Two-Point Conversion Decisions: The Break-Even Math and the Charts

A Decision With a Right Answer

After a touchdown, an offense faces a choice that looks like a coin flip but is actually a math problem: kick the extra point (PAT) for one near-automatic point, or run a single play from short range to try for two. For decades the kick was the default and going for two was treated as a gamble or a desperation move. Analytics flipped that framing. With a few honest probabilities, the two-point decision becomes one of the cleanest expected-value calculations in football - and the answer is "go for two" far more often than tradition assumed, especially once the score and time are accounted for.

The Break-Even Math

Start with expected points, the simplest version of the decision. Each option's expected value is its payoff times its probability of success:

Setting these equal gives the break-even conversion rate - the two-point success probability at which the two options are worth exactly the same:

The logic: if the extra point were automatic (probability 1.0), you would need to convert the two-point try at least 50% of the time for it to be worth as much in expected points. Because the PAT is not perfectly automatic - it is a real kick that is occasionally missed - the break-even bar sits a little below 50%. The lower the PAT success rate, the lower the two-point conversion rate you need to justify going for two.

This is precisely why the math shifted when the league moved the extra-point kick back to a longer distance. A longer PAT is missed more often, which lowers EV(PAT), which lowers the break-even bar for the two-point try - making "go for two" more attractive on pure expected value than it was when the kick was a near-formality from point-blank range.

To see the mechanics with clearly hypothetical numbers (for illustration only - not real-season figures):

From Expected Points to Win Probability

Expected points is the right lens early in a game, when one point here or there is fungible and you simply want to bank the most points over the long run. But football is not won on cumulative expected points - it is won by being ahead when the clock hits zero. Late in a close game, the decision should be made on win probability, and that can recommend the opposite of what raw expected value says.

The reason is that points are not linear in value. Specific margins matter enormously because of how football scoring is structured around 3-point field goals and 7-point touchdowns. Being up by exactly 3, or down by exactly 1 instead of 2, can be worth far more than the half-a-point expected-value difference suggests, because it changes what the opponent needs to do on their next possession.

This is what the well-known "go for two" charts encode. These charts - a coaching staple long before analytics formalized them - map the score differential after a touchdown to a recommendation, telling a coach when to kick and when to go for two based on getting to the most valuable margins. The analytics versions replace rules of thumb with win-probability models, but the spirit is the same: chase the margin that maximizes your chance to win, not the one that maximizes expected points.

When Trailing Situations Flip the Decision

The clearest cases where win probability overrides expected value come when a team is trailing and specific point margins unlock or foreclose comebacks. A few canonical situations illustrate the logic:

• Down 8, score a touchdown (now down 2): The textbook chart says go for two now rather than later. If you convert, you tie and force overtime conditions in your favor; if you miss, you still know early that you need another score, and you have time to plan for it. Kicking to go down 1 leaves you needing a two-point try eventually anyway - so you would rather learn the result of the two while there is time to react.
• Down 14, two touchdowns to come: Going for two on the first touchdown lets you "play the math early." Convert, and the second touchdown plus a PAT wins it; miss, and you can go for two again on the second to tie. Taking the two-point attempt first gives you the information when you can still adjust.
• Down 5, score a touchdown (now up 1): Going for two to push the lead to a field-goal-proof margin of 3 can be worth more than the safer 2-point lead, because it forces the opponent to score a touchdown rather than a field goal to retake the lead.
• Late and need two scores regardless: When you will need a two-point conversion at some point to win or tie, attempting it earlier is almost always better, because a failure leaves you the most remaining time to chase the points another way.

These are exactly the spots where a coach who manages only to expected points - or to the old instinct of "take the safe point and figure it out later" - leaves win probability on the table. The trailing-team math is the part of the two-point decision where analytics most often disagrees with tradition.

Caveats and How to Read It

• Conversion rates are team-specific. The break-even math uses a probability of converting the two. A team with a power running game or an elite quarterback near the goal line may clear the bar comfortably; a struggling short-yardage offense may not. League averages are a starting point, not a team's true rate.
• The charts assume average rates. Published "go for two" charts are built on leaguewide conversion and kick rates. A coach who knows his own offense is well above or below average should adjust the recommendation accordingly.
• Game state dominates late. Time remaining, timeouts, and which team has the ball next all feed the win-probability calculation. A chart that ignores the clock can mislead in the final minutes.
• Variance is the point, not a flaw. Going for two is higher-variance than kicking. When trailing, that added variance is usually good - you need swings to come back; when protecting a lead, the safer point is often correct for the same reason.
• It connects to the rest of aggressive decision-making. The same win-probability framework drives fourth-down calls; two-point decisions are one branch of the broader move toward maximizing win probability over comfort.

The bottom line

The two-point decision is an expected-value problem at heart: kicking is worth one point times the PAT make rate, going for two is worth two points times the conversion rate, and the break-even conversion rate is just the PAT success rate divided by two - a little under 50%, and lower the more often PATs are missed. That math alone makes going for two more attractive than tradition assumed, especially since the longer extra point lowered the kick's value. But expected points is only the right lens early; late in close games the decision should follow win probability, because reaching specific margins around field goals and touchdowns matters far more than the fractional point difference. That is what "go for two" charts encode, and it is why trailing teams should often take a needed two-point try as early as possible - converting the information value while there is still time to adjust. Use expected value early, win probability late, and always your own team's real conversion rate rather than the league average.

Further reading

• The Fourth-Down Revolution - the same win-probability framework applied to going for it
• Win Probability and Leverage - how the live percentage that drives late decisions is built
• Special Teams Analytics - where the extra-point and field-goal probabilities come from
• All NFL Analytics Explainers - the full library of metric guides
• nflfastR - open play-by-play data for computing conversion and kick rates yourself