By Chris Sherman
I don’t really have anything to add as far as a preview for the Green Bay game that you haven’t already read this week. Except, maybe, that Freddie Mitchell’s AT ratio (Arrogance / Talent) this season is on pace to break his own record.
Here’s a little something to nibble on as we prepare for tomorrow’s game:
Thought process (skip down for actual analysis)
So I was thinking to myself a while back as to how we can truly evaluate the rankings of teams in comparison to one another.
You can’t really do it by simply their W-L record. With an unbalanced schedule (meaning that not everyone plays the same teams), Win-Loss record can be easily inflated or deflated based on the competition faced.
You can’t really put together a weighted average based on their opponent’s W-L record (i.e. counting a win against a 4-12 team as worth less than a win versus a 12-4 team) because you run into a circular reference problem (chicken-egg problem). For instance, is the NFC-South’s record inflated because they played a weak NFC-West? Or is it that the NFC-West’s record is deflated because they played a powerful NFC-South? The answer is probably somewhere in the middle but no one can prove with any confidence how much a win is worth based solely on their record.
I decided that the only way to determine how teams stack up against each other is by head-to-head matchups and by winning percentage amongst common opponents. After determining how each team compares to every other team on these grounds, I decided to combine the results of each matchup as a pseudo-W-L-T record. This should give a decently accurate representation of how the teams rank all together.
Now you can stop skipping:
-This compares W-L record of common opponents for each other team in the NFL
-Head to Head games are also included and count double
-I removed those week 17 games in which a team didn’t play their starters
The results weren’t as encouraging for the Eagles as I had hoped. We got a lot of wins, but, being the only team not to tie anyone, we picked up a fair number of losses too.
| Rank | Team | Conf | WinPerc | W | L | T |
| 1 | NE | Amer | 100.00% | 29 | 0 | 2 |
| 2 | PIT | Amer | 89.66% | 26 | 3 | 2 |
| 3 | ATL | Nat | 88.89% | 24 | 3 | 4 |
| 4 | BAL | Amer | 88.46% | 23 | 3 | 5 |
| 5 | NYJ | Amer | 79.31% | 23 | 6 | 2 |
| 6 | GB | Nat | 79.17% | 19 | 5 | 7 |
| 7 | NO | Nat | 75.86% | 22 | 7 | 2 |
| 8 | IND | Amer | 75.00% | 18 | 6 | 7 |
| 9 | PHI | Nat | 74.19% | 23 | 8 | 0 |
| 10 | CHI | Nat | 70.83% | 17 | 7 | 7 |
| 11 | NYG | Nat | 65.38% | 17 | 9 | 5 |
| 12 | TB | Nat | 64.00% | 16 | 9 | 6 |
| 13 | KC | Amer | 62.50% | 15 | 9 | 7 |
| 14 | MIA | Amer | 55.56% | 15 | 12 | 4 |
| 15 | MIN | Nat | 48.00% | 12 | 13 | 6 |
| 16 | SD | Amer | 47.83% | 11 | 12 | 8 |
| 17 | JAX | Amer | 41.67% | 10 | 14 | 7 |
| 18 | OAK | Amer | 40.74% | 11 | 16 | 4 |
| 19 | SEA | Nat | 37.50% | 9 | 15 | 7 |
| 20 | DET | Nat | 35.71% | 10 | 18 | 3 |
| 21 | CLE | Amer | 33.33% | 9 | 18 | 4 |
| 22 | STL | Nat | 29.17% | 7 | 17 | 7 |
| 23 | TEN | Amer | 27.27% | 6 | 16 | 9 |
| 24 | HOU | Amer | 26.09% | 6 | 17 | 8 |
| 25 | WAS | Nat | 25.93% | 7 | 20 | 4 |
| 25 | CIN | Amer | 25.93% | 7 | 20 | 4 |
| 27 | BUF | Amer | 22.22% | 6 | 21 | 4 |
| 28 | SF | Nat | 18.18% | 4 | 18 | 9 |
| 29 | ARZ | Nat | 17.86% | 5 | 23 | 3 |
| 30 | DAL | Nat | 14.81% | 4 | 23 | 4 |
| 31 | DEN | Amer | 11.11% | 3 | 24 | 4 |
| 32 | CAR | Nat | 7.69% | 2 | 24 | 5 |
The most obvious and important criticism of this analysis tool is that it doesn’t take into account sample size of games. While most teams share at least 5 games, some can go as low as 2 shared games.
So, I added a little weighting system to count results with more shared opponents count more. It simply multiplies the ‘win’, ‘loss’, or ‘tie’ by the sample size of games (counting H2H games twice (each)) divided by 16.
REMEMBER, THAT THIS WEIGHTING SYSTEM DOESN’T WEIGHT THE QUALITY OF A VICTORY, ONLY THE STATISTICAL CERTAINTY OF IT.
| Rank | Team | Conf | weight_WinPerc | weight_W | weight_L | weight_T | change |
| 1 | NE | Amer | 100.00% | 12.7 | 0.0 | 0.4 | 0 |
| 2 | ATL | Nat | 91.90% | 12.1 | 1.1 | 0.9 | 1 |
| 3 | BAL | Amer | 87.43% | 10.4 | 1.5 | 2.1 | 1 |
| 4 | PIT | Amer | 86.70% | 11.0 | 1.7 | 1.3 | -2 |
| 5 | IND | Amer | 82.89% | 9.7 | 2.0 | 2.3 | 3 |
| 6 | GB | Nat | 78.97% | 9.6 | 2.6 | 1.8 | 0 |
| 7 | CHI | Nat | 77.27% | 9.6 | 2.8 | 1.6 | 3 |
| 8 | PHI | Nat | 77.14% | 10.1 | 3.0 | 0.0 | 1 |
| 9 | NO | Nat | 72.91% | 9.3 | 3.4 | 1.3 | -2 |
| 10 | NYJ | Amer | 72.33% | 9.3 | 3.6 | 0.3 | -5 |
| 11 | NYG | Nat | 65.37% | 8.4 | 4.4 | 1.2 | 0 |
| 12 | TB | Nat | 64.55% | 7.6 | 4.2 | 2.2 | 0 |
| 13 | OAK | Amer | 61.54% | 7.5 | 4.7 | 1.8 | 5 |
| 14 | KC | Amer | 57.75% | 6.8 | 4.9 | 2.3 | -1 |
| 15 | SD | Amer | 56.42% | 6.3 | 4.9 | 2.8 | 1 |
| 16 | MIA | Amer | 45.60% | 5.5 | 6.6 | 1.1 | -2 |
| 17 | MIN | Nat | 42.00% | 5.3 | 7.3 | 1.5 | -2 |
| 18 | SEA | Nat | 40.91% | 4.5 | 6.5 | 3.0 | 1 |
| 19 | STL | Nat | 37.72% | 3.9 | 6.5 | 3.6 | 3 |
| 20 | JAX | Amer | 34.62% | 3.4 | 6.4 | 4.3 | -3 |
| 21 | CLE | Amer | 31.86% | 4.1 | 8.7 | 1.3 | 0 |
| 22 | HOU | Amer | 30.86% | 3.1 | 7.0 | 3.9 | 2 |
| 23 | SF | Nat | 27.39% | 2.7 | 7.1 | 4.2 | 5 |
| 24 | WAS | Nat | 24.38% | 3.1 | 9.5 | 1.4 | 1 |
| 25 | TEN | Amer | 23.84% | 2.3 | 7.2 | 4.6 | -2 |
| 26 | DET | Nat | 22.49% | 2.9 | 10.1 | 0.9 | -6 |
| 27 | BUF | Amer | 22.45% | 2.8 | 9.5 | 0.9 | 0 |
| 28 | CIN | Amer | 19.05% | 2.5 | 10.6 | 0.9 | -3 |
| 29 | DAL | Nat | 14.21% | 1.7 | 10.2 | 1.3 | 1 |
| 30 | ARZ | Nat | 14.15% | 1.9 | 11.4 | 0.8 | -1 |
| 31 | DEN | Amer | 10.14% | 1.3 | 11.6 | 1.1 | 0 |
| 32 | CAR | Nat | 7.92% | 1.0 | 11.6 | 1.4 | 0 |
The Eagles triumph over New Orleans in this scenario but still rank 8th. The Jets are the big losers here as they move down 5 spots.
A team’s actual ranking is probably somewhere in between their rankings on these two charts.
If you’re curious, the Eagles’ result vs. Green Bay was indeed a loss but only by the margin of the head-to-head game in week 1 before Antonio Dixon, Moise Fokou, Max Jean-Gilles, Owen Schmidt (Eldra Buckley played FB when Weaver went down), Jamar Chaney (Omar Gaither played when Bradley went down), Juqua Parker, and, oh yes, Michael Vick were starters. In games amongst common opponents with Green Bay, Philadelphia was 7-3 and Green Bay was 7-4 (our week 17 game against Dallas was discounted because our starters weren’t playing).
There are still drawbacks to this analysis tool, and before next season, I’ll be thinking of ways to improve it. If you have any suggestions to make it more accurate, please post them below.
Otherwise, if you’re curious about the individual breakdown of Philadelphia’s other matchups, here you go:
| Team | Opponent | GP | Wins | Losses | Ties | Margin of Victory |
| PHI | NYG | 15 | 1 | 0 | 0 | 16.67% |
| PHI | WAS | 15 | 1 | 0 | 0 | 16.25% |
| PHI | DAL | 14 | 1 | 0 | 0 | 28.57% |
| PHI | JAX | 10 | 1 | 0 | 0 | 43.64% |
| PHI | HOU | 10 | 1 | 0 | 0 | 43.64% |
| PHI | DET | 10 | 1 | 0 | 0 | 32.73% |
| PHI | IND | 10 | 1 | 0 | 0 | 25.45% |
| PHI | TEN | 10 | 1 | 0 | 0 | 15.45% |
| PHI | MIN | 10 | 1 | 0 | 0 | 4.55% |
| PHI | OAK | 5 | 1 | 0 | 0 | 80.00% |
| PHI | CAR | 5 | 1 | 0 | 0 | 60.00% |
| PHI | TB | 5 | 1 | 0 | 0 | 40.00% |
| PHI | STL | 5 | 1 | 0 | 0 | 40.00% |
| PHI | SEA | 5 | 1 | 0 | 0 | 40.00% |
| PHI | DEN | 5 | 1 | 0 | 0 | 40.00% |
| PHI | KC | 5 | 1 | 0 | 0 | 20.00% |
| PHI | BUF | 5 | 1 | 0 | 0 | 20.00% |
| PHI | SF | 4 | 1 | 0 | 0 | 75.00% |
| PHI | ARZ | 4 | 1 | 0 | 0 | 55.00% |
| PHI | ATL | 4 | 1 | 0 | 0 | 25.00% |
| PHI | CLE | 2 | 1 | 0 | 0 | 100.00% |
| PHI | CIN | 2 | 1 | 0 | 0 | 100.00% |
| PHI | BAL | 2 | 1 | 0 | 0 | 50.00% |
| PHI | GB | 12 | 0 | 1 | 0 | 10.90% |
| PHI | CHI | 10 | 0 | 1 | 0 | 22.73% |
| PHI | NE | 5 | 0 | 1 | 0 | 60.00% |
| PHI | MIA | 5 | 0 | 1 | 0 | 40.00% |
| PHI | NYJ | 5 | 0 | 1 | 0 | 20.00% |
| PHI | SD | 5 | 0 | 1 | 0 | 20.00% |
| PHI | NO | 4 | 0 | 1 | 0 | 5.00% |
| PHI | PIT | 2 | 0 | 1 | 0 | 50.00% |
No comments:
Post a Comment