Steelers vs. Giants: A Look Back at Some Interesting Coaching Decisions
Big Ben and Eli Manning faced off against each other Sunday, leading the teams that have won four of the last seven Super Bowls; Manning and Roethlisberger will always be linked as the No. 1 and No. 3 QBs taken in the 2004 NFL draft.
Let's look at the plethora of interesting decisions between the coaches.
What first tickled my fancy was when the CBS announcer mentioned that, "Mike Tomlin says he makes all these decisions strictly on feel." This was before the Steelers went for it on 4th-and-1 from the Giants 48-yard line, down 7 in the second quarter. After I shuddered at the quote, I checked out the 4th-down calculator to confirm my notion that this was one of the easier football decisions to be made:
This early in the game, score and time are not as big of a factor (especially in a close game) so let's look at expected points. Going for it, based on a 74% conversion rate, results in an expectation of +1.15 points > -0.22 expected points from punting. Unless you think you can only convert 4 out of 10 times or worse, you definitely go for it. Good gut feeling from Tomlin. Keep in mind that all these probabilities are baselines, but they are the necessary starting point for any sort of decision-making analysis.
Let's go back though. On the first drive of the game, the Giants punted on 4th-and-1 from the Steelers 44-yard line - another no-brainer. Expectation going for it is +1.40 > -0.11 points for punting. Only reason not to go for it that early is if you think you cannot convert at least 38% of the time (about half of the measured league-wide conversion rate).
As we get later in the game, though, time left and score matter more in these fourth down decisions. Up 7 with about a minute and a half left in the 3rd quarter, Tom Coughlin faced another 4th-down decision - 4th-and-Goal from the 5.
If you ask anyone in the NFL, this would be a no-brainer. A field goal is almost a guarantee and it puts your team up two scores. But looking at the numbers, it may actually be the better decision to go for it. 4th-and-Goal converts at roughly 36% from the 5-yard line (the field is smaller so this is lower than elsewhere on the field). Expected win probability going for it is 87%, 2% higher than the 85% win probability of kicking a field goal. In fact, the numbers say to go for it if you believe you can convert one out of five times. The main reason for this is that going up two touchdowns with just more than a quarter left equates to a 96% chance of winning.
The last and most controversial decision came from the Steelers on 4th-and-1 from the 3-yard line. Down 20-17, Mike Tomlin seemed to feel that kicking the field goal was the proper decision:
Similar to before, going for it actually results in a +6% margin in expected win probability (53% vs 47% of kicking the field goal). A field goal only ties the game which does not mean they would win with any certainty. Tomlin trots out the field goal unit. BUT WAIT, A FAKE FIELD GOAL. Steelers pitch the ball to the fastest kicker on their team (I have to assume), Shaun Suisham, who is tackled short of the line-to-gain.
First, I want to give Mike Tomlin credit for actually going for it. That was the right decision. As long as you believe you can convert over 46% of the time (which the league-average is 68%), you go for it. But, I question the decision to fake the field goal. That's a high-risk play to begin with and does it actually convert 46% of the time? I doubt it. The Giants were ready for it and a kicker will rarely outrun a defender in the NFL. That play works predominantly due to its surprise nature. A straight run or QB-sneak most likely converts at a much higher rate.
Obviously, hindsight is 20-20 and I hate to think that if it had converted I would not be criticizing him - there are few things I hate more than evaluating a decision based on the outcome. It's exceedingly difficult to measure the expected success rate of such a fake; in reality, it probably is best measured by a binary variable. What I mean is that the play is most likely successful only if the defense is caught off guard. Will an NFL special teams unit be surprised by a fake more than 46% of the time? Let's try to estimate it anyway. I pulled 264 fake field goal and fake punts out of the play-by-play data since 2000. One note: these are likely not all of the fakes and not all of these may even be fakes because they are not specifically listed that way in NFL gamebooks. That being said, here are the results I found by looking at fakes when the distance-to-go was 2-yards or less. I limited it this way because then defenses are probably more suspicious of potential fakes.
Sample sizes here are pretty small, especially as we go left to right. Fakes convert at a rate of 55.7% from inside 2 yards-to-go. Fake field goals, though, only converted 8 out of 24 times, 7 of which were on fake field goal rushing plays. If we go by the numbers (again, tiny sample size), the fake field goal run converts at 41.2% of the time, which is less than the aforementioned break-even rate at 46%.
Given that the Steelers actually pulled out the victory, this play will ultimately be forgotten completely. If they had not, I'm sure the ridicule for the fake field goal would have been endless.
When in doubt, go strictly based on feel.
Keith Goldner is the creator of Drive-By Football, and Chief Analyst at numberFire.com - The leading fantasy sports analytics platform. Follow him on twitter @drivebyfootball or check out numberFire on Facebook.
Get more great NFL analysis over at Advanced NFL Stats.