You are here
Running the Numbers: Heart, Determination & Predictions
You can’t measure heart, they say. You can’t quantity determination or leadership, either. Those intangibles are what renders all of analytics useless, some claim.
And while it’s difficult to isolate intangibles like “heart,” it’s not like the numbers are incapable of capturing them in any sense. Characteristics like heart, while extremely vague and in some ways meaningless, are measured insofar as they affect on-field play.
If Tony Romo has heart, we’d expect that to help his play, right? We’d expect it to aid his production. He’d presumably have superior efficiency, more touchdowns, or some other improvement in play, right? If a trait affects on-field performance, it can be measured.
But if heart, leadership and the like don’t affect a player’s production – if heart can’t help Travis Frederick block better or leadership can’t aid Sean Lee and his teammates in making more tackles – then why do we care about it?
For the record, I think certain traits that are challenging to quantify in isolation, such as a player’s motivation, can be really important in building the right sort of team. But the idea that such characteristics can’t at all be measured is silly. If they have an effect on the field, that will be reflected in the numbers. If they don’t, why bother?
A Look at Tony Romo
As usual, I’ll be using the rotoViz GLSP model to project Cowboys and Eagles players based on their comps – similar players facing comparable defenses. Here are Romo’s 25 closest comps versus the Eagles. Note that “heart” has already been factored in.
The app combs past data for quarterbacks with similar stats to Romo, so it’s no surprise to see a lot of the same names on the list: Matt Ryan (5), Peyton Manning (3) and, of course, Romo himself (4). The average stat line for these comps is 25-of-40 for 296 yards, 1.96 touchdowns, and 0.92 interceptions.
If we look at just how Romo has performed against defenses similar to Philly’s, the average line is 27-of-45 for 339 yards, 2.25 touchdowns, and 1.50 interceptions. That sort of game is probably about equal to the first line because, although Romo himself has more yards and touchdowns than all of his comps, he also has significantly more picks. That suggests we could be in for a bit of a rollercoaster ride on Sunday, which might or might not be a good thing.
Another piece of evidence that suggests a potentially volatile game from Romo is that many of his comps have performed close to the extremes – either a dominant performance or a really poor one. That’s reflected in Romo’s touchdown probability (based on the comps).
Of Romo’s 25 closest comps for this game, 40 percent have thrown either three or four touchdowns. That’s a high number. On the flip side, 44 percent have tossed either zero or one touchdown. While Romo’s average line is pretty standard or just slightly better than normal, the numbers suggest he’s set up for an outlying performance – either four-touchdown dominance or an ugly 250/1/3 sort of line.
Using the same methodology, here’s the average line for the other skill position players in this game. Read
RB Joseph Randle: 31 rush yards, 0.28 rush TD, 1.8 receptions for 10 yards, 0.12 receiving TD
WR Dez Bryant: 4.9 receptions for 84 yards, 0.48 TD
WR Miles Austin: 3.9 receptions for 61 yards, 0.20 TD
WR Terrance Williams: 3.6 receptions for 59 yards, 0.28 TD
TE Jason Witten: 4.6 receptions for 47 yards, 0.28 TD
QB Nick Foles: 22-for-34 for 278 yards, 1.92 TD, 1.36 INT
RB LeSean McCoy: 72 rush yards, 0.52 rush TD, 4.0 receptions for 31 yards, 0.12 receiving TD
WR DeSean Jackson: 4.2 receptions for 66 yards, 0.24 TD
WR Riley Cooper: 2.4 receptions for 28 yards, 0.16 TD
WR Jason Avant: 4.2 receptions for 58 yards, 0.28 TD
TE Brent Celek: 3.2 receptions for 40 yards, 0.24 TD
Predicting a Score
So as usual, we can take the individual projections and predict a final score. The Cowboys would have 1.96 Romo passing touchdowns, 0.28 Randle rushing touchdowns, and 0.35 “other” touchdowns (returns, for example). That’s 2.59 touchdowns, or 18.1 points. We can project kicker Dan Bailey at 1.69 field goals, his average over the past 16 games. That’s a total of 23.2 points for Dallas.
The Eagles check in at 1.92 Foles passing touchdowns, 0.52 McCoy rushing touchdowns, and the same 0.35 “other” touchdowns. When you add in kicker Alex Henery’s 16-game average of 1.81 field goals, you get a total score 25.0 points.
However, the model, which still uses player stats dating back to last year to find comps, doesn’t fully account for the tempo of the Eagles’ offense. They’ve been involved in an amazing 139 plays per game this season, 14 more than the average team. That’s basically a 10 percent increase in overall output, which would bump the raw projections to 25.5 points for Dallas and 27.5 points for Philly.
Considering the Eagles are at home (typically worth around three points), my final prediction is …
Eagles 28, Cowboys 24 Read