT, AND OVERWHELMINGLY STATISTICALLY SIGNIFICANT DEPARTURES FROM THE DECISIONS THAT WOULD MAXIMIZE TEAMS’ CHANCE OF WINNING.”

It is a different scenario but serves to illustrate the point that NFL coaches aren’t necessarily rational.

The assumption that the correct decision is the one a majority of people would take ignores the empirical data on the irrational human psychological condition.

It all comes back to:

“Uh, I’d take the free dollar from coin 1 every time over the 48% chance of winning nothing.”

So would most people. But if Chris and I played this game, and I did the exact opposite of him, I’d be richer. The motivation of maximizing profits which drives Chris’ decision (I don’t doubt by his answer, for instance, that Chris wants money) is inconsistent with the decision made, since he chooses the strategy that nets him less money. I’m just sayin’ we should recognize same when it comes to NFL coaches and post touchdown strategies.

“Uh, I’d take the free dollar from coin 1 every time over the 48% chance of winning nothing.”

Actually I don’t think you’re alone. However that raises some serious doubts about, to quote another, “the assumptions of rational behavior.” People who want money should choose the game that produces the most of it.

“The problem with your logic is that NFL teams don’t score 100 touchdowns a game. Thus the sample size for this experiment between 2pc and XP is too small for a 4 percent increase in points to win consistently.”

Yes but coaches will win more games by going for 2 points than they will lose games by going for 2. The sample size only demonstrates the better strategy, besides that it is largely irrelevant. If a coach is presented with a 52% chance to win a game and a 48% chance to win a game, why would he choose the latter? Where 2PC strategy beats XP strategy a majority of games, why choose the XP strategy?

“Meaning that the coach who goes for two needs to be successful 66% of the time to get a win in this game. Considering the overall success rate is only a little over 50% that is not a good proposition. It would not be good enough for that coach to be 54% or 52% successful on the season. The team needs the points today!”

My response is that isn’t entirely fair because you’re conflating %s here. Yea, a coach needs to be 66% successful to win here… but he’s going to be 66% successful a majority of the time he goes for 2 points over a three touchdown history. A coach who goes for 2 at a base 52% success rate over three touchdowns has the following likelihoods of points scored after TD:

4>2>6>0

In other words, in a majority of circumstances, he scores more than 3 points. Stated otherwise: if you know you score three touchdowns in a game as does your opponent, and he goes for the XP, you will beat him more often than not if you go for 2.

To turn the wording on its head, yea you have to be successful 66% of the time to win, but you also have to fail 66% of the time to lose. If the percentage of success on 2PC is higher than 50%, are you more likely to go 2 out of 3 or 1 out of 3?

There is no number of touchdowns scored whereas you can match a XP strategy against a 2PC strategy and have the former beat the latter in a majority of instances. It remains true of 1 TD, 3 TD, 57 TD, 2 TD, etc.

“If the 2pc’s are just over 50% sucessful–a big if when you consider that a team who constantly went for two would have opposing teams preparing for this much more than the average team currently prepares to defend 2pcs–that means that they would most likely make one 2pc, miss one 2pc and then have a chance slighly above 50% (maybe) to make the third and win.”

Before we dive into the nuanced or esoteric arguments about game preparation, just remember that those kinds of cases can be made both ways. Does the team who gameplans the week prior against the one of 16 opponents who has a 2PC strategy decrease the odds of success more than does the team who practices 4th and 2 every single practice, every game, over the course of a 16 game season, increase their odds of success? If gameplanning to stop a 2 yard gain decreases the likelihood of success, doesn’t practicing a 2 yard gain increase it? And wouldn’t, ostensibly, a 2PC strategy team practice that a lot more (both in practice and in actual game situations) than would any team accustomed to defending an XP in 15 out of their 16 games?

Now, regarding the output, you’re right! As the most likely scenario in a 3TD scenario is one made, one missed, and then one 52% opportunity, that’s the REASON you go for two! The alternative is you go for the XP and then you necessarily tie with the XP strategy… just so you can have the opportunity to literally flip a coin in overtime. In other words, you are sacrificing a 52% chance to win for a 50% chance to win. Why do that?

“This gives the defense and special teams no chance to win. If you think your team is good why not trust your defense to stop the other team?”

The coach should not be interested in spreading out the wins between his defense and special teams and offense equally, rather he should be interested in winning the most games capable by whatever means capable. There is nothing special about special teams (or offense that matter) that entitles either unit to win the game. They are simply tools used towards the ultimate end which is: Ws. A coach should seek to “give[] defense and specical teams” a chance to win only and exclusively in those circumstances where doing so… actually increase his odds of winning. Consider: Down by 4 points on 4th and 1 on the 1 yard line with 10 seconds remaining. You can give your offense a “chance to win” by going for it or you can give your special teams a chance to win it by kicking the FG, then sending them out for the onside kick, then sending them out for another FG. The proper strategy is to go for it, since equalizing opportunities for your respective unit to win the game is irrelevant.

Response to the defense question: 1) A lot of teams don’t think their defense is good and 2) Your extra point strategy has either no impact or a net positive impact on your defense. I say no because why would your defense care what the offense does after a touchdown? I say net positive because, obviously, the point swings would dictate the game just as much as any other point differential (if you go for 2 and it ultimately puts the other opponent down by 4 points rather than 2 points your defense will play to stop the TD late in the game rather than playing to stop the FG). But from a defensive perspective… don’t you always want your offense to assume the strategy that increases the dispairty between our points and theirs positively? If team A averages 17 points and team B averages 16 points, the defense for team A is better off, since they will need to do one point less worth of work than the Team B defense.

“As a coach I think it is a better idea to play it safe and take the points unless wacky circumstances dictate otherwise.”

I think that is certainly the position of every single NFL coach, so you are not alone.

“Keep in mind that over the course of the year a 4% increase in extra point production only equates to a 1.6 increase in points.”

Sure, the question is, why would a coach given the choice between X points and X+1.6 points choose the first? You can either say the strategy is inconsequential (in which case there’s no good reason to pick an XP strategy over a 2PC strategy or vice versa, since it doesn’t matter, right?) or it is. If it is consequential — it is — than there’s only one right strategy, and it can’t be the one that generates fewer points.

“That means one or two points a season depending on how the chips fall. Those 1 or two points might come in a game where the score is not close which would mean that they don’t matter.”

They might or they might not. But in blowouts or monster wins your post-touchdown strategy is irrelevant. But you don’t decide to go for two on your first five touchdowns after they’re scored; you decide to go for 2, or not, after you score the first touchdown. Then you decide after you score the 2nd touchdown. Then you decide…

Point being, coaches are presented an opportunity something like 50 times a year to take one decision and take another, the former increasing your odds of winning strategically. The strange thing is, the entire world of NFL football seems to agree that you should choose the strategy that decreases your odds of winning. It is one of the truly fascinating aspects of the game that either dramatically calls into question the rationality of coaches or, alternatively, suggests they simply aren’t aware that the statistics have turned in their favor on the 2PC.

Whatever strategy you think is better, I think we can all agree that whether 2PC is better than XP has SOMETHING TO DO WITH the actual likelihood of converting 2PCs. Prior to 5 years ago, the 2PC rate was below 50%. Phil Simms, among others, poined out repeatedly how stupid it was for coaches to go for 2 in the majority of circumstances… and he was right! Coaches were going for two more often than they should have been based on the likelihood of success. As a result, coaches began going for two less often. This led to the absolutely absurd result of: in the history of the 2PC, coaches have gone for two more often when they should not have and less often when they should have. Anyone assuming the sanity of NFL coaches should wonder about that result.

The problem with your logic is that NFL teams don’t score 100 touchdowns a game. Thus the sample size for this experiment between 2pc and XP is too small for a 4 percent increase in points to win consistently.

Consider this: If a coach has three touchdowns in a game and kicks the extra point and makes it 3 times he has 21 points.

If the opposing coach’s team scores three touchdowns and goes for 2 each time he could have anywhere between 18 points and 24 points. He would need two out of three 2pcs to be successful to win the game.

Meaning that the coach who goes for two needs to be successful 66% of the time to get a win in this game. Considering the overall success rate is only a little over 50% that is not a good proposition. It would not be good enough for that coach to be 54% or 52% successful on the season. The team needs the points today!

If the 2pc’s are just over 50% sucessful–a big if when you consider that a team who constantly went for two would have opposing teams preparing for this much more than the average team currently prepares to defend 2pcs–that means that they would most likely make one 2pc, miss one 2pc and then have a chance slighly above 50% (maybe) to make the third and win.

Basically they would be losing games and winning games based on their ability to punch it in from two yards out. This gives the defense and special teams no chance to win. If you think your team is good why not trust your defense to stop the other team?

As a coach I think it is a better idea to play it safe and take the points unless wacky circumstances dictate otherwise.

Keep in mind that over the course of the year a 4% increase in extra point production only equates to a 1.6 increase in points.

That means one or two points a season depending on how the chips fall. Those 1 or two points might come in a game where the score is not close which would mean that they don’t matter.

Don’t underrate the consistency of the XP.

Uh, I’d take the free dollar from coin 1 every time over the 48% chance of winning nothing.

I think you’re absolutely right that the biggest hurdle to convincing coaches to optimize their scoring strategy is simply that you can’t convince them to change the status quo for what is, admittedly, not all that much benefit. The results aren’t amazingly consequential given that, over the course of an entire NFL season, we are talking about just a few more points.

But what a Coach should be asking themselves after a touchdown is really only and exclusively: What can I do to maximize the points I score here? Run the permutations of teams under similar circumstances and a 2PC strategy that converts over 50% of the time wins a majority of games over an XP strategy that converts ~ 98% of the time. Two teams score one touchdown, you take that result 100 times over, and the guy who goes for 2 wins 52 games and the other guy wins 48 games.

The issue for me is, while this is irrefutably true:

“Basically what the above googlable phrases are saying is that in order to make an investor take more risk you have to give them the chance at more return.”

It shouldn’t be! As has been argued by Peter Bernstein based on the research of Richard Thaler and Daniel Kahneman and Amos Tverseky, the human risk-aversion (perhaps stated more accurately as LOSS-aversion) basis for decisions is… irrational. If Head Coaches are in the business of scoring points and winning games, all that should be required to convince them of the efficacy of strategy A over strategy B is evidence that strategy B maximizes scoring. When it comes to risk, the correct answer is: Maximizing returns is the proper move. But that isn’t how things work in practice, as human beings routinely sacrifice bigger returns because we’re generally (illogically) risk-averse.

The crazy thing is that no one would ever get this wrong if the problem were simplified. I give you an opportunity to play a game whereas you have a choice to flip two coins, one is weighted to land heads 98% of the time and the other is weighted to land heads 52% of the time. The former will result in you winning 1 dollar everytime heads lands, the latter 2 dollars everytime heads lands. No one would select the first coin.

So why should/do coaches always do so?

Sorry for the douchiest post in history.

(although it does make sense when you think about it)