Filed under:

# Eagles vs. Cowboys 2015: Let's Get Simulated

They told me there would be no math.

The Dallas Cowboys come to Philadelphia with as many question marks from their week one performance as the Philadelphia Eagles.  How will the offense respond without Dez Bryant?  Can the offensive line really generate a running attack? Can the defense win games?  Likewise, the Eagles need to demonstrate that they can compete at the start of the game, run the ball, and defend the pass.

How will this translate on the field Week 2?  Here is this week’s simulated forecast.

Last Week Recap

First, let’s revisit last week’s simulation.  Going into Atlanta, the Eagles were three point favorites, with Vegas succumbing a bit to the preseason hype surrounding Sam Bradford’s better-than-expected performance.  Based on this and last year’s stats, the simulation also favored the Eagles, forecasting a 67% chance of victory.  Alas, the Falcons had better ideas.

I re-ran the simulation using the game’s actual stats and the results are a bit different: Philadelphia won just 51.5% of the time (compared to 58% last week) and their Bayes odds dropped to 60% (again, inflated because they were three point favorites.  The underdog Falcons won the game with offensive and defensive efficiencies that forecasted a 40% chance of victory.  SMH kickers (and slow starts and Jordan Matthews’ hands and…).  Here are the simulation results using the game's actual numbers:

This Week’s  Simulation

Unlike last week, this week’s simulation is based on last week’s stats, however fair or unfair that may be.  As we move from week to week, we will use the season’s cumulative stats, so the simulation results will be more meaningful as the season progresses.  Keep that in mind when reading below.  Again, any statistical "advantage" is based solely on one week’s worth of stats.  Interact with the viz to see how simulation runs, turnovers, and changes to the point spread impact results:

Philadelphia Offense versus Dallas Defense

For the readers who asked last week, here are my efficiency formulas, which are based on Chip Kelly’s offensive success formula:

Offensive Efficiency = (Rushes + Completions – Turnovers) / (Total Plays + Accepted Penalties)

Defensive Efficiency = (Rushes + Incompletions) / ((Total Plays – Turnovers) + Accepted Penalties)

The Eagles’ first half against the Falcons was abysmal, but the second half almost made up for it.  The Eagles offense ran 68 total plays, rushing just sixteen times while also committing two turnovers and seven (accepted, enforced) penalties, a 66.7% rate of efficiency.

The Dallas Cowboys defense was about as efficient against the New York Giants (65%).  No clear statistical advantage here.

Dallas Offense versus Philadelphia Defense

This is where the statistical advantage presents itself.  The Cowboy’s offense was 81.2% efficient last week.  Despite turning the ball over three times, Tony Romo threw 36 completions and only nine incompletions, while the offense as a whole committed just one (accepted) penalty.  In contrast, the Eagles defense was 67% efficient last week, forcing two turnovers and 12 incompletions in 70 plays.  Advantage: Cowboys.

Point Spreads and Bayes Odds

Vegas knows.  The Eagles enter the game against Dallas as five point favorites.  Since 1994, five point favorites have won 66% of the time.  With the Eagles returning home after a close loss in Atlanta and the Cowboys coming to Philadelphia after a one point win over a divisional opponent, Vegas seems a bit bullish on the Eagles.  We can use this with the simulation results to update each team’s win probability.

What It All Means

Tony Romo’s command of the Dallas offense last week proves too much for the Eagles to overcome in this week’s simulation.  When the turnover margin is zero, the Cowboys win 77.7% of the time.  When taking into account the point spread, this means the Eagles have a 34.8% chance of winning the game.

But fortune quickly turns to favor the Eagles if they force one more turnover then they commit.  With a +1 turnover margin, their Bayes Odds improves to 54.3%.  With Dallas turning over the ball three times last week, this is a real possibility.

So, ultimately, this simulation is not too kind to the Eagles.  They will need to prove that last week’s performance against Atlanta was an aberration.