Outcome bias is a powerful habit in forming mainstream opinion, and one of the most common NFL-related ones entails "momentum." It's easy to look at a team that wins the Super Bowl and point to a late-season three-game winning streak as a sign that the future champs "figured things out" or "got the ball rolling." Fill in your favorite cliche and voila, instant storyline.
But how often does so-called momentum play a role in playoff success? This isn't a new question, and many analysts and fans have realized that success at the end of the regular season is not a prerequisite for a deep postseason run. Last season's Baltimore Ravens won it all despite losing four of their final five regular season games.
More advanced metrics than simple wins and losses seem to also have confirmed this hypothesis. Last year, Football Outsiders used their weighted DVOA metric to dispel the notion that December performance has any bearing on playoff results. The teams with a higher weighted DVOA won games at about the same rate as teams with a higher DVOA, which measures the full season.
But then again, shouldn't we expect that to be the case? Better teams usually win, and better teams should have a higher DVOA and weighted DVOA. After all, the best teams usually play better for longer stretches of time (#analysis). If the Denver Broncos or Seattle Seahawks win the Super Bowl this season, no one will point to late-season momentum as the catalyst, even though they are the top two teams in weighted DVOA this year.
Instead, it might be more informative to examine how much of a role late-season performance played in underdogs who pulled out playoff victories. Teams are underdogs usually because they have proven themselves worse over the full course of a season. But because of injuries and internal development, the December versions of teams should theoretically provide a more accurate representation of a team's capabilities.
We can use EPA per game from the final month of the season (Weeks 14-17) to see how underdogs compare to their vanquished opponents. The idea is to look for patterns in the offensive or defensive EPAs of playoff underdogs who came out victorious from 2002-2012. 2002 is the cut-off date because that's when the NFL had an even number of teams, so there were no late-season byes to throw things off. The final-month EPA/G of the underdogs can also be compared to the EPA/G of the favored teams they defeated, to see which units were playing better headed into the final months.
Per Pro-Football-Reference, there were 44 postseason upsets between 2002-2012. Even if a team pulled off multiple upsets (ex: 2012 Ravens), they were considered a separate "team" in each upset in order to compare them against the different favorites they defeated. Graphing the offensive and defensive EPA/Gs of the two teams in each matchup against each other, we see a fairly clear trend emerge:
In upsets, the favored teams are generally better offensively, while the underdogs' recent performances are superior defensively. Moreover, as the first two graphs indicate, most of the EPA/G marks are reasonably close to what we would see in a full season, so it's not as if there were a bunch of outliers skewing the data. Only three units—the '08 Eagles defense, '09 Jets defense and '11 Saints offense—had per game EPAs that were unsustainable anomalies, and they're the ones that don't fit on the chart.
This dovetails with the common narrative that defense wins out in the postseason. It's not as if terrible defenses suddenly turned into the 2000 Ravens during the postseason, but some of the more famous teams on the list saw good defenses elevate their games to league-leading levels. The '05 Steelers, the first sixth seed to win the Super Bowl, went from -2.83 defensive EPA/G during the first 12 games to -5.66 per game in the last four. Ditto for the runner-up '03 Panthers, who went from -0.78 to -9.35, and the champion '10 Packers, improving from -3.61 to -5.65.
The prototype isn't perfect; in fact, the '07 Giants and '12 Ravens, two of the biggest surprise champions in recent seasons, weren't particularly great offensively or defensively in the final quarter of the regular season. Nevertheless, if we're fishing for upsets or darkhorses, it appears the best bet is to look for good defenses that get better as the season progresses. So what does that mean for this year's playoff field?
The Seahawks made the largest jump, which seems to re-affirm their status as NFC favorites. The Bengals are also intriguing on this graph—their defense got slightly better from the first 12 games, the result of the front seven solidifying itself after an initial rough patch following Geno Atkins' season-ending injury. The Colts defense also finished the season strong, allowing just 20 points in its final three games after a rough midseason stretch.
On the opposite end of the spectrum, the Patriots, Packers and Panthers regressed into net-minus defenses in the final quarter. New England and Green Bay suffered some key injuries, but Carolina's slump is a bit more worrisome. Granted, the Panthers had to play the Saints twice, but the unit was totally healthy for the final month. They also got games against the Jets and Falcons to even things out. Even with a Divisional round home game, Carolina stands out as vulnerable amidst a loaded NFC field.
This methodology isn't foolproof, and the one-and-done nature of the NFL postseason means that these late-season trends might not manifest themselves in a 60-minute sample size. Momentum might not be the right word to explain these upsets, as its debatable whether momentum even exists over long stretches of time. But while wins and losses are too black-and-white to be an indicator of possible postseason success, a team's late-season profile should typically provide a more accurate representation.
- Home
- team analysis
- Can Late-Season Momentum Explain Playoff Upsets?
Can Late-Season Momentum Explain Playoff Upsets?
By
Unknown
published on 1/03/2014
in
playoffs,
team analysis
Subscribe to:
Post Comments (Atom)
Looks like San Diego could fit this prototype also.
Instead of momentum the question should be is the team in question healthier going into their first playoff game than they were going into week 13. Most teams will lose an average of 3 players to the IR during the final 4 weeks. While on the other side of the ledger many teams will rest key players during the final week in an attempt to get/stay healthy for the playoffs. I suspect the relative performance of a team in the regular season versus the playoffs is mainly determined by the relative contributions of the players they get back from injury versus those that they lose to injury going into the playoffs.