With 7:15 left in the 4th quarter against DEN, KC's Knile Davis ran for a 4-yard TD, narrowing the Broncos' lead 21-16 pending the extra point or two-point conversion. Andy Reid elected for the extra point, and following the kick the Chiefs trailed by 4 points rather than 3 or 5 points resulting from a two-point try.
NFL coaches typically adhere to what's known as the Vermeil Chart for making two-point decisions. The chart was created by Dick Vermeil when he was offensive coordinator for UCLA over 40 years ago. It's a very simple chart that simply looks at score difference prior to any conversion attempt and does not consider time remaining, with one caveat. It applies only when the coach expects there to be three or fewer (meaningful) possessions left in the game.
With just over 7 minutes to play, there could be at most three possessions left, especially considering that at least one of those possessions would need to be a KC scoring drive for any of this to matter. (In actuality, there were only two possessions left, one for each team.) Even the tried-and-true Vermeil chart says go for two when trailing by 5. But it's not the 1970s any more and this isn't college ball, so let's apply the numbers and create a better way of analyzing go-for-two decisions.
Except for rare exceptions I've resisted analyzing two-point conversion decisions with the Win Probability model because, as will become apparent, the analysis is particularly susceptible to noise. Now that we've got the new model, noise is extremely low, and I'm confident the model is more than up to the task.
First, let's walk through the possibilities for KC intuitively. If KC fails to score again or DEN gets a TD, none of this matters. Otherwise:
Kick the XP:
- if DEN doesn't score, need a TD to win
- if DEN scores a FG, need a TD to tie
- if DEN doesn't score, need a TD to win
- if DEN scores a FG, need a TD plus a two-point conversion to tie
- if DEN doesn't score, need a FG to tie, TD to win
- if DEN scores a FG, need a TD to win
Every permutation but one points in favor of going for two. Only the scenario where DEN gets a FG does it hurt to go for two, and even then KC can still achieve the tie with a second two-point conversion. But we can go around in circles all day debating what-ifs and what counts as a "meaningful" possession. What does the model say?
Let's find the breakeven probability of success that would make the conversion attempt worthwhile. If the breakeven is below the typical success rate of just over 45%, it would generally be a good idea to go for two.
A few definitions:
B = the breakeven probability of success
XP = KC's win probability by kicking the extra point
S = KC's win probability after a successful conversion
F = KC's win probability after a failed conversion
Setting the WP for kicking the XP equal to the 'lottery' of going for two, we get:
Solving for B gives us:
In other words, the breakeven probability is the ratio of two differences of win probabilities: the difference between the XP and a failed conversion, and the difference between a successful conversion and a failed conversion. This makes intuitive sense because as the benefit of the XP diminishes, the breakeven gets lower. And as the benefit of a successful conversion decreases, the breakeven gets higher.
Just the smallest bit of statistical noise can hurt the analysis. When either of the differences mentioned above are very small, the breakeven probability can swing wildly. Noisy inputs can make the denominator very small, creating very large leverage in the result. Fortunately, with new more advanced modeling techniques WP 2.0 offers, we can get smooth, stable results as the chart below will illustrate.
In KC's situation, the numbers look like this:
XP = 0.273
S = 0.328
F = 0.250
Putting those numbers through the meat grinder gives us a breakeven of 32%, much lower than the league-wide average of 45%+. So the model supports the Vermeil chart (in this case) as well as our intuition. It also agrees with the Krasker chart [link currently down] and the Sackrowitz chart.
Here is what the breakeven probability success is for a score difference of -5 points throughout the second half. As you can see it's almost always below the league average rate, and plummets in the end-game. Krasker and Sackrowitz both agree on this result as well. The red dot indicates the KC-DEN scenario.
A couple final notes.
1. What eventually happened is the only thing that really makes a failed conversion attempt costly: DEN kicked a field goal on their last possession. Don't be suckered into thinking that makes the XP the right decision. You can only evaluate decisions based on the information available at the time.
2. The difference in terms of absolute WP between the XP and the conversion attempt was only 1.2 percentage points. You might think that's tiny, and in the grand scheme you'd be right. But coaches stay up watching film until their eyes bleed, looking for every last 0.1% edge. It's not a good idea to toss away a percentage point here and a percentage point there. Analyses like this might only matter 1 or 2%, but they add up fast, even in a single game. And before you know it, you can turn a 7-win team into a playoff team, just with some better decisions.
3. I've created a complete model for two-point conversions, covering every possible scenario including timeout considerations. Hopefully, there will be some interesting situations to look at this season!
Search here for over 1,500 research and analysis articles at AFA and the AFA Archives.