"Moneyball" is based on a book by the same name by godfather of advanced baseball stats., which describes what happened when an actual team decided to base its decisions about ball players on new ways to think about the game, ideas developed by Bill James, the
One of the many, many things James used to do in his Baseball Abstracts is look at six indicators to help predict whether a team would improve or decline in the next season. Today we'll use those indicators to see what they can tell us about the NFC East heading into the 2017 season.
Just a small word of caution up front. James developed the indicators for baseball, not for football. I can't vouch for their validity in baseball, much less in the NFL, but I thought they'd provide an interesting starting point for discussion, as they provide at least some quantitative basis for preseason predictions. Also, because the indicators look at stats from a macro perspective, they can help highlight trends that we might miss as we get lost in the minutiae of roster battles, position rankings and draft pick evaluations.
So without further ado, here are the indicators, and how each NFC East team stacks up:
1. Pythagorean Wins
The Pythagorean formula predicts a team's W/L record based on points scored and points allowed. When a team wins more games than it is projected to win, that team will tend to decline in the following season. Conversely, when a team wins significantly fewer games than projected by the formula, that team will tend to improve in the following season. The table below summarizes the win projections and variances for the NFC East teams.
|Variance||Proj. Wins for 2017|
Research by Bill Barnwell shows that over a period from 1983-2010, teams that overperformed their Pythagorean projection by more than two wins declined by 2.5 wins the next season. For the Cowboys and Giants, this would mean a drop to 10.5 and 8.5 wins respectively. Barnwell's data doesn't expect any changes for the Redskins, but bumps up the Eagles to 9.5 wins and second place in the division.
Many fans expect that in 2017, the Cowboys will simply pick up where they left off in 2016. And that's okay. We're fans. It's our job to be optimistic.
But repeating a 13-3 season is much harder than it sounds. Since 1990, when the league moved to a 12-team playoff format, 38 teams finished the season with a 13-3 record, and only four of those teams were able to repeat that 13-3 record the following year. Similarly, the Giants will find that repeating an 11-5 season is a lot easier said than done.
Overall, this indicator suggests some regression for last year's top teams, but also likes the Eagles to improve significantly this season, probably even leapfrogging the Giants in the standings.
2. "It’s not how you start, it’s how you finish"
Teams who play better in the second half of the season tend to improve the following season. It follows that teams who play worse in the second half tend to decline the following season. Because the NFL season is only 16 games versus the 162 in a baseball season, this indicator is closely related to the first indicator.
So we'll use the Pythagorean Formula once more to calculate the projected wins for the NFC East teams based on the last eight regular season games from 2016 of each team.
|Variance*||Trend for 2017|
|* Last eight regular season games|
For the Cowboys, Giants, and Redskins, this metric simply confirms what we saw in the first indicator, so we shouldn't expect this metric to impact our win projection. Not so for the Eagles. Where the 16-game Pythagorean suggested a 2.5-win improvement in 2017, that number was based almost entirely off their play in the first half of the season. The Eagles' second half of the season suggests they may have a hard time in 2017 if they continue playing like they did in the second half of the 2016 season.
Of course, this is where we run into the plague of NFL stats: small sample size. Applying the Pythagorean Formula to a 16-game schedule is already dubious at best, and it doesn't get any better when using it against an eight-game schedule, but there's nothing we can change about it except to exercise some caution in not overinterpreting these results, which is why I didn't add any win projections to this table.
3. "Plexiglass Principle"
In simple terms, teams that show a dramatic improvement (or decline) from one season to the next have a tendency to relapse (or bounce back) in the following season. What sounds simple enough can actually be quite tricky when a team has so little continuity from one season to the next that any comparison between the two seasons becomes a fool's errand. Take the Eagles, who imploded in a Chip Kelly fireball in 2015. Or the Cowboys, who trotted out a succession of QBs with two left hands in 2015. For both teams, a better comparison for continuity purposes is 2014, when the Eagles still had a coach and a QB they believed in, and the Cowboys didn't have a dumpster fire burning in the QB room.
Which is why I'll use 2014 as the base season for the Cowboys and Eagles, but stick with 2015 as the base for the Giants and Redskins. And I'll be using Football Outsiders' DVOA metric to evaluate the year-on-year improvement.
|Team||2014/5 DVOA||2016 DVOA||Change||Incremental Wins
Back in 2012, Danny Tucillo of FO looked at Year 2 DVOA changes for all teams since 1987 to arrive at the incremental win projections above.
All four NFC East improved their DVOA in this analysis, though the Giants are the only team that improved it so much that a bounce-back effect can be expected in 2017.
4. Regression to the mean
This statistical phenomenon describes the fact that if a variable is extreme on its first measurement, it will tend to be closer to the average on a second measurement. In sports, this means that teams with a winning record tend to decline; teams with a losing record tend to improve.
The formula we'll use to calculate this is straightforward: Starting with 8 wins, add 0.25 wins for every win above 8 from the previous year, or subtract 0.25 wins for every win below 8 from the previous year. For example, a 13-win team from 2016 would be predicted to have 8 + (13-8)/4 = 9 wins in 2017.
|Team||2016 W/L||Base wins||0.25 per game
above or below .500
|Trend for 2017
The Cowboys and Giants, because they farther away from the statistical average than the other two teams, are hit harder by the Regression To The Mean indicator.
5. Team Age
Young teams improve; old teams don't. This early in the season, we have only the vaguest idea of what the rosters for the NFL teams will look like, and looking at team age doesn't make a lot of sense before the final rosters are set. But here we are in June and want to make a prediction, so I went and looked at the current age of the 22 starters for each team as per the latest ourlads.com depth charts.
|Team||Combined Starter Age
||Avg. Starter Age||Trend for 2017
By this method, the Cowboys, Giants, and Redskins all have a combined age of 26.0 for their 22 projected starters, while the Eagles easily have the oldest team at 27.5 average years of age. This may not initially feel like a lot, but in a league that is obsessed wit youth, it is. Even worse for the Eagles, they've accumulated much of that age on the O-line.
The combined O-line age for the Cowboys and Giants is 126 years (25.2 average age), the Redskins are a little older at 134 (26.8), while the Philly Geriatrics Club is at a combined 150, which gives them an average age of 30. By NFL standards, that's ancient. As players become older, the risk of injury increases exponentially, and the Eagles might find that out the hard way this year.
6. Record in close games
Bill James' sixth indicator is Triple A performance, but since developmental leagues don't exist for the NFL, I'm picking 'record in close games' as the sixth indicator. It's awfully hard to consistently win a high percentage of games decided by seven points or less, and the vast majority of teams will see their performances in these types of games regress to the mean from year to year. Here's how the East fared in games decided by seven points or less last year:
|Team||2016 Close game record
||Trend for 2017
Whether the Eagles badly sucked or were just unlucky, and whether the Cowboys and Giants were really good or simply lucky is for us to think about and occasionally hyperventilate about. At the end of the day, it's unlikely that extreme outliers in performance can be maintained year after year.
Summing up the Indicators
to the mean
|Team Age||Close Games|
What this exercise shows is that there are some larger-scale statistical trends that can have an influence on how a season progresses. Unfortunately, we can't simply sum up the different metrics to come to a win projection, as some of these metrics overlap, and they certainly don't all have the same weight.
But what the larger-scale trends suggest is that the Cowboys and Giants are likely to regress, even if it's not clear by how much. Similarly, the indicators suggest the Eagles will improve, but by how much? You can draw your own conclusions from the data presented, my read on this is that the Cowboys will likely take the division with around 11 wins, the Giants could struggle to get nine wins and might and end up behind the Redskins. The Eagles look like the surprise team in the East, and I wouldn't be surprised to see them come in second with around 10 wins - if their O-line holds up.
Having said all that, the NFL is not baseball. And picking six largely random stats to make a projection is very arbitrary. There are so many different factors determining who wins and who loses each Sunday that an accurate projection is nearly impossible, and that's without even factoring in the biggest driver of NFL game results: randomness.
How do you see the NFC East playing out this year?