The Sinking of the Viking Ship.

A Look At the 1998 NFC Championship.
by: John T. Reed
Author, Football Clock Management.
© More from this issue

          
A bunch of readers have asked me to analyze the clock-management decisions in the Vikings-Falcons NFC Championship game held on January 17th. They disagree with two Vikings' decisions, one at the end of each half.

Atlanta won in overtime and advanced to Super Bowl XXXIII. Would better clock management have sent Minnesota to the Super Bowl instead? Yes. With 1:17 left in the first half, Minnesota fair-caught a punt at their own 18. They ran three pass plays. The first two were incomplete. On the third, Randall Cunningham was sacked and fumbled the ball to Atlanta. The Falcons scored a touchdown on their first play after the fumble at :59.

Critics complain that Minnesota should not have been trying to score from their own 18 with only 1:17 left, especially since they were ahead 20-7 at the time. Rather, they should have just run out the clock.

I agree. On your last possession of the first half, you should first decide whether you are going to try to score or just try to kill as much clock as possible. A team that is up 20-7 should try to score only when their probability of success is very high.

Minnesota's probability of scoring from their own 18 with just 1:17 left would have been below 50 percent. To have a greater than 50 percent probability of kicking a field goal, they would have had to get to about the Falcons' 30. That means they had to go 52 yards.

Under optimum conditions (stopping the clock after every play), a team can run about one play every six seconds. With 1:17 left, that means Minnesota could run 13 plays (77÷6). Going 52 yards in 13 plays requires an average gain per play of 4 yards. The Vikings had all three timeouts remaining. They averaged 4.2 yards per play on the day. But can they stop the clock after every play 13 times? Probably not.

What about the risk? As they started the series, Minnesota had a win probability of over 90 percent. (I looked it up in win probability tables created by Hidden Game of Football co-author Pete Palmer.) Adding a score cannot take them all the way to 100 percent, so another score is worth only about 5 percent in terms of increased win probability. On the other hand, what are they risking if disaster strikes? If the Falcons get a turnover and score a touchdown, which is what actually happened, the Minnesota win probability drops by about 20 percent.

The only time you take a gamble where you may end up 5 percent better off or 20 percent worse off is if the probability of the better-off outcome is more than four times the probability of the worse-off outcome. In other words, trying to score rather than killing the clock requires that the probability of the Vikings marching 52 yards and kicking a field goal must be more than four times the probability of the Falcons getting a turnover and scoring.

Could the Vikings have just taken a knee starting at 1:17? No, that's too much time when your opponent still has three timeouts. Could they have just run quarterback-keep-sweep-slide plays starting at 1:17? Those extend the period when you no longer need a first down because the sweep-slide play takes longer than taking a knee. No, they couldn't do that either for the same reason. Also, the Vikings were backed up at their own 18 so they did not have room to run more than one sweep slide, two if they took an intentional safety.

The Vikings needed one first down to get into the take-a-knee period. Had they failed to get the first downs, advancing the ball to, say, their own 25, and punted for a net-30 yards, the Falcons would have first and 10 at their own 45. If Atlanta used their timeouts, as they should, during Minnesota's three downs, the time left would be about :57. If Atlanta had subsequently settled for a field goal or failed to score at all, they would have lost the game.

In other words, the Vikings should have run out the clock instead of trying to score because of their already large lead, their poor field position, and the small amount of time remaining. Had the game been tied, the decision on trying to score before half would have been almost a 50-50 proposition.

Remember also that the coaches did not fumble the ball. Randall Cunningham did. If he had held onto the ball, the Falcons may have scored only a field goal before half and Minnesota would have won the game in regulation. The coaches bear some responsibility for Cunningham's fumble because they called three pass plays, and turnovers are more likely on pass plays. But Cunningham bears most of the responsibility. Hardly anyone would be upset about the three pass play calls if Cunningham had held onto the ball.

The other decision that upset fans was the opposite choice at game's end. The Vikings chose to run out the clock at the end of regulation with the game tied at 27. Their final series started at their own 20 after a Falcon touchdown and touchback with :49 left. Minnesota had all three timeouts.

On first down, Cunningham ran up the middle for 7 yards. Timeout Minnesota at :37. A second-down pass to Moss fell incomplete at :30. Cunningham then ran a sneak that lost a yard. Even though they had two timeouts remaining, the Vikings let the clock run out putting the game into overtime.

Should the Vikings have tried to score in regulation? Again, they needed to get to the Falcons' 30 to have a greater-than-50 percent probability of kicking a field goal. That's 50 yards in :49, enough time for about nine hurry-up plays, thereby requiring an average gain of 50/9 = 5.6 yards per play.

Which is more likely, gaining that 5.6 yards per play over nine plays or winning in overtime? Winning in overtime. The odds in any NFL overtime are 50-50 almost by definition. The chances of gaining 50 yards and kicking a field goal in :49 seconds have to be less than 50 percent. If the Vikes had obtained possession by takeaway, you might argue that they had emotional momentum and should take advantage of it by trying to win. But you don't get much momentum from an opponent is touchdown and touchback.

Mathematically, this was a close call. From an emotional and job-security standpoint, going for the win in regulation is best. Ara Parseghian still smarts from criticism of his decision 33 years ago to settle for a tie against Michigan State in the pre-overtime era. He's more infamous for that game ("Tie one for the Gipper") than famous for winning the national championship that year (in most polls) or for his .836 winning percentage at Notre Dame.

In the business world, people used to say, "Nobody ever got fired for buying IBM." It may also be true that in the NFL nobody ever got fired for trying to win in regulation.

One play that should have been considered by the Vikings at the end of both halves is the fake take-a-knee pass play. You set that play up by taking a knee once or twice, then, on the last play of the half, you behave as if you are going to take another knee, but suddenly throw a deep scoring pass. The play should only be used before half-time or at the end of a tie game.

Bill Walsh ran the fake take-a-knee against the Cardinals in 1982 and Buddy Ryan ran it against the Cowboys once. The play was inappropriate in both of those cases because the teams that ran it were ahead and it was the end of the game. Walsh immediately apologized saying, "I don't know what I was thinking at the time." Russ Francis said the play violated a "gentlemen's agreement" in the NFL. But before half-time or at the end of a tie game, there is no gratuitous running-up-the-score aspect to it. In those two situations, the fake take-a-knee play is the moral equivalent of the fake-spike play Dan Marino used to beat the Jets in '94.

In the '99 NFC Championship game, the fake take-a-knee play (after the Vikings got the first down needed to put them into the take-a-knee period) would not have been smart at the end of the first half because of the Vikings' lead, the risk of an interception, and the danger that it would fire up the Falcons in the second half. But it might have been worth a try at the end of the game because of the tie score. The fake take-a-knee play can only be run when the take-a-knee play itself is plausible before half-time or near the end of a tie game. Those are fairly rare situations, but they both happened in this game.

There was another, bigger clock-management issue in the NFC Championship game. None of the people who asked me to analyze the Vikings' decisions mentioned it.

If a Vikings' coach had asked my opinion of the game, the conversation might have gone like this:

Vikings: We lost the NFC championship in OT. Would better clock management have won it for us in regulation?

Reed: Did your opponent score near the end of either half?

Vikings: Actually, they scored near the end of both halves - at :56 left in the first half and at :55 left in the game.

Reed: Were you ever ahead in either half?

Vikings: Yes. We were ahead for most of the game. We trailed only for about three minutes in the first quarter.

Reed: Did you go to a slowdown when you were ahead?

Vikings: Somewhat in the second half.

Reed: The principles of good clock management require that you go to a maximum slowdown whenever your win probability is greater than 50 percent (with the exception of some last possessions of the first half). It's greater than 50 percent whenever you're ahead or about to take the lead.

Here's how much time the Vikings' drives took when they were ahead, and how much time they could have taken in a maximum slowdown.

Took Could have taken

3:41 6:30
1:28 1:29
0:18 2:20
0:56 0:56

6:23 11:15 1st 1/2 total

Had the Vikings gone to a slowdown when they were ahead in the first half, as they should, the half would have ended 11:15 - 6:23 = 4:52 earlier. That meant the first half would have ended with Minnesota up 17-7, not 20-14. If the scoring in the second half had been the same as it was in the actual game, the final would have been Vikings 24-20.

My "Could have taken" numbers are not exact because NFL play-by-play sheets do not show enough detail. I arrived at the "could have taken" figure in general by assuming 45 seconds per clock-running play, 5 seconds per clock-stopped play, and 25 seconds per clock-stopped-temporarily play. Here are the second-half Vikings-ahead drives:

Took Could have taken

3:21 3:37
6:55 8:35
1:26 1:42
1:12 1:12
4:00 4:35

16:54 19:41 2nd 1/2 total

The Vikings appear to have slowed down in the second half, but they were not "fanatics about it." With a maximum slowdown, the game would have ended about 19:41 - 16:54 = 2:47 earlier. That would have eliminated Atlanta's tying touchdown, which came at :55.

Every second you leave on the clock when you should be in a slowdown may be the one your opponent uses to beat you.

John Reed can be reached at johnreed@johntreed.com.

NEW BOOK!

Digital Download - 304 Pages of Football Forms for the Winning Coach