Week 15 True Power Rankings: Chiefs return to second

As we showed you on Tuesday, the Kansas City Chiefs jumped from fourth to second place in our aggregated pundit power rankings with their 27-24 overtime win against the Baltimore Ravens on Sunday — and after folding data from the number-crunchers, the Chiefs move from third to second in the Week 15 True Power Rankings.

Here’s how the crunchers see the NFL — now that Week 14’s games are in the books.

The Chiefs held steady — or improved — in all of the data-based models. But the Los Angeles Rams’ 15-6 loss to the Chicago Bears dropped them out of second place with the crunchers, and the Chiefs took their place.

Once again this week, the top teams remain the same. The Bears’ win served them well with all the models except the unweighted PE model — where they remained in second — and the Pittsburgh Steelers’ loss to the Oakland Raiders came within a whisker of dropping them out of the top tier.

We do see one odd thing here. The New England Patriots actually rose one spot with the crunchers after losing to the Miami Dolphins on that crazy last-second play, because the Patriots remain strong in Elo. The Patriot’s Elo score did, in fact, drop after the loss, but their one-point loss to the Dolphins cost them less than the Los Angeles Rams’ nine-point loss to the Chicago Bears.

This is another example of why rankings of statistical data can sometimes be misleading. We’ll come back to this when we get to this week’s cruncher grades.

Now let’s combine the cruncher rankings with Tuesday’s pundit rankings to create the Week 15 True Power Rankings.

With the Chiefs in second place with both the pundits and the crunchers, it shouldn’t be a surprise that the Chiefs would rise to second place in the combined rankings.

The Houston Texans drop out of the top tier after losing to the Indianapolis Colts, and the Ravens move in with the top teams.

Wait... what?

Yes, the Ravens move up two spots after losing to the Chiefs. We can see part of why this happened by looking at the inverse version of the Ravens chart we saw in Tuesday’s pundit power rankings.

This version of the chart compares the individual cruncher rankings against the aggregated pundit rankings — which is the black line.

The pundits haven’t been all that crazy about the Ravens all season — and started bailing on them during the three losses that led up to Joe Flacco’s injury. But the crunchers have been fairly high on them all along — as we see here, more so in some models than in others — and eventually, that resulted in the pundits undervaluing the Ravens quite a bit in comparison.

But the pundits have been getting back on board since the Ravens starting winning again. Then they decided the Ravens showed them something by nearly knocking off the Chiefs on the road — so they gave them a little boost. The crunchers, on the other hand, thought they should come down a little — but not a lot — because they lost narrowly to a better team. Also note that the Ravens’ DVOA ranking rose a bit after the loss, too. This is because DVOA is less concerned about wins and losses than some of the other models.

But again... rankings can be deceptive. The Ravens, Steelers and Texans are in a really tight group; the difference between the three is just 0.4 ranking points. If the pundits hadn’t decided to be generous, the Ravens probably would have stayed out of the top tier, and the Texans would have remained.

And here’s your weekly reality-check: the Week 15 cruncher grades.

The Chiefs drop measurably — but not significantly — in this week’s grades.

Again, we see a group of teams at the top of the league, a group at the bottom and a whole lot of average in between. The Bears and the Ravens move in with the top teams, and the Patriots fall out.

But why would the Patriots go down in the cruncher grades, while going up in the cruncher rankings we saw earlier? Or for that matter, why did the Ravens rise in the cruncher grades, while dropping in the cruncher rankings?

It’s all about the math. Rankings have a basic problem: they don’t tell us anything about the difference between two values — only that one is greater (or less) than another. So when you average rankings together, some error can start creeping in. But the cruncher grades are calculated using the actual values behind the rankings, rather than the rankings themselves. Further, since these data models use different scales, we must convert the values to standard deviations from average, so we can compare (and average) them together.

This helps eliminate some of the problems you run across with rankings and can give a more truthful picture of what’s going on. That’s why we always refer to the cruncher grades as a reality check.

That’s it for now. Big game Thursday!

Join us again on Tuesday. We’ll fire up the spreadsheet again and see how the NFL shakes out after Week 15.