The first experience of a 2-pointer scored by the defensive team on an extra point play -- by the New Orleans Saints last weekend -- was almost accompanied by a second one the same weekend in the Baltimore Ravens' scrum.
Some time ago, I did a calculation of the odds of this happening, just to round out the whole winning percentage computation in different score-and-time scenarios. Based on that, here's what the "reverse deuce"* lay-of-the-land is.
On 2-point attempts: There were 6 reverse-deuce-like events -- 3 pick-sixes and 3 fumble returns -- on a standard play from the 2 or 3 yard line (allowing for plays inside the opponent's 3 removes possible error from recording 2 vs 3, and it gives us a sample that's at least plausible) from 2000-2014, out of almost 6,000 attempts. The two-point conversion thus has about a 1/1000 chance of turning 2 the other way.
Of course, situation-wise, only 3rd down and 4th down really reflect the urgency of a 2-point conversion, where it's one shot or nothing. If we confine ourselves to these plays, we find 2 turnovers (one of each type) in just shy of 1800 attempts. So the odds inch up a bit to 1/900.
On 1-point attempts: There were 14 returns for a TD on a blocked FG from 28-38 yards out in the same span; just over 4,900 kicks were attempted at that range. So about 1/350 kicks can be expected to be returned for a pair the other way.
Last season, there were almost 1,200 TD plays, so take this as the baseline. Through Week 5, 12 of 150 TDs (about 8%) were attempted 2P conversions. (EDIT: Through Week 13, the numbers are 80 in 994 post-TD attempts, so 8% seems to be pretty stable.) At that rate, we should get something like 100 2P and 1,100 modified extra points (mXP) from the 33-yard line.
Overall, then, we should see about 3 reverse deuces on a blocked mXP per year and (on average) 1 RD on a 2-point attempt every 9-10 years.
Had Baltimore and New Orleans both done the deed, I would have been surprised that our first taste happened in the same weekend, but I wouldn't have been shocked at the event.
However, the return called back on a penalty raised a further question, namely: how often does a blocked field goal get returned for a score then subsequently called back on a penalty? This need not be distance-dependent, but it will play into the calculations above (and I had failed to consider it).
Out of 29 field goal blocks returned for touchdowns from 2000-2014, 3 of them were called back -- one for offsides, one for 12 men on the field, one for an illegal block in the back; one of those was even in the target distance range. The rest stood as TDs. We should expect to see a similar percentage of penalty call-backs no matter the distance in question, so anticipate another blocked extra point return to be called back on a penalty no sooner than 3 years from now and no later than about 6 years out.
In conclusion, we should see a reverse deuce a little over 3 times a year. And don't be surprised if a team chucks an interception on a 2-point attempt; that's going to happen in the next few years, assuming these rules stay in place.
One additional note: Obviously the stakes of the play change during an extra point, so an intercepting or fumble-recovering defender has more incentive to try to return that turnover for a score during the 2-point play than during a typical play. For instance, if a player picks off a pass in the endzone and knows he can return it to the 5-yard line, he probably won't take the chance under normal circumstances, but he absolutely should on the 2-point play. This probably doesn't come up much, and I have no real way to evaluate the odds. But there were, in all, 113 turnovers on the short plays referenced, so even if 1% of these were a defender giving up on a play where he would have returned it all the way had he taken the chance, we get only an 17% increase in the odds and a resultant change to once every 8 or 9 years.
* Not in the Urban Dictionary sense, but in the original college football sense. I don't know when this term fell out of style, nor who has decided to populate this particular entry. (Hell, I'd even consider a better UD definition to be related to what The Cat did.) Regardless, "reverse deuce" so accurately describes the result of the play that I feel comfortable using the phrase.