Same Decision, Different Results

Within about thirty seconds of real time, both two NFC North teams were faced with the same dilemma: 4th-and-1 in "field goal range," up three points with four minutes left in the game.

1. Detroit Lions: The Lions faced off against the Redskins on Sunday and went for it from the 12-yard line. Matt Stafford kept it and went up the middle, gaining two yards and converting. The decision here is pretty straight forward from a numbers perspective - especially on a 4th-and-1 - but so many coaches make the wrong choice.


4th-and-1 converts at 74%. If the Lions succeed, their win probability jumps to 87%, if they fail it drops to 81% as the Redskins would still need to move the ball from deep in their own territory. While a field goal from the 12 is almost a sure thing, the difference in win probability between a success and failure is negligible for one main reason: being up six points without the ball in today's NFL is worse than being up three with the ball - and sometimes, without the ball. Since 2000, there have been almost 50 similar situations when a team is up three points and coaches predominantly kick the field goal:
I've written about similar situations in each of the last two weeks, and don't want to beat a dead horse, but being down six forces teams to be more aggressive, and as a result, more efficient. Once inside "field goal range," coaches are about 1-1.5% more likely to kick a field goal in this situation for every yard closer to the end zone.
2. Green Bay Packers: The Packers faced the same situation not thirty seconds later, but faced it at the Cincinnati Bengal's 30-yard line. We can add the punt into this analysis if it is a particularly conservative coach, which would result in an 80% win probability for the Packers. Here, the decision is even more clear from a field goal perspective - since 47-yard field goals only convert at 62%, going for it will result in an expected win probability increase of 15% over a field goal attempt. 
What's interesting here is the Packers chances of winning are even higher than the Lions if they convert, even though the Lions would be much more likely to score a touchdown. This is just a result of having more field to play with, which equals more first down opportunities, which leads to a greater chance of running out the clock. The Packers, like the Lions, did the right thing and went for it, but Jonathan Franklin fumbled fighting for extra yardage when it did not look like he would convert, leading to a Cincinnati game-winning defensive touchdown. After the play, the Packers win probability dropped to 25%, the biggest play in the game.
Despite the Packers losing, that doesn't made they made the wrong choice. In fact, it is refreshing to see both of these coaches make the win-maximizing decision when given the opportunity.

Keith Goldner is the Chief Analyst at - The leading fantasy sports analytics platform - and creator of Drive-By Football.  Follow him on twitter @drivebyfootball or check out numberFire on Facebook

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

15 Responses to “Same Decision, Different Results”

  1. Anonymous says:

    Shouldn't the "break even" for both choices be 0% because the tables show that failing the 4th down leads to a higher and equal chance to win (respectively) than making the field goal?

  2. Anonymous says:

    this may be ancient territory i'm covering, and im sure the chances give slight sway in any case given the scarcity of these events, but are the possibilities of a fumble/turnover return touchdown built into the win probability of the fail scenario?

    because a naive way of looking at the numbers is to say, well worst-case scenario of going for it (presumably turnover on downs) still produces a favorable probability, but it's not true when true worst-case scenario involves, as you put it, "the game-winning defensive touchdown."

    i imagine a lot of coaches don't make the optimal move up 3 for the same reason they aren't aggressive down 3, because "putting the ball in play" assumes a greater risk of this disaster outcome. so, for sake of comparison with the coach's mindset, i just wonder if the calculator takes the same absolute risk into account.

    of course, any field goal try implies some opportunity for a block/miscue return td as well, but i'm not sure if the odds cancel out.

  3. Anonymous says:

    I doubt the risk of a turnover returned for a touchdown is less for a fourth and short compared to a field goal attempt.

  4. John Black says:

    Do this take into account that what happened in the past has nothing to do with what happens now? That was sarcasm because I know for a fact that in football it is an changing game and can not be predicted. coaches and players are in the trenches who make decisions with instinct and gut feel which is how it has to be since all times. I know this aint a grammar blog but I think instead of "I've wrote about similar situations", you you meant "I've written about similar situations".

  5. Anonymous says:

    Still, how much higher of a % chance of victory does McCarthy have with a QB sneak than a handoff to an undersized rookie RB?

  6. Unknown says:

    Great article Keith. Even without the WP Calculator, just watching the game I had no doubt Green Bay made the right call going for it.

  7. Unknown says:

    Great article Keith. Even without the WP Calculator, just watching the game I had no doubt Green Bay made the right call going for it.

  8. Anonymous says:

    @matt of course. i assumed the risk of turnover was greater for a fourth and short than a field goal attempt. that was the gist of the question. but it's also possible that that risk is small enough to not be a significant enough factor to swing the win probabilities any from what ordinary failures (no gain/missed kick) include

  9. Unknown says:

    There are so many red flags in this analysis I'm not sure where to start. Just focusing on the Green Bay 4th down, the statement that starts with "the Packers chances of winning" requires a huge and ill advised leap of faith. If I understand this correctly, what the model produces is the percentage of NFL teams that have gone on to win or lose in this situation. This doesn't take into account that Green Bay was handing the ball to a rookie running back who before that snap had 12 NFL carries under his belt. It doesn't account for Green Bay's kicker kicking very poorly coming into the year and this would have been his longest attempt of the year. The 22 players on the field aren't even considered in the outcome. The model results are interesting and informative but based on my casual observations so far this year are poor at predicting the winner in close games.

    Did anyone see coach McCarthy's Monday press conference? He was asked about this decision and acknowledged that he had information from his research department about the odds but went on to say that you have account for the personnel on the field when you make decisions.

  10. Anonymous says:

    Whatever happened to your "Game Win Probabilities" that used to be published in the now-gone Fifth Down on the New York Times website?

  11. Anonymous says:

    i'd still really like to know if "wp fail" takes into account defensive turnover/td or if it just assumes a turnover on downs

  12. Brian Burke says:

    That's a good point. WP Fail does not anticipate a turnover return for a TD. However, it does anticipate a turnover, because failure to convert is a turnover by definition.

  13. J.D. Krull says:

    The analysis of the Detroit game implies that they would be better off taking a knee than kicking the almost-guaranteed short field goal. That's hard to believe. That combined with the explanation that "being down six forces teams to be more aggressive, and as a result, more efficient" makes me think there is a bug in this system. Couldn't a team, in theory, behave exactly the same down three as down six? Thus, the underlying data factors in inefficient coaching decisions, but is used to evaluate whether coaches make efficient decisions. I'm wondering if the data needs an extra layer of finessing.

    That said, great work Brian!

  14. Brian Burke says:

    JD- They could but they don't. I would think it's harder to believe coaches play perfectly efficient football.

  15. J.D. Krull says:

    Think about it like a chess game. In chess, a player’s choice of the optimal move assumes that the opponent will also make HIS optimal move. The analysis is weaker if you assume that the opponent only makes an “average” move, which is apparently what the WPA analysis does.

    But I’m sure adjusting the data to assume optimal decisions becomes an order of magnitude harder to do, and even without such adjustments, the evaluations are right the vast majority of the time. Maybe as this type of analysis becomes more popular, the next frontier will be adding refinements to assume optimal decisions in all cases…and maybe some of what Mike talked about (factoring in the personnel on the field instead of assuming generic actors). Even with its modest limitations, you've come up with some spectacular analysis.

Leave a Reply

Note: Only a member of this blog may post a comment.