The Myth of Momentum

Thursday December 11th, 2014

By Keith Goldner

numberFire.com

Everyone wants to get hot at the right time.

With the feel-good stories of the Giants taking down the Patriots in 2011, and the bottom-seeded Packers winning the Super Bowl in 2010, conventional wisdom holds that it is extremely important to finish the NFL regular season strong. Despite coaches resting players in Week 17, many still believe that momentum entering the playoffs is imperative.

We’re here to debunk that myth by answering the following questions: Does a winning streak entering the playoffs mean a team is more likely to win a playoff game? Multiple playoff games? Reach the Super Bowl? Win the Super Bowl?

Of course, we need to be careful. Better teams are more likely to have longer winning streaks and are also more likely to win games, including the Super Bowl. So, throughout the analysis, we will need to account for factors like team strength, home field advantage, and byes.

To evaluate momentum, we’ll look at basic statistics (like winning streaks) as well as our own internal efficiency metric at numberFire.com, Net Expected Points (NEP). A quick refresher: NEP compares every single play over a season to how a league-average team should perform on that play. Every situation on a football field has an expected point value; that is, how many points an average team would be expected to score in that situation (given down, distance-to-go and yard line). For example, the Chiefs may be playing the Steelers, facing a third-and-2 on the 50. That’s a ton of variables, but numberFire has data from the past dozen years of every single play, so most situations have come up at least once. According to our data, an average team may be “expected” to score 1.23 (estimated number) points on that drive. However, Jamaal Charles reels off a 32-yard run to bring the Chiefs into the red zone, increasing the “expected” point value of the next play to 4.23 points (still an estimated number). Jamaal Charles then gets credit for the difference, in this case 2.96 points, as his NEP. That’s Net Expected Points.

Winning In The Playoffs

Misleading Statistic: Since 2004, teams on at least four-game winning streaks won 60.9% of their first-round games.

Since 2000, such teams won exactly 50.0% of their playoff games.

We can cherry-pick any statistic we want to try to prove a point, but without doing a full study and removing all biases, we won’t know if there is true causality.

To determine if momentum has an effect on various outcomes, we use linear regression. Without getting into the mathematical minutiae, we will use winning streaks and the numberFire Momentum Indicator (MI) as predictors for each possible outcome (winning the Super Bowl, winning a playoff game, etc.).

So, essentially, we are using math to answer questions like: Does a team that ends the regular season with a 4+ game winning streak have a better chance of winning the Super Bowl than those not on a streak—adjusted for team strength?

To adjust for team strength, home field advantage, byes and other factors, we use our internally calculated win probabilities from NEP. This does not include any momentum, so if momentum is predictive, we will see it in the momentum variable.

For every single regression we ran, our team strength variables were significant. That means our efficiency ratings are predictive of playoff performance (as they should be).

The numberFire Momentum Indicator looks at how many games each team should have won (based on our internal win probabilities) in their final four games versus how many games they actually won. For instance, a team may have won three out of their last four, but was only projected for 2.2 wins. That would result in an MI of +0.8.

We will use something called the p-value, which we will denote as P, to determine if our winning streaks add value. For a variable to be significant, P needs to be less than 0.05. There are different thresholds of significance, but 0.05 is pretty standard among statisticians (many times statisticians use 0.01 or even 0.001).

Winning any single game in the playoffs: Momentum is NOT a factor

P: All above 0.05

Winning Week 17: 0.40

2-Game Win Streak: 0.97

3-Game Win Streak: 0.98

4+ Game Win Streak: 0.76

numberFire Momentum Indicator: 0.38

Total playoff wins: Momentum is NOT a factor

P: All above 0.05

Winning Week 17: 0.08

2-Game Win Streak: 0.42

3-Game Win Streak: 0.94

4+ Game Win Streak: 0.45

numberFire MI: 0.45

Winning week 17 is the closest thing we found to predictive, but this is actually misleading. Total playoff wins aren’t a great indicator of playoff performance because the best teams (those with byes) do not have the opportunity to win as many games as those without byes. Since it is often the best teams that have locked up byes, they rest their players in Week 17 (making them more likely to lose). So, it makes sense that a team winning in the final week could mean more total playoff wins, simply based on opportunity.

Reaching and Winning the Super Bowl

So momentum does not help on an individual game basis, but what about the Giants and Packers and their respective win streaks? Does momentum predict a team reaching or winning the Super Bowl? Instead of specific win probabilities, we used our efficiency ratings from NEP to adjust for team strength.

Misleading Statistic: Since 2000, 15 conference champions and seven Super Bowl champs won at least their last two games in the regular season.

There have been 80 teams to enter the playoffs on at least a two-game winning streak. If we assume all teams have an equal chance of winning the Super Bowl—while this obviously is not true, it works for this exercise—we would expect 13 of those two-game winning streak teams to be conference champions and seven to win the Super Bowl.

Reaching the Super Bowl: Momentum is NOT a factor

P: All above 0.05

Winning Week 17: 0.70

2-Game Win Streak: 0.58

3-Game Win Streak: 0.99

4+ Game Win Streak: 0.38

numberFire MI: 0.57

Winning the Super Bowl: Momentum is NOT a factor

P: All above 0.05

Winning Week 17: 0.44

2-Game Win Streak: 0.98

3-Game Win Streak: 0.58

4+ Game Win Streak: 0.43

numberFire MI: 0.36

The next time you feel like your team needs positive momentum heading into the playoffs, think again. A weak finish in the regular season means little, as long as those losses don’t affect playoff seeding, byes or home-field advantage. After all, six out of the last eight Super Bowl winners lost one of their final two regular-season games. Yes, that is another misleading statistic.

Keith Goldner is the chief analyst at numberFire.com. Follow him @keithgoldner.


Follow The MMQB on Facebook, Twitter and Instagram.

[widget widget_name="SI Newsletter Widget”]

mmqb-end-slug-square
SI Apps
We've Got Apps Too
Get expert analysis, unrivaled access, and the award-winning storytelling only SI can provide—from Peter King, Tom Verducci, Lee Jenkins, Andy Staples, Grant Wahl, and more—delivered straight to you, along with up-to-the-minute news and live scores.