Why Not Two Point Conversions?

Deus Irae said:

The issue with analytics is always the same. No matter how they might slant in one direction or another, they cannot, or at least currently do not, take the immediate context into account. Is it Brady going against the worst defense in history? Context tells you to favor analytics that imply a QB win. Is it Luke Falk going against the best defense in history? Context tells you to be very careful about analytics that imply a QB win.

Even Truthseeker's attempt to justify going for 2 when down 14 demonstrates the folly of relying on analytics. Using his numbers, the chance of losing if you go for 2 ends up being 27.56%. The chance of losing if you go for 1 is just 8.23%. So, barring unusual circumstances, you're a fool if you go for 2. Yet Truthseeker thinks it's a no brainer to go for it, because the analytics has him twisted to the point of ignoring the loss potential in favor of the win potential. And that's not a shot at Truthseeker. That's just an inherent problem with relying on the analytics. In football, a loss is what kills you. If you're tied, you still have a shot in overtime.

Click to expand...

I certainly agree with you that all circumstances need to be taken into account and that, when that is done, what appears obviously correct with no context can be questionable or even wrong within a specific context.

I am disappointed that you pulled the loss percentages out of the broader context to present a misleading overall picture. But it is what it is.

TommyD4207 said:

This. He coached that game like they were 50-point underdogs. As soon as they went for 2 the first time, KC must have known they had the game won, since Baltimore was going to play desperate.

Click to expand...

In what universe is going for 2 points as a strategy a sign of desperation ?

If Harbaugh thinks with Lamar as QB they have an edge compared to other teams on 2 point tries then good for him to test that theory out.

It might fail or it might succeed. One game and 3 attempts means absolutely zilch. You need a way bigger sample size for that. Hell, most probably even one season is not enough to say anything about it.

Another factor to consider is that the TD is not the only way to get points. I think looking at statistics for TD only, and comparing the extra point vs 2-point conversion has to also factor in relationship to FG's.

1 TD with extra point is worth more than 2 FG's. Given the finite number of possessions in a game, by missing a 2 point conversion, you basically reduce the need for the other team to score a TD. For example, if you are down 3-0, score a TD, and fail the 2 point conversion, the other team just needs a FG to tie, but if they score a TD and nail the extra point, a FG no longer ties the game, you need to score a TD.

With maybe 10-12 possessions in a game, you need maximize the odds of most points scored per individual possession. The XP gives the highest odds per possession after scoring a TD).

Deus Irae said:

Teams have been trying that this year, and losing because of it, or because of it in part. The math really only matters if you can get enough attempts at the play in one game. Since you're not going to be putting up 10 TDs a game, you're better off taking the PAT, because you're less likely to be chasing the lost point(s).

Generally speaking, going for two when you don't have to is a stupid idea.

Click to expand...

Question: If you line up for a two-point PAT, do you get two points if you drop kick it through the uprights?

TruthSeeker said:

I certainly agree with you that all circumstances need to be taken into account and that, when that is done, what appears obviously correct with no context can be questionable or even wrong within a specific context.

I am disappointed that you pulled the loss percentages out of the broader context to present a misleading overall picture. But it is what it is.

Click to expand...

I didn't present a misleading picture.

goheels22002 said:

Question: If you line up for a two-point PAT, do you get two points if you drop kick it through the uprights?

Click to expand...

Zero points, actually.

The new rule does not change the definition of how a try is scored or what each kind of try is worth. Rather, it says that tries by kick can only result in a score if the team has chosen the 1-pt line of scrimmage.

This also means that if you go for 1 and there’s a bad snap or whatever and you end up running the ball into the end zone you get *2* points. This also means you score 2 off a fake kick, not that you’re likely to try that because the snap is from the 15.

mitsudriver said:

This conversation reminds me a lot about blackjack. I spent a lot of time learning different strategies and counts and put it into action and made quite a bit of money over the long run. The LONG RUN. If I was to go to a casino only 16 times a year (16 NFL games) there’s a good chance that I end the year down because I wasn’t able to play enough hands.

Working with percentages requires massive amounts of trials to REALLY make sense. If 2 point conversions are successful 50% of the time are you ok with losing a game by 3 because you experienced 3 fails in a row? 50% does mean 1 out of 2 but it also means 100 out of 200.

Click to expand...

Law of Large Numbers.

If we were struggling to get points I can see that but as of now getting the almost guaranteed one point is more valuable. I can understand Baltimore’s reasoning given they couldn’t stop the Chiefs and needed all the points they could get.

Ravens next game plan is to onside kick on every kickoffs

At this point, Gostkowski is hitting 87.5% of his extra points.
Average teams make 48% of their two-point conversions.
So if the Pats have an average offense, their expected value on two-point conversions would be 0.96 point per play, and kicks 0.88 points per play.
Further, the historic stats show that running plays convert at around 63%. So when the Pats run, they would expect 1.26 points per play.

I've seen the issue of injuries raised. Running more two-point plays rather than kicks would risk injuries. I have two words in reply: Gronkowski's arm.

So why don't they do it?
I assume it's a couple things.
Maybe it would be a negative psychological message to the kicker.
Maybe none of the other kids do it, and there really is a strong culture of fitting in. You see the same thing with the stats on fourth-and-short and punting, and how long have people questioned that fourth-and-short in Indy?
Maybe it's an issue of greater downside than upside - risk avoidance bias. The upside of an extra point is not perceived as valuable as the downside of losing a point.
Maybe it's the small-sample-size issue. Statistically teams should be better off always going for two over the course of a season. The expected value is greater when compared to the league-average expected value for kicking extra points. However, if you score three touchdowns, with Gostkowski's current success rate you have a 67% chance of scoring three points and only a 0.2% chance of scoring zero (three straight misses). If you attempt the conversion, you have an 11.1% chance of scoring six points, but a 14.1% chance of scoring zero (using the overall 48% success rate). So, statistically, in two games out of a sixteen game season, you would expect three straight conversion attempts to fail. That's playing not to lose, but that could be the argument.

Deus Irae said:

I didn't present a misleading picture.

Click to expand...

Here's what you said:

Deus Irae said:

Even Truthseeker's attempt to justify going for 2 when down 14 demonstrates the folly of relying on analytics. Using his numbers, the chance of losing if you go for 2 ends up being 27.56%. The chance of losing if you go for 1 is just 8.23%. So, barring unusual circumstances, you're a fool if you go for 2. Yet Truthseeker thinks it's a no brainer to go for it, because the analytics has him twisted to the point of ignoring the loss potential in favor of the win potential. And that's not a shot at Truthseeker. That's just an inherent problem with relying on the analytics. In football, a loss is what kills you. If you're tied, you still have a shot in overtime.

Click to expand...

Just as a reminder, the numbers came out like this when down 14 points close to the end of the game:

Summary of 1-point conversion on 1st attempt:
Win/Lead: 0%
Tie: 91.77% (89.11 + 2.66)
Loss by 1 point: 5.29 %
Loss by 2 points: 2.94%
Total: 100.00%

Summary of 2-point conversion on 1st attempt:
Win/Lead: 44.84%
Tie: 27.60% (2.66 + 24.94)
Loss by 1 point: 0.00%
Loss by 2 points: 27.56%
Total: 100.00%

I apologize for not taking the next obvious step and completing the analysis of ties. When I googled what percentage of games end in a tie, I saw a 4.6% tie rate. Let's use that as a fair number. When considering win vs. loss, I think it's fair to split it down the middle (not sure what reason could justify a different split). So the probability of a win is 47.7%, loss is 47.7% and tie is 4.6%

Updated win probability when overtime is considered:
Summary of 1-point conversion on 1st attempt:
Win/Lead: 43.78% (91.77% * 47.7%)
Tie: 4.22% (91.77% * 4.6%)
Loss : 52.01% (5.29 % + 2.94% + 43.78%)
Total: 100.01% (rounding)
Win probability (win + 1/2 of tie) = 45.89%

Summary of 2-point conversion on 1st attempt:
Win/Lead: 58.01 (44.84%+13.17%)
Tie: 1.27 (27.6% * 4.6%)
Loss: 40.73% (27.56% + 13.17)
Win probability (win + 1/2 of tie) = 58.735%

Total: 100.01% (rounding)

In this specific scenario (score 2 TDs in regulation while opponents don't score) gives a 58.73% win probability vs. 45.89% win probability if you go for 2 after the first TD. Or you can go with the 58.01% vs. 43.78% if you don't want to figure in ties. Either way, I think that's a huge difference. Your mileage may vary. I also think that this reasoning applies in similar circumstances.

Hopefully this too long and too technical reply makes it easier to understand why I thought the 8.23% loss probability vs. 27.56% loss probability you cited was misleading, although technically accurate.