SANTA CLARA, Calif. -- The NFL's Harbaugh brothers faced fateful fourth-down coaching decisions during conference championship games Sunday.
Neither decision worked out well, but statistical analysis from ESPN's analytics team favored Baltimore's John Harbaugh over San Francisco's Jim Harbaugh when it came to playing the percentages.
John Harbaugh went for it on fourth-and-6 from the New England 33-yard line with 2:53 left in the fourth quarter of a game the Patriots led, 23-20. Jim Harbaugh opted to punt on fourth-and-1 from his own 31-yard line in overtime. Update: I initially listed the yard line incorrectly on the 49ers' fourth-and-1.
I'll focus on Jim Harbaugh's decision for our purposes.
The win-probability tool ESPN has developed takes into account how decisions in similar situations have affected game outcomes in the past. Comparing the expected fourth-and-1 conversion rate to the expected resulting field position following a punt can produce percentage gains or losses in win probability.
To be fair, the 49ers could have made different assumptions based on things they knew at the time. Football is not a game of blackjack, where the right decision is mathematically demonstrable for every hand. How a coach feels about a certain situation might correctly override what the percentages say in certain situations.
The point here is to consider the options and learn more about the process, not to say Jim Harbaugh definitely erred.
Albert Larcada of ESPN's analytics team calculated the percentages this way:
Going for it. If San Francisco converted with a 1-yard play, its win probability would have been 69.1 percent. If it failed to convert, its win probability would have been 20.9 percent. Since 2001, teams going for it on fourth-and-1 pick up the first down 64.7 percent of the time. Going for it would have carried a win probability of 52.1 percent.
Punting. On the season, San Francisco punter Andy Lee had an average net punt of 44 yards. Using that number, the Giants would be expected to have first-and-10 from their own 25-yard line following the punt, producing a win probability of 48.9 percent.
For the 49ers, the difference between 52.1 percent and 48.9 percent win probability seems insignificant, particularly if the 49ers thought their chances for converting were lower.
The Giants' defense had stopped teams six times in 11 chances on fourth-and-1 plays this season, including five of six times since Week 12. The 49ers had converted twice in four chances on fourth-and-1 plays this season.
In this case, the 49ers knew their defense was playing more effectively in the second half. Pinning the Giants deep and playing for field position did carry appeal. But Lee's punt traveled 47 yards to the 22, and a 14-yard return gave the Giants possession at their own 36. When the 49ers did force a punt, their return specialist, Kyle Williams, lost a fumble.