The Big Mo: Does it Really Exist?

 

(Note: this study was completed in 2010, and covers the 1990-2009 seasons. I plan to update the numbers though the 2012 playoffs, but I don't suspect that will change the conclusions.)

 

Each December, broadcasters, sportswriters, and fans all obsess over whether or not NFL teams have momentum heading into the playoffs. In this past season, the San Diego Chargers were hot, riding an 11-game winning streak into the playoffs. This was considered a good thing. The New Orleans Saints, after starting the season 13-0, dropped their last 3 regular season games and were written off by many analysts. Likewise, the decision of the Indianapolis Colts to rest their starters in their final two games, leading to their only two losses of the season, was widely criticized as costing the team a season’s worth of momentum heading into the playoffs.

 

The fact that the Chargers got bounced from the playoffs in their first game and the Saints and the Colts went to the Super Bowl should give people pause to reconsider their belief that momentum heading into the playoffs is important. It should, but it probably won’t; sportswriters and sports fans often don’t want facts to get in the way of a good belief system. But anecdotal evidence isn’t proof; to investigate whether momentum heading into the playoffs is important, we need to look systematically at a large set of data. That’s what I intend to do in this article.

First, how do we define momentum? The simplest way to define it is this: What is the team’s record in the last few games of the season? For the purposes of this article, I’ll use five games as the cutoff – momentum is defined by how well the team did in the last five regular season games. (Later in the article I’ll discuss what happens if you use a different number of games, like four).  We’ll limit the study to the 1990-2009 seasons, because 1990 was the year the NFL went to the current 12-team playoff system.

Table 1 shows the playoff performance of all teams, based on their momentum.

Table 1 – Momentum of playoff teams, 1990-2009

Momentum

# of teams

Playoff record

SB Wins

SB Win %

Regular Season

0-5

0

0-0

1-4

5

3-5

.375

0

0.0%

43-37-0

.538

2-3

39

28-37

.431

2

5.1%

387-236-1

.621

3-2

87

76-80

.487

7

8.0%

932-459-1

.670

4-1

76

71-70

.504

6

7.9%

860-354-2

.708

5-0

33

42-28

.600

5

15.2%

400-128-0

.758

 

That’s a lot of information, so let’s go over  what each of the columns means. “Momentum” is the record of the team in the last five games of the season (note there was one team that had a tie; for simplicity their 3-1-1 record was counted as 3-2). “# of teams” is the number of times a team entered the playoffs with that momentum. “Playoff record” is the combined playoff wins/losses and winning percentage of the teams with that momentum. “SB Wins” is the number of times a team with that momentum won the Super Bowl. “SB Win %” is how often a team with that momentum wins the Super Bowl. “Regular Season” is the combined full regular season record of the teams with that momentum.

 

What do we notice from this table? First, very few teams head into the playoffs in the middle of a complete meltdown. Since 1990, no one lost their last five regular season games and made the playoffs.  A few teams lost 4 of 5 heading into postseason, and as a whole they didn’t do too well (in terms of Playoff Record and Super Bowl Win Percentage). Teams with two wins out of five did a little better, but still below average. 3-2 and 4-1 teams were better still, and were almost identical.  And teams on a 5-win (or more) streak before the playoffs did the best, winning 60% of their playoff games and winning Super Bowls at nearly twice the rate of any other group.

 

So momentum exists, case closed? Not so fast. Take a look at the last column: the combined regular season records of the teams. Note that, on the whole, the more momentum a team has going into the playoffs, the better the team is. This is not surprising – if you pick any point in the season and find a team with a five-game winning streak, that team will likely be better than teams that didn’t go 5-0 at that point.

 

So it appears Table 1 has a “quality bias” -- the teams with good momentum do better in the playoffs not because they have momentum, but because they are better teams.

 

[Note that we’re assuming that if a team has a better record, it is a better team and will do better in the playoffs. A reasonable assumption, but let’s see how that works out for the playoffs (Table 2):

 

Table 2 – Playoff Performance based on Regular Season Record

Regular Season Record

# of teams

Playoff record

SB Wins

SB Win %

8-8

8

3-8

.273

0

0.0%

9-7

40

23-40

.365

0

0.0%

10-6

59

32-58

.356

1

1.7%

11-5

51

51-49

.510

2

3.9%

12-4

38

49-31

.613

7

18.4%

13-3

29

37-23

.617

6

20.7%

14-2

12

21-8

.724

4

33.3%

15-1

2

2-2

.500

0

0.0%

16-0

1

2-1

.667

0

0.0%

 

Clearly, the better your record in the regular season, the better you’ll do in the playoffs. Well, maybe you don’t want to go 15-1. J]

 

Testing the Theory

 

Again, my theory is that teams with momentum do better in the playoffs not because they have momentum, but because they are better teams. How can we eliminate this “quality bias”? One way is to figure out how many wins we would expect from each category of momentum based only on the quality of the teams, and compare it to their real life performance. If the expected number of wins is equal to the actual number of wins, then momentum has nothing to do with the teams’ performance – it was all based on the quality of the teams.

 

As an example, let’s take all of the teams with perfect momentum heading into the playoffs: 5-0 in their last five games. There were 33 such teams over the period studied. We’ll break these 33 teams in groups based on their regular season records and calculate their expected playoff wins. I’ll do a couple of examples.

 

Six of these teams went 11-5 during the season. From Table 2, we would expect them to win 51.0% (.510) of their playoff games. This group of teams played 13 actual playoff games; therefore we expect them to win 6.63 (51% of 13) and lose 6.37 (49% of 13) of their playoff games (in reality they went 8-5).

 

Nine of these teams went 12-4 during the season; from Table 2, we would expect them to win 61.3% (.613) of their playoff games. This group of teams played 20 actual playoff games; therefore we expect them to win 12.26 (61.3% of 20) and lose 7.74 (38.7% of 20) of their playoff games (in reality they went 13-7).

 

And so on. Doing this for each group of teams, based on their regular season records, and summing up the wins/losses results in an expected playoff record of 40.1 wins and 29.9. In reality, their record was 42-28, slightly better than expected. So the 5-0 teams won two more games than we would have expected based merely on their regular season records. Two wins out of 70 games is not much of an advantage, and is well within random fluctuation.

 

Doing this for each group of teams, and converting to win/loss percentages instead of raw wins and losses, we get the following (Table 3). Note that I’ve also included the same set of calculations for Expected and Actual Super Bowl wins.

 

Table 3 – Playoff Performance: Actual vs. Expected

Momentum

Expected Playoff W/L %

Actual Playoff W/L %

Difference (W/L %)

Expected SB Wins

Actual SB Wins

Difference (SB Wins)

0-5

 

 

1-4

.330

.375

+.045

0

0

0

2-3

.428

.431

+.003

1.2

2

+0.8

3-2

.480

.487

+.007

6.2

7

+0.8

4-1

.529

.504

-.025

8.1

6

-2.1

5-0

.573

.600

+.027

4.5

5

+0.5

 

If there was an advantage of having momentum, then you would expect the Difference columns to increase as the Momentum increased. For Playoff W/L percentage, the values are pretty flat, with a small bump down at 4-1 and a small bump up at 5-0. For Super Bowl wins, the values are all within 1 (which is as close as they can get), except for the teams with 4-1 momentum. But we’re talking about a very small number of events; expecting 8 wins and getting 6 is not particularly significant. Still there does seem to be a small disadvantage to being 4-1 going into the playoffs, even as compared to teams with less momentum. Puzzling!

 

The key thing to take away from Table 3 is that if momentum really existed, you would expect the Difference columns to have large negative numbers for the poor momentum teams, and large positive numbers for the high momentum teams. We really don’t see that here.

 

 

There’s Momentum, and then there’s Momentum

Our definition of momentum could be improved. For example, take two teams with “3-2” momentum. One team’s last five games are, in order

 

W W W L L

 

The other team’s last five games are:

 

L L W W W

 

Clearly their momentum is different, but in this system they are both “3-2” teams.

 

To resolve this, let’s define Momentum Points (MP). You get five MPs for winning the last game of the season, four for winning the next-to-last, three for winning the game before that, two for the game before that, and one for the game before that. So a 5-0 team would get 15 MPs. In the examples above, the first team would get 1+2+3=6 MPs, and the second team would get 3+4+5=12 MPs. This more clearly distinguishes momentum. Maybe now we’ll actually see some advantage to momentum?

 

Here is a table of the performance of teams based on Momentum Points. It is the same format as Table 1, just using MPs instead of the record in the last five regular season games:

 

Table 4 - Momentum of playoff teams, based on Momentum Points

Momentum Points

# of teams

Playoff record

SB Wins

SB Win %

Regular Season

0

0

1

1

0-1

.000

0

0.0%

8-8-0

.500

2

0

3

6

7-5

.583

1

16.7%

60-36-0

.625

4

6

3-6

.333

0

0.0%

57-39-0

.594

5

10

5-10

.333

0

0.0%

99-60-1

.622

6

9

8-9

.471

0

0.0%

94-50-0

.653

7

20

19-19

.500

1

5.0%

210-110-0

.656

8

16

17-13

.567

3

18.8%

176-80-0

.688

9

22

18-21

.462

1

4.5%

233-119-0

.662

10

39

35-35

.500

4

10.3%

433-191-0

.694

11

29

20-27

.426

2

6.9%

317-146-1

.684

12

26

28-24

.538

2

7.7%

279-136-1

.672

13

10

4-10

.286

0

0.0%

108-51-1

.678

14

13

14-12

.538

1

7.7%

148-60-0

.712

15

33

42-28

.600

5

15.2%

400-128-0

.758

 

Teams with more momentum seem to do better. But again there may be a “quality bias” involved. So let’s run the data again, this time adjusting for the quality of the teams in each group of teams. This will be the equivalent of Table 3, but grouping the teams by Momentum Points instead of last-five-game records.


 

 

Table 5 - Playoff Performance: Actual vs. Expected (Momentum Points)

Momentum Points

Expected Playoff W/L %

Actual Playoff W/L %

Difference (W/L %)

Expected SB Wins

Actual SB Wins

Difference (SB Wins)

0

 

 

1

.273

.000

-.273

0.0

0

0.0

2

 

 

3

.435

.583

+.148

0.3

1

+0.7

4

.359

.333

-.026

0.1

0

-0.1

5

.419

.333

-.086

0.3

0

-0.3

6

.492

.471

-.021

0.6

0

-0.6

7

.429

.500

+.071

1.2

1

-0.2

8

.522

.567

+.045

1.5

3

+1.5

9

.477

.462

-.015

1.3

1

-0.3

10

.520

.500

-.020

3.9

4

+0.1

11

.494

.426

-.068

2.7

2

-0.7

12

.493

.538

+.045

1.5

2

+0.5

13

.499

.286

-.213

0.7

0

-0.7

14

.525

.538

+.013

1.5

1

-0.5

15

.573

.600

+.027

4.5

5

+0.5

 

Once again, if momentum counted for anything,  you would expect lots of high-negative numbers in the top of the Difference columns, and high-positive numbers at the end of these columns. If anyone can see a pattern here that benefits the high-momentum teams, you have better eyesight than I do! Really, there is nothing here that indicates that teams with lots of Momentum Points do better than expected. To illustrate that, here’s a graph of the “Difference (W/L %)” column this table (that is, the difference between the expected and actual Playoff W/L percentage) plotted against Momentum Points. If momentum mattered, you would expect the line to slope upwards from left to right. I certainly don’t see that.

 

 

 

 

Last Notes

 

One more data point. Since 1990 there have been 36 playoff games between teams with identical regular season records. In four of these games the teams had the same Momentum Points coming into the game. Of the remaining 32 games, the team with the better momentum went 14-18. So in the one case where momentum might make a difference, between two evenly-match teams, the team with the momentum won just 44% of the time!

 

Finally, what if five games isn’t the magic number?  What if four or six or three is the right number of games to define end-of-season momentum? Unlikely, but just to make sure I ran the same calculations against three, four, and six games, using both definitions of momentum: record-in-the-last-X-games and Momentum Points. The results were the same – there was never any evidence that momentum made a difference in post-season performance. I’ll spare everyone the tables and charts, but if someone’s really interested in them they can be generated.

 

Conclusions

 

Momentum going into the playoffs is a myth. It’s a useful myth, giving fans something to talk about, sportswriters something to write about, and TV and radio folks something to fill the airways with in the weeks before the playoffs start. If you want to believe the myth, that’s fine; sweat over your team’s late season swoon, or gloat over their winning streak. But when the playoffs start, don’t be surprised if the better team wins, regardless of how they were playing in the final weeks of the season.