Earlier this week, Nick Eatman suggested that this is the dreaded "trap game" for the Cowboys. I certainly understand his reasoning; the Cowboys are traveling to Kansas City to take on a team who, although talented, was just 2-14 last year. However, Dallas actually isn't the favorite in this game. The 'Boys are three-point underdogs, and while that might or might not seem fair, the truth is that the guys in Vegas who set those lines are pretty good at what they do.
So why in the world are the Cowboys not the favorite to beat a team coming off of a 2-14 season? Sure, Andy Reid is in town and Kansas City brought in a capable quarterback in Alex Smith, but it still seems odd. As I did last week, I'm going to run the numbers to project individual player performances, and perhaps that can shed some light on the situation. [embedded_ad]
Before getting started, let me again recap my methodology. When trying to project players, we typically look at how they did in the past against a certain team. When the Cowboys play the Giants later this year, Tony Romo's Week 1 stats should certainly be a consideration in projecting him in that game. But that's hardly a large enough sample size to make accurate predictions. We can't just predict that Romo will repeat those stats.
To solve the sample size issue, I look at how quarterbacks similar to Romo have performed against defenses comparable to his opponent. I use apps like this one from rotoViz that use an algorithm to find the matchups most similar to "Romo vs. Chiefs." That allows for all sorts of cool insights because 1) we have a large sample of games to study, and 2) we can analyze players in terms of probabilities. We intuitively know that Romo isn't a lock to throw for three touchdowns, for example, but by studying how similar quarterbacks have performed against defenses comparable to that of Kansas City, we can assign Romo a probability of tossing three scores.
That general methodology, finding historical player comps, has proven to be accurate in the past. It's the best way that I know to project players, and we can use the projections to predict a final score.
Tony Romo Comps
Earlier this week during the "On Air" podcast, I predicted a close win for the Cowboys. That was just my opinion before digging through the numbers (something I still haven't done as I write this sentence). So let's get into those numbers right now.
Looking at Romo's comps, it's clear why the Cowboys are the underdog in this game.
We can throw Rex Grossman's 2010 performance out of this sample because he got injured in the game. For the other 24 comps, here's the average stat line: 22-for-36 (61.1 percent) for 240 yards (6.67 YPA), 1.58 touchdowns, 0.83 interceptions.
That's not horrible, but it's certainly not Romo-esque. The most important number in there is the yards per attempt (YPA), which is really predictive of team success. At 6.67 YPA, Romo's comps (four of which are Romo himself) have averaged more than a full yard less than his career average (7.91 YPA).
Breaking it down further, we can look at the probability of Romo reaching each level of efficiency.
Based on his comps, Romo has a 62 percent chance to not even reach 7.0 YPA. For whatever reason, he and comparable quarterbacks have struggled against Chiefs-like defenses.
And how about the probability of Romo throwing X number of touchdowns?
So even though he's projected at 1.58 touchdown passes, Romo's most likely outcome is probably just one touchdown pass, followed by zero! There is a fairly large deviation here, though, as exactly one-quarter of Romo's comps have thrown for at least three scores and eight percent have tossed at least four touchdowns.
The Other Guys
Using the same methodology, here's how the comps for the other players turned out:
- DeMarco Murray:18 carries, 79 yards, 0.28 TDs, 2.8 receptions, 23 receiving yards, 0.04 receiving TDs* *
Nearly 80 yards is a nice number for Murray – that would extrapolate to 1,264 over a 16-game season – but I think there's a good chance that he crosses the 100-yard mark. The Chiefs figure to mimic the Giants' game plan to stop Dez Bryant, and that should open up things in the running game. No one would confuse me for a proponent of running the ball early, but I actually think the Cowboys could find initial success on the ground on Sunday.
- Dez Bryant:7.6 targets, 4.4 receptions, 63 yards, 0.68 TDs
Last week, I wrote "I think there's a high probability of Bryant either 1) posting abnormally low numbers because the Giants do everything they can to take him out of the game or 2) completely going off because the Giants don't double-team him."
So yeah, that happened. And it's probably true again this week, although Bryant's 0.68 TD projection is probably pretty accurate.
- Miles Austin:7.3 targets, 4.3 receptions, 63 yards, 0.40 TDs
Austin's production will once again be tied to how the defense defends Bryant.
- Jason Witten:7.5 targets, 4.9 receptions, 60 yards, 0.40 TDs
If Kansas City sits back in Cover 2 and Cover 2 Man-Under, you'll probably see another big game from Witten. He's unlikely to score two touchdowns in another game this year, though.
And here's how the Chiefs' skill players are projected to perform based on their historical comps:
- Alex Smith:19-for-26 for 223 yards, 1.25 TDs, 0.83 INTs
- Jamaal Charles:16 carries, 79 yards, 0.64 TDs, 1.9 receptions, 18 receiving yards, 0 receiving TDs
- Dwayne Bowe:6.8 targets, 3.9 receptions, 52 yards, 0.2 TDs
- Donnie Avery: 6.9 targets, 3.7 receptions, 54 yards, 0.32 TDs
- Anthony Fasano:3.9 targets, 2.4 receptions, 25 yards, 0.24 TDs
Projecting a Final Score
Now, we can take the final stat projections and use them to predict the final score. Romo and Murray are projected to combine for 1.96 touchdowns. That's down from last week. When you factor in the opportunity for non-offensive scores and touchdowns from other players, we're at right around 2.10 touchdowns.
Since kicker production is so fluky, we can just use Dan Bailey's average of 1.81 field goals from 2012 to predict his points. When you add it all up, that's 20.1 projected points for Dallas.
For Kansas City, Smith and Charles are projected to score 1.89 touchdowns. We'll call it 2.00 with the opportunity for other scores. Kicker Ryan Succop made 1.75 field goals last year. That gives Kansas City a final point projection of 19.3 points.
Now you can see why Vegas has the Cowboys as underdogs; the two teams are projected to score right around the same amount of points, but the Chiefs are at home. That's historically "worth" three points to a team, hence the three-point spread.
So while it seems that most fans think the Cowboys "should" win this game, it's very close to a coin flip. I'd be inclined to go with the home team in such a situation, but … I can't. Cowboys 21, Chiefs 20.