Strategy Discussion: The Two-Point Conversion

[ivanvamp]

I've been thinking about the two-point conversion for a while and what it means for the NFL and game strategy. Peter King wrote about it in his most recent MMQB piece, so I thought it might be worth discussing here.

Here's an article that uses stats from 2000-2009 (Advanced NFL Stats: Almost Always Go for 2-Point Conversions?), so we're missing several years' worth of data. I will work on the assumption that not too much has changed, so the numbers are still fairly relevant.

Here are some interesting tidbits:

"There have been 718 2-point conversion attempts from 2000-2009, including playoff games. Overall, they've been successful 46.3% of the time. But this is slightly misleading because it includes aborted kick attempts. If we weed those out, along with some other mysterious plays, such as Josh McCown's kneel-down while trailing by 5 points in the final few seconds of the Cardinals-Vikings 2003 game, we get a different answer. For all normal 2-point conversions, the success rate is 47.9%."

Ok, so we start with a normal success rate of 47.9%. However, that can be broken down further into rushing vs. passing plays.

2-point passes: 43.4% success (525 attempts)
2-point rushes: 61.7% success (183 attempts)
TOTAL: 339-708 (47.9%)

If you weed out the QB scrambles/runs and just go with conventional running plays, you get this:

2-point passes: 43.4% success (525 attempts)
2-point QB runs: 74.5% success (47 attempts)
2-point RB runs: 57.4% success (136 attempts)
TOTAL: 339-708 (47.9%)

Now, what's clear from these numbers is that teams going for 2-point conversions throw FAR more than they run, and yet running is FAR more successful. So it is a mystery why they throw as much as they do. Other than that the league has become so pass-heavy it's almost their default position.

Looking at New England's data from 2012 (2012 Patriots Defensive Stats, Breaking Down The Patriots Performances on 1st Down, 2nd Down, 3rd Down,4th Down and Total Number of Plays - thanks Ian!), we can see how the Patriots do on either 3rd and 2 or 4th and 2 situations (I want to limit it to those because 2nd and 2 is COMPLETELY different than, say, 4th and 2. 3rd or 4th and 2 is a far better comp to a 2-point conversion). Here were the Patriots' numbers in 2012 on 3rd and 2 or 4th and 2:

3rd and 2: 26 plays, 14 conversions (53.8%)
4th and 2: 1 play, 1 conversion (100.0%)
TOTAL: 27 plays, 15 conversions (55.6%)

So 55.6% of the time, the Patriots converted from 2 yards out. Since 2001, here are the Pats' 2-point conversion numbers:

2-point passes: 7-17 (41.1%)
2-point rushes: 9-12 (75.0%)
TOTAL: 16-29 (55.2%)

These numbers line up very well with the NFL in general.

So let's put this together. The league as a whole converts less than 50% of their 2-point conversions. But the Patriots convert 55.2% of theirs, and 55.6% of their 3rd/4th and 2 plays. The Pats scored 67 touchdowns in 2012. Gostkowski has made 355 out of 356 career PATs, so let's just call him 100% at PATs. That means that the Patriots would have scored 67 points after touchdowns in 2012.

But at a 55% conversion rate, over 67 touchdowns, they would have converted 37 2-point conversions. 37 x 2 = 74.

Therefore, the Patriots would have gained an additional seven points over the course of a season by going for 2. If they ran it more, they'd convert even more, based on their past success rate.

So the question here is this: Why don't the Patriots go for 2 every time? It doesn't make sense to go for two regularly if your success rate is less than 50%. But for the Patriots, it's greater than 50%. So it makes sense for them to go for it every time. Of course there will be failures. But they will succeed more often than they will fail, and that will be a net positive.

Now I know Belichick knows these numbers and he obviously knows football far more than I do or pretty much anyone else for that matter. But the math here doesn't lie. Why do you think they don't go for 2 regularly? I know it's not orthodox, but that's never seemed to bother Belichick before.

 

[ivanvamp]

Can a mod edit the title of this thread? I really can spell "conversion".

:bricks:

 

[QuantumMechanic]

The goal of a coach is to win games, not score the most points.

So you need to look at the variance of the strategy as well as the expectation of value of points.

A good team (because it is good) generally wants to follow a low-variance strategy -- in other words, because its average outcome is high, it wants to follow a strategy that gives outcomes tightly-clustered around that average. A bad team wants a high-variance strategy -- if it's gonna win it needs outcomes widely spaced around its average -- in the hopes that it'll get an outcome well above its average.

This is why when a bad team plays a good one it will try to slow the game down and minimize the number of possessions. The more possessions, the lower variance and the more likely the better ability of the good team will dominate luck. Conversely, with a fewer number of possessions the more likely it is that luck will trump skill, which is what the bad team needs to have happen.

So even though going for 2 a lot more often would give more expected points, the higher variance of the strategy may offset the benefit of more points for a good team like NE.

 

[ivanvamp]

The goal of a coach is to win games, not score the most points.

So you need to look at the variance of the strategy as well as the expectation of value of points.

A good team (because it is good) generally wants to follow a low-variance strategy -- in other words, because its average outcome is high, it wants to follow a strategy that gives outcomes tightly-clustered around that average. A bad team wants a high-variance strategy -- if it's gonna win it needs outcomes widely spaced around its average -- in the hopes that it'll get an outcome well above its average.

This is why when a bad team plays a good one it will try to slow the game down and minimize the number of possessions. The more possessions, the lower variance and the more likely the better ability of the good team will dominate luck. Conversely, with a fewer number of possessions the more likely it is that luck will trump skill, which is what the bad team needs to have happen.

So even though going for 2 a lot more often would give more expected points, the higher variance of the strategy may offset the benefit of more points for a good team like NE.

In other words, the guaranteed 1 point every time is a better strategy than the 55% chance of 2 points every time, because the 1 is a virtual lock with every TD, while you could go stretches (an entire game maybe) of not converting a 2-pointer, even though that might be offset by a long stretch of converting them?

That makes some sense.

That said, if they convert 2-pointers at about 75% when they run, it would seem like they could lessen the variance considerably simply by running more. But then game theory suggests that opponents would adjust to that and play the run more. And so it would go...

 

[patchick]

The league as a whole converts less than 50% of their 2-point conversions. But the Patriots convert 55.2% of theirs, and 55.6% of their 3rd/4th and 2 plays.

...

at a 55% conversion rate, over 67 touchdowns, they would have converted 37 2-point conversions.

I don't think the 3rd & 2 rate is relevant -- on a typical 3rd & 2 you still have to defend the whole length of the field, so it's a different defensive scheme from the 7-yard field that is a conversion attempt. In fact, according to the graph in this article the leaguewide success rate on 3rd & 2 is close to 60%:

Advanced NFL Stats: First Down Probability

The Pats' 2-point conversion rate is definitely good, but a small sample and not wildly divergent from the league average. Also, going for it every time could easily lower your success rate by forcing you to use a broader set of plays. So I'm not at all convinced that the expected value is really above 1 point.

 

[DocHoliday]

Strategy Discussion: The Two-Point Converstion

Maybe I'm missing something, but using the conversion rates of 3rd/4th and is different than a 2 point conversion attempt since there's far more field to cover.

A better comparison would be 3rd and 4th and goal 2 yards away.

 

[Palm Beach Pats Fan]

We are likely to see some of this from Chip Kelly this year in Philly.

At Oregon, while he didn't ALWAYS go for 2, he did it way more than anybody else. Plus early in games he almost always went for two.

Someone with chime in soon- Tim Tebow, 2 point conversion specialist!

(I'd prefer Ridley, myself).

 

[Synovia]

The Patriots have been overachievers the past two years. It doesn't have the talent to compensate for injuries, and it wins so much because it puts in 99% effort in the regular season and plays with terrific schemes to mask its deficiencies.

But in the playoffs a good team at 99% will not beat emotional, talented teams that play at 100%. It's what happened against the Giants in 2011 and the Ravens in 2012.

I realize this is a sig, and not a thread post, but it may be the dumbest thing I've read on this forum, and I needed to point that out.

The idea that the winningest team in football over the last decade has some sort of talent and effort gap is utterly absurd.

If all 12 playoff teams are of even talent, the #1 seed has a 12.5% chance of winning the superbowl. The Patriots are doing much better than that.

 

[ivanvamp]

I don't think the 3rd & 2 rate is relevant -- on a typical 3rd & 2 you still have to defend the whole length of the field, so it's a different defensive scheme from the 7-yard field that is a conversion attempt. In fact, according to the graph in this article the leaguewide success rate on 3rd & 2 is close to 60%:

Advanced NFL Stats: First Down Probability

The Pats' 2-point conversion rate is definitely good, but a small sample and not wildly divergent from the league average. Also, going for it every time could easily lower your success rate by forcing you to use a broader set of plays. So I'm not at all convinced that the expected value is really above 1 point.

Maybe I'm missing something, but using the conversion rates of 3rd/4th and is different than a 2 point conversion attempt since there's far more field to cover.

A better comparison would be 3rd and 4th and goal 2 yards away.

You both make good points. I realize the differences between 3rd/4th and 2 and a 2-point conversion (there's obviously even a difference between 3rd and 2 and 4th and 2). I was just trying to find something reasonably comparable. And the larger point still stands: the Patriots convert on these (and 2-point conversions) at a greater than 50% rate.

 

[DocHoliday]

Strategy Discussion: The Two-Point Converstion

I realize this is a sig, and not a thread post, but it may be the dumbest thing I've read on this forum, and I needed to point that out.

The idea that the winningest team in football over the last decade has some sort of talent and effort gap is utterly absurd.

If all 12 playoff teams are of even talent, the #1 seed has a 12.5% chance of winning the superbowl. The Patriots are doing much better than that.

You assume talent is evenly distributed among equally important positions.

That's not the case. Tom Brady has covered for this team since 2010. Take Tom off the 2010, 2011, and 2012 squads and they're barely playoff worthy.

But hey, surely you're the bastion of intelligence. Waiting to hear your response to this clearly idiotic argument I have.

 

[JR4]

I would almost bet that if they used 2 point conversion every time their
success rate would go down significantly.

I think it was BB in some conference call talking about red zone offense. I
think he said something like when you are close to the goal line, getting
across is a lot about trickery.
If so, then if you use something over and over you lose your trickery
advantage. I would think.

 

[Synovia]

You assume talent is evenly distributed among equally important positions..

I assume no such thing.

The only thing I assume is that the best team isn't going to win every game. At best, a team has about a 75% chance of winning a specific game, and every playoff game is lower than that. That means you're going to lose some.

This is basic probability. The Red Sox lose spring training games to Boston College. Do you think BC has more guts? Do you think BC has more talent?

 

[DocHoliday]

Strategy Discussion: The Two-Point Converstion

I assume no such thing.

The only thing I assume is that the best team isn't going to win every game. At best, a team has about a 75% chance of winning a specific game, and every playoff game is lower than that. That means you're going to lose some.

This is basic probability. The Red Sox lose spring training games to Boston College. Do you think BC has more guts? Do you think BC has more talent?

When a team loses you don't assign it to basic probability. A team loses because it isn't good enough to beat its opponent and the small percentage of chance that goes into each play. I can't even fathom how you'd take this approach to evaluating a FOOTBALL team. It's such reductive nonsense.

You haven't addressed the talent issue at all, only the part you thought (and failed) to make a proper response to.

Tom Brady has covered this teams deficiencies and without him (up until this year anyway) the Patriots are a middling team.

 

[patfanken]

Thanks for starting this thread. What a blessed relief from the media crush of the last few days

You topic reminds me of my own feelings about the extra point and that it should be radically changed, given that the success rate is now over 99%. It is without question the LEAST interesting play in an NFL game outside of the kneel down.

I would propose several changes to the extra point rule that would turn it into one of the most compelling, interesting and most talked about plays every game.

1. Eliminate the kick for the extra point.

2. Give the offense the following choices for their extra point.

a. They can run or pass the ball in from the half yard line for one point. I would imagine that the success rate would be about 85% on that. A number high enough to make it usually attainable yet with enough failures to make it interesting enough to watch

b. Put the ball on the 4 yard line and offer the opposition 2 points for the conversion. Even though I would imagine the success rate would be somewhere in the 42-47% rate, the fact that the one point conversion wouldn't be automatic anymore might make this option more appetizing.

3. Put the ball on the 10 yd line and offer 3 points for a successful try. This would add a lot of excitement if a team is down by 8 or 9 at the end of a game. Now with a chance to tie or win the game, the extra point now becomes the most important play of the game

4. The end result of all this would be a LOT more 2nd guessing by fans and "experts", and isn't THAT what's make the game more interesting for all of us. Rather than lowering the number of possible choices a HC makes during the game, we should be looking for ways to increase them. Here is one way.

5. Fiddle away anyway you want. Maybe you'd put the ball on the 3 yard line instead of the 4....or perhaps the 5. Maybe you'd give the offense 4 points for making the try from the 10.

I've been fiddling with the idea that if the offense doesn't get in from the half ydline, the defense gets the extra point. Or here's one. Put the ball on the 2 and have the try be for 2 points.....BUT, if they are stopped the defense gets one point.

This is all stuff that adds to the strategy and makes for lots of discussion after the game for fans, and ULTIMATELY makes a horrid play into a very interesting one.

Now isn't this a better discussion than what our 3rd string QB is wearing for socks.

 

[BTTA]

Tom Brady has covered this teams deficiencies and without him (up until this year anyway) the Patriots are a middling team.

If I have to replace Brady with an average NFL QB for the past few years, then I want $10-$12M in CAP space taken out of the QB position, to upgrade the talent elsewhere.

 

[Patsfanin Philly]

Could it be that the rush success rate is so much higher is that teams are defending the pass at the expense of defending the run? If more teams would attempt the rush rather than the pass you might see the success rate of passes go up while the rush success rate declines as teams crowd the box to stop the run...

 

[ivanvamp]

Could it be that the rush success rate is so much higher is that teams are defending the pass at the expense of defending the run? If more teams would attempt the rush rather than the pass you might see the success rate of passes go up while the rush success rate declines as teams crowd the box to stop the run...

Yes I'm sure that would happen.

 

[DocHoliday]

If I have to replace Brady with an average NFL QB for the past few years, then I want $10-$12M in CAP space taken out of the QB position, to upgrade the talent elsewhere.

The point is that Tom Brady won games in spite of this team from 2010 - 2012, perhaps with some mild help in 2012.

Compare to other teams that can bail out their QB. It's doubtful the extra 10-12 M in cap space of not having Tom Brady would bring in enough players (one or two impact players?) to make up for what he brings to this team.

Until this team can overcome less than stellar play from Tom on a consistent basis it has no chance at winning a title.