Great Playoff Runs

by Vince Verhei

In the Super Bowl XLII Audibles discussion [1], I raised the question of whether the Giants had faced the most difficult schedule of any Super Bowl champion. To review, the Giants beat the following teams to win the Super Bowl: Tampa Bay, Dallas, Green Bay and New England. What are the odds that the Giants, or any other team, would have completed that four-game sweep?

To a degree, we can answer that question with two tools: The Pythagorean theorem and the log5 method. Like many other sports statistics, these were both originally developed by baseball statistician Bill James, then used by others to examine other sports. In a nutshell, the Pythagorean theorem uses each team's points scored and allowed to predict that team's winning percentage, while the log5 method predicts the odds of one team defeating another, taking the strength of both teams into account. More on the theorem can be found by reading this article [2] on the pro-football-reference.com blog. (I should add that all numbers in this study were taken from PFR [3].) More on the log5 method can be found in this article [4] at Diamond Mind Baseball.

There are two ways to determine the difficulty of the Giants' playoff slate. We can estimate the odds of of the Giants running the table, but that will skew the results; the Giants finished the regular season with a Pythagorean rating of just .536, the lowest of any Super Bowl champion. (The highest rating of any Super Bowl champion belongs to the 1985 Chicago Bears at .8784; the 1975 Pittsburgh Steelers are right behind at .8783.)

Instead, we're going to measure each Super Bowl winner's playoff schedule in two ways: the odds of that particular team running the table, and the odds of a .750 team (about the average Pythagorean rating for a Super Bowl winner) pulling off the same feat. There are plenty of factors that could throw this number off -- injuries, home-field advantage, strength of schedule, overall league strength, etc. -- but it should be fairly accurate for most teams.

(I should also add that this is not the prediction method used elsewhere on this site, including the Playoff Odds Report, which uses Weighted DVOA and does account for home-field and other factors.)

The following table lists the Giants' four opponents, the Giants' odds of beating that team, and the "Typical" Super Bowl winner's odds of beating that team:

So the Giants had a 41.1 percent chance of beating Tampa Bay, a 34.2 percent chance of beating Dallas, a 30.8 percent chance of beating Green Bay, and merely a 15.8 percent chance of beating New England. (Note that these numbers were calculated using New York's final regular season Pythagorean rating. We could add the results of each progressive playoff game, but A) the numbers would change very little, and B) we're going to compare New York to other Super Bowl winners, whose ratings would also improve as the playoffs progressed.) Multiplying all those percentages leaves us with a very small number; we would predict that the Giants had only 0.7% chance of winning the Super Bowl against that schedule. (That is not a typo: 0.7%, as in, if they faced that schedule 1,000 times, they would win about seven Super Bowls.) This is the lowest expected rate for any Super Bowl winner going into the postseason. The fact that the real-life Giants actually did win the Super Bowl does not necessarily mean they were "lucky" to do so. It means that faced with a longer and more difficult road than any Super Bowl winner before them, they still emerged triumphant.

Our "typical" Super Bowl champion, facing that same schedule, would have about a 6.5 percent chance of winning the Super Bowl. That makes this a very difficult slate, but not the most difficult of all time. That honor goes to John Madden, Ken Stabler, and the rest of the 1976 Oakland Raiders, which is all the more remarkable given that the Raiders only played three playoff games.

First, the Raiders beat the Steve Grogan/Sam Cunningham Patriots (Pythagorean rating: .751) 24-21 in the divisional round, the famous "Sugar Bear Hamilton roughing the passer" game. Next, the Raiders squared off against the Pittsburgh Steelers, whose Steel Curtain defense was at its absolute peak. The defending champion Steelers had started 1976 very slowly, just 1-4. Then they won their final nine regular season games, and only once in those nine games did they allow seven or more points. They pitched five shutouts, including three in a row. They finished with a Pythagorean rating of .896 -- yes, even higher than the 2007 Patriots. They whipped the Baltimore Colts 40-14 in the opening round of the playoffs, but in the process lost both of their leading rushers, Franco Harris and Rocky Bleier, to injury. Without their ground game, they were no match for the Raiders, who won the AFC Championship 24-7. Oakland then finished off the Minnesota Vikings (Pythagorean rating: .786) to win the Super Bowl. In those three games, the Raiders beat two teams that were typically good enough to win Super Bowls, and in between they beat one of the greatest teams of all time. The "typical" champion, facing this schedule, would have just a 5.8 percent chance of winning the Super Bowl.

Here are the numbers for each Super Bowl champion, ranked by the odds of that particular team winning the Super Bowl against that particular schedule:

A few final notes:

• The last three Super Bowl winners have each been among the five least likely champions of all time. This is largely because they each have had to win four games to pull it off; most Super Bowl winners have only had to win three. The Giants and Steelers each had to beat teams that flirted with perfection, while the Colts beat three teams -- the Ravens, Patriots, and Bears -- with Pythagorean ratings greater than .750.
• The easiest schedule of all time, based on the odds of a typical Super Bowl winner, belongs to the 1979 Pittsburgh Steelers, whose first playoff opponent, the Miami Dolphins, was actually their most difficult, with a Pythagorean rating of .662. They then beat two completely mediocre teams -- the Houston Oilers (.553) and the Los Angeles Rams (.526) -- to win the Super Bowl. The next easiest playoff slates belong to the 1995 Cowboys and the 1999 Rams.
• Winning the Super Bowl is really, really hard. Only two champions -- the 1985 Bears and 1999 Rams -- began the postseason with a 50 percent chance of winning the Super Bowl. One team had an historically great defense, the other an historically great offense. They both faced relatively poor competition. And yet each was basically an even-money bet to not win the Super Bowl. (This year's Patriots would have joined them. They had a 51.7 percent chance of beating the Jaguars, Chargers and Giants. Things didn't work out that way.)
• New England fans can take some solace in the fact that two teams with even higher Pythagorean ratings than the 2007 Patriots have also lost Super Bowls: the 1969 Vikings (.923) and the 1968 Colts (.919) More evidence that public perception of the difference between the AFL and NFL before the merger was really off.
• Going back to the days before the Super Bowl, this was not the first Giants team to pull off a big upset and spoil a perfect season. In 1934, the 8-5 Giants (Pythag: .680) beat the 13-0 Chicago Bears (.945) in the NFL Championship. This became known as the Sneaker Game [5]. Freezing rain overnight left the playing surface at the Polo Grounds as slick as a hockey rink. Trailing 10-3 at halftime, Giants head coach Steve Owen frantically acquired sneakers for his team to wear instead of their football cleats. The extra traction proved to be a huge boost, and Chicago head coach George Halas was never able to adjust. The Giants scored four touchdowns in the second half, winning by a final score of 30-13. The log5 method predicts that the Giants had just an 11.0% chance of winning that game.