An Ode to Bad Two Point Conversion Decisions

Dear Readers:

Mark Dantonio’s decision to go for 2 against Ohio State yesterday was another in a long series of bad 2 point conversion decisions. It was the proverbial straw that inspired me to write a bit about this phenomenon.

Let me first state that most NFL and college coaches are terrible at this particular decision-making data point. It amazes me that, in this age of advanced metrics and years of available statistics, coaches are still making decisions without any actual framework to justify those decisions. What it usually boils down to is that most coaches make decisions that put off losing as long as possible rather than making decisions that will increase their team’s probability of winning decisively.

This is also not a “fire McCarthy” post. I have long been one that criticizes Mike McCarthy’s in-game decision making (2 point conversion decisions, decisions to settle for field goals, terrible challenges), but I have not been one to call for his firing. I am a firm believer in not making such decisions unless one has a better solution. Every better coach already has a job, so firing McCarthy is likely to fall into the “careful what you wish for” category. He does have the second best winning percentage of any active coach – Yes, plenty of that is due to Aaron Rodgers.

That all stated, let’s analyze four 2 point conversion decisions, the first being in the Michigan State-Ohio State game yesterday. The other three are decisions made by Packers head coach Mike McCarthy over the past two seasons.

Ohio State at Michigan State (November 19, 2016)

In this game, Michigan State scored a touchdown that, with an extra point, would have tied the game with about 4:40 left in regulation. Mark Dantonio decided to go for 2 in this situation. When asked about it later, his rationale was that he planned to be aggressive late in the game if the situation arose.

This was way too early to be that aggressive. There are a number factors that play into the pros and cons of going for 2 in this situation, all of which suggest that this was the incorrect decision.

If the Spartans had successfully converted, they would have been up by 1 point with over 4 minutes left in the game. The probability that no more scoring occurs at that point is pretty small. Being up 1 at that point is only meaningful if both teams score field goals or touchdowns in the final 4 minutes. In the case of trading touchdowns, Ohio State would then be forced to go for 2 that late in the game. The larger problem with being up 1 with that much time is that Ohio State’s incentives have changed. If Ohio State is down by 1 instead of tied, they now have incentive to get in position to at least kick a field goal. They are much more likely to go for it on fourth and short to keep a drive alive instead of punting and settling for overtime. Being up by 1 with that much time remaining may not actually improve the Spartans win probability – it may actually decrease it with that much time left.

If the Spartans fail to convert (as they did), they are down 1 in a situation wherein they must either kick away and play defense or try for an onside kick (or inexplicably kick short). If Ohio State gets the ball, they have a decent probability of scoring at least 3 points, which means the Spartans have to score a touchdown to win the game with roughly two minutes or less to play. If Ohio State does not score, they will likely punt the ball and leave the Spartans with (at most) two minutes to get at least a field goal.

This last scenario is exactly what happened. The Spartans had about two minutes to get into field goal range starting from their 25 on the touchback. Gaining roughly 50 yards in two minutes is not all that difficult, but the probability of getting into field goal range AND making the field goal is lower than all of the other scenarios that might have played out by kicking the extra point after the earlier touchdown.

Ultimately, the Spartan 2 point attempt was unsuccessful. One must critique these decisions on their merits rather than the outcome. The outcome was bad, but that does not always mean the decision was incorrect. In this case, the decision was incorrect because of the amount of time left in the game AND the impact of a successful conversion on Ohio State’s incentives. Stated another way, the small upside was greatly outweighed by large and various downsides.

Indianapolis Colts at Green Bay Packers (November 6, 2016)

After the offense struggled most of the game, Aaron Rodgers kicked it into gear late in the 3rd quarter. His touchdown pass to Davante Adams with 7:40 left in regulation brought the score to 31-19. The normally conservative McCarthy decides to go for 2 here. Again, the decision needs to be analyzed without knowing the outcome.

If the Packers kick the extra point, they are down 11 with 7:40 to go. If they manage to prevent the Colts from scoring again (big if), it is still a two possession game. They can try for a 2 point conversion on a subsequent touchdown if it happens and is still necessary.

If the Packers successfully make the conversion, they are now down 10 points with 7:40 to go. They can tie it up with a field goal plus a touchdown and PAT assuming the Colts do not score again. Again, that is a big assumption given the way the game was playing out. It is too early to be worrying about that 1 point when you are already down 12.

Materially, they are in the same position whether they kick the extra point or successfully convert the 2 point play.

However, if the Packers do not convert, they are down 12 points with 7:40 to go. The Packers now have to score at least TWO touchdowns or a touchdown and TWO field goals to tie or take the lead, assuming the Colts do not score again.

The potential downside far outweighs any potential upside.

Ultimately, the decision did not really matter. The Packers never got the chance for a tying or go-ahead score when the Colts ran out the clock on them at the end of the game. They did score another touchdown, after which McCarthy opted for the extra point (to go down 5). Ironically, because of his earlier decision, going for 2 on this second touchdown would have made sense. Being down 5 or 6 is essentially the same, while the potential to be down 4 means an additional touchdown and PAT puts you ahead 3 instead of 1 or 2. That late in the game, you can begin worrying about the costs and benefits of 1 additional point.

Effectively, McCarthy made two probabilistic errors on two point decisions in that game, neither of which actually cost his team the game.

Green Bay Packers at Atlanta Falcons (October 30, 2016)

McCarthy made another bizarre two point conversion decision in this game. The decision could have cost his team the game, but Rodgers wasn’t able to get the offense into field goal range. The decision was ultimately moot but incorrect nevertheless.

With 3:58 left in regulation, Rodgers hit Janis for a 7 yard touchdown to go up 30-26. McCarthy opted to go for two here. Let’s analyze the decision without knowing the outcome.

If the Packers kick the extra point, they are up 5 points. If Atlanta scores a touchdown on the ensuing drive, they likely try a two point conversion to go up three. However, that attempt may fail. Either way, the Packers can tie with a field goal (if Atlanta tries and makes the two point attempt) or win with a field goal if Atlanta kicks the extra point or fails on a conversion attempt. More importantly, the Packers have forced the Atlanta coaches to make the unenviable decisions and the Atlanta players to execute.

If the Packers go for the two point conversion and make it (which they did), they are now up 6 points. The only way being up 6 rather than 5 helps is if Atlanta happens to score a touchdown and miss their extra point, in which case they are tied with a chance to win with another score. That’s a pretty small probability (unless their kicker is Blair Walsh). They obviously won’t go for two in that situation (as detailed in the previous paragraph). Regardless of outcome here, the Packers still have to kick a field goal to win (or not lose) in regulation.

Again, the downside to this decision is the real problem. If the Packers miss the two point conversion here, they are up only 4 points instead of 5. If the Falcons score again and kick the extra point, the Packers now have to kick a field goal to TIE rather than WIN the game. Obviously the probabilities become muddied after that, assuming the game goes to overtime if the Packers make a field goal.

The decision was incorrect even though they successfully converted. Ultimately, win probability is only slightly higher by kicking the extra point, but slight differences in probability are important.

Green Bay Packers at Arizona Cardinals (January 16, 2016)

Plenty has been written about this game. Barnwell’s article on the playoff games that weekend pretty much sums up the argument with more rigorous probabilities, so I won’t go into as much detail here.

The moral of the story is that the Packers had not one, but TWO, Hail Marys on the same drive to be in a position to win or tie the game. That alone should embolden you to go for two in this scenario. Barnwell lays out the probabilities – you’ve got a slightly better probability of winning if you go for two here than if you kick the extra point and go to overtime.

His argument is the correct one – the Packers were clearly the inferior team that day. If you are the inferior team, you have the best chance of winning a shorter game. Going to overtime lengthens the game, thus making it more probable that the better team will ultimately win. That is why the best team usually wins playoff series in all other major professional sports. The worse team can get lucky once or twice, but in the long run the better team wins the series.

McCarthy made some reference to the idea that he thought his team had the momentum. If that is the case, you don’t think Aaron Rodgers, your MVP Super Bowl-winning QB, can get two yards?

The issue here with most coaches, including McCarthy, is that they have NO SYSTEM. They spout off about momentum, being aggressive, etc. But they really have no system in place, except to delay losing as long as possible.