In Week 1, I used individual stat projections to argue that the Dallas offense would score 23 points* *against the Giants. They did, adding two defensive touchdowns, which are notoriously hard to predict.
In Week 2, I used the same model – this one based on similarity projections – to predict both teams' final scores. The result was a 20-19 game in favor of Dallas (a score which normally would have been switched since Kansas City was at home, something I didn't do because, hey, I'm a fan just like you). The Chiefs indeed won by one point, 17-16.
The model also predicted that Tony Romo's average stat line against Kansas City would be 22-for-36 (61.1 percent) for 240 yards (6.67 YPA), 1.58 touchdowns, 0.83 interceptions, with the most probable outcome actually being just a single touchdown pass, which did indeed happen.
So what does all that say about my ability to predict games and player stats? Absolutely nothing. The Cowboys have played only two games, so I could just be lucky at this point. We don't know right now. And the model has been off on some things as well. It's going to be really difficult to consistently predict individual player's stats, in particular, because they're so volatile from week to week.
But I've been pretty accurate on the final score projections thus far, and I'm confident that trend will continue. That's not to say the model will never be off; it will be. But hopefully I'll be able to perform better than someone using traditional stats or, even worse, just guessing.
Tony Romo's Comps vs. St. Louis
As I did last week, I'll examine player comps – similar players versus comparable defenses – to project the most important guys. And when we look for players with similar recent stats to Romo playing against defenses comparable to that of the Rams, this is what we get:
The average line for those guys is 24-for-37 (64.5 percent) for 282 yards (7.62 YPA), 1.92 touchdowns, and 0.64 interceptions, significantly better than last week.
We can break down the comps further to estimate Romo's probability of achieving certain levels of success. Here are the touchdowns.
Unlike last week, Romo's most likely outcome is two touchdown passes. There's probably around a three-in-five chance that he tosses either one or two. He's also got nearly a one-in-four chance to throw either none or at least four.
The Other Guys
Using the same methodology, let's take a look at the average line for the Cowboys' other skill players:
- RB DeMarco Murray:65 rushing yards, 0.48 rushing touchdowns, 3.4 receptions for 29 yards, 0.12 receiving touchdowns
I've been high on Murray all year, so I think there's a really good chance that he turns things around. He's contributing quite a bit as a receiver, and I think he'll get going on the ground this week against the Rams.
As a side note, I've heard some talk about Murray underperforming because he's a "straight-line runner." I think we all already knew that, right? It doesn't take a scout to see that Murray doesn't juke many defenders. But you know who else is a straight-line runner? Jamaal Charles. And Chris Johnson. And even Adrian Peterson, to a degree.
Not every back is LeSean McCoy. Size and speed matter most for backs, and Murray has that. He'll be fine.
- WR Dez Bryant:6.4 receptions for 87 yards, 0.68 touchdowns
This is actually a quality line for Bryant. If you extrapolate it over 16 games, it comes out to 102 receptions for 1,392 yards and 11 touchdowns. Given the short stature of the Rams' cornerbacks, which isn't a factor in this projection, I think Bryant's probability of scoring a touchdown is probably closer to 75 percent.
- WR Miles Austin:4.6 receptions for 62 yards, 0.32 touchdowns
Like every week, Austin's production is strongly tied to how the defense defends Bryant. If they double Bryant on most plays, as they should, Austin needs to step up.
- TE Jason Witten:5.7 receptions for 59 yards, 0.24 touchdowns
It's worth noting that a large portion of Witten's comps in this game are Witten himself. His recent matchups with the Saints, Bucs and Redskins in 2012 are all listed due to the similarity of the defenses. [embedded_ad]
And the Rams' player projections:
QB Sam Bradford:24-for-36 (66.7 percent) for 279 yards (7.75 YPA), 1.88 touchdowns, 0.88 interceptions
RB Daryl Richardson:40 rushing yards, 0.28 rushing touchdowns, 3.9 receptions for 29 receiving yards, 0.16 receiving touchdowns
WR Tavon Austin:5.9 receptions for 68 yards, 0.36 touchdowns
WR Chris Givens:3.3 receptions for 47 yards, 0.36 touchdowns
TE Jared Cook:5.8 receptions for 68 yards, 0.68 touchdowns
*Note that I altered the possible comps for Cook by including only his two games in St. Louis in the sample. He's a talented player in a dramatically different situation, so it doesn't make sense to find players who are comparable to the Tennessee version of Cook.
Projecting a Final Score
So now we've got most of the info we need to project a final score. With Romo projected at 1.92 touchdowns and Murray at 0.48 rushing touchdowns, we have a bulk projection of 2.40 scores. We can refine that to 2.60 touchdowns to account for those from other players, either offensively, defensively, or on special teams. That's 18.2 points.
Since kicker production is a crapshoot, we can use Dan Bailey's average of 1.81 field goals in 2012 to project him at 5.4 points. When we add that to the touchdowns, we get a final projection of 23.6 points.
For the Rams, Bradford and Richardson's scores add up to 2.16 touchdowns. The Rams are probably slightly more likely than Dallas to have a non-starter rush for a touchdown, however, just because Richardson isn't a workhorse back like Murray. So we can project the Rams at right around 2.46 touchdowns (17.2 points).
Rams kicker Greg Zuerlein averaged 4.3 points on field goals in 2012. Adding that to their touchdown projection, we get a final point projection of 21.5 points.
That's a raw score prediction of 23.6 to 21.5 in favor of Dallas. As I mentioned last week, being the home team is typically "worth" right around three points (so you can see why the Cowboys are four-point favorites in this game). Knowing that, here's my final Week 3 prediction:
Cowboys 24, Rams 20