Advanced NFL Power Rankings: 49ers Are Consensus No. 1

Greetings, football fans! I’d like to introduce something to you that I’ve been doing over at Bolts from the Blue for the last three years or so. I call it the NFL Power Rankings Consensus, and well, the name kind of speaks for itself.

Simply put: I troll the web for various NFL power rankings, load them up into my computer and crunch the numbers. I then make observations based on what the numbers tell me. Don’t worry, it’s nothing too complicated, and I’ll walk you through the various displays, explaining as we go along.

This week there are 18 different sources for the power rankings. But, not all rankings are created equally: some are based on running statistics through computers to generate rankings, and others are based on gut feelings on how well teams are playing, like the one from SB Nation.The latter is probably what you’re used to when you think of “Power Rankings.” I’ve broken down the stats-based rankings and the perception-based rankings into two categories -- “Objective” and “Subjective” rankings, respectively. This will become more obvious as we move along.

Before we get to looking at where all the teams rank, let’s take look at the bigger picture in the NFL. This table takes the average rankings for each team, then averages the rankings of each team in the division. It’s used as a way of showing where the tougher and weaker divisions lie each week.

As you can see, the NFC East is the dominant division in the NFL, while the NFC West – the former laughingstock of the NFL — is respectable, and it’s not all because of the 49ers. Two weeks into the season, the NFC is becoming the powerhouse conference in the NFL now as the top three divisions are NFC divisions, and the AFC occupies the bottom 3 positions.

This table is kind of self-explanatory: it shows which teams have moved up the most this week compared to last week.

Like the “Stock Rising” chart above, this one shows which teams are slipping.

We’ll start off with the Subjective Average. These rankings are the ones that has actual people ranking the teams based on their gut feelings. You’ll see in a moment that some of these sites may even just throw darts at a board, or have a chicken pecking at numbers. I then take all those rankings and average them together, then sort the table based on the average ranking. This method evens out the peaks and valleys and allows us to get a more reliable “big picture” view of the perception of where teams rank in the league.

Put simply, Standard Deviation is a tool in statistics to measure how much variance there is in a set of numbers. To put more simply: I use it to look at who’s ranked the most- and least-consistently. In Standard Deviation, the lower the number, the more consistent the values in the set. For our purposes, this means the lower the Standard Deviation, the more consistent the rankings. The opposite is true too: the higher the number, the less consistent the rankings.

This table is sorted by most-consistently ranked to least-consistently ranked.

One thing I’d like to point out about the Standard Deviation method here: it’s not perfect. Because teams are bound on each end of the spectrum (they can’t be any less than 1 or more than 32), we tend to see more consistent numbers as you get to each end of the spectrum. It’s not a steadfast rule as there are plenty of times there’s no consensus on the best or worst teams in the NFL, it is kind of the rule of thumb. Just something that’s the nature of the numbers.

I call these rankings the “Objective” rankings because unlike the rankings above, these are rankings based on statistical models of some sort. I use the term “Objective” loosely here because there is still some inherent bias in how the various models are constructed. Each ranking site has their own criteria in what they feel is success in the NFL.

All that said, these rankings are more objective as they remove perception and bias and produce rankings based on how well teams actually play, not how they’re perceived to play.

Now let’s compare the two sets of data. We’ll start with the “Overall Average” chart. This takes the Subjective Average and the Objective Average, and then averages those two numbers. This way it gives equal weight to both averages so that even though Subjective Average has a larger sample size, it doesn’t drown out the Objective Average.

The Overall Standard Deviation does not average anything out, it just takes the Standard Deviation across all the rankings. Yes this will even things out slightly, but it goes to show which teams have consensus across the board.

This compares the Subjective average ranking with the Objective average ranking. This tells us just how far off people are from the formulas. A positive number means people think they will be or are better than the stats indicate, a negative number means the math thinks they’re better than what people think. The closer to 0 this number is, the more in agreement the Subjective and Objective rankings are.

Generally speaking, the Subjective Rankings tend to react to a team’s performance quicker than the mathematical models can. So that’s why in the charts it’s generally considered a good thing if the Subjective Rankings are higher than the Objective Rankings. The opposite can most certainly hold true in that some teams are actually better or worse than they’re perceived. But it’s difficult to tell which teams fall into each category.

This is something a little new I’m trying out. I take a look at what should be the best match-ups based on the rankings, and also what should be the biggest blow-outs.