What's In A Schedule? Taking a Look at the Vikings' "True" Strength of Schedule

The Vikings are 5-2! That's pretty good, I think.

To that end, there's not much controversy over the fact that the Vikings—for the most part—have played well so far. We see this discussion every year for different teams; whether or not their record is a "true" representation of how good they are.

Typically, that discussion revolves around undefeated teams, like the 9-0 Kansas City Chiefs in 2013 or the current Carolina Panthers before (and honestly, after) they took it to Seattle.

Many times, that discussion pans out like the skeptics expect: the overhyped team on a weak schedule falls flat and limps into (or misses) the playoffs. More often than not, the hype about hype is overblown. That can be for several reasons, but it can all circle around one: it's very difficult to go undefeated and not be good.

An average team typically has a 65-70% chance of beating an awful team in a neutral stadium (this week, for example, the Giants are favored at -135 against the Bucs at Tampa Bay, which is roughly a 57% win rate on the road and would be about 80% at home).

In a seven-game schedule against the worst teams, the average team has an 8.2 percent chance of emerging undefeated, a 24.7 percent chance of earning six wins and a 31.7 percent chance of coming out with five wins.

A good team (like the Patriots) can be expect to beat a bad team a little more than 9 out of 10 times. In the same schedule, they would end up undefeated about half the time, have six wins 37 percent of the time and underperform to get five wins 12 percent of the time.

So, if the only information we had about a team were their wins and losses, as well as their "strength of schedule," we could get a reasonable estimate on how good they were. You can do that in a lot of ways that are more sophisticated than I am capable of doing, but for the most part if you have a chart of what the likelihood of a certain class of team having a certain record is, you can really crudely do it.

Speaking of crude, here is that chart!

Awesome. So anyway, a five-win team against a slate of teams that is all bad is most likely a good-ish team! If we wanted to separate the tiers from 0.5 (bad team) to 0.9 (very good team) into five sets (below average, average, above average), there's a 21 percent chance that a five-win team is a little below average, 25 percent chance they are average, 22 percent chance they are above average and 10 percent chance they are very good.

Remember, if you're expected to be .500 against bad teams, you yourself are a bad team—which is why I didn't get into the tiers below the .500 rate.

This may be why NFL.com said the Vikings were set to tumble and why Ben Goessling said the Vikings may not be "real."

That all seems to be pretty simplistic though, yea? What about how the Vikings played? What about the eye test? What about the fact that most measures of the Vikings strength of schedule (opponent wins) don't take into account the fact that many of the Vikings' opponents themselves had tough strengths of schedule (think about it—the Bears had the Packers, Cardinals and Seahawks to start out, while the Lions dealt with Arizona and Seattle as well, along with Denver)?

To some extent, DVOA does take care of most of that and both articles linked above cite Football Outsiders' DVOA metric. Their opponent adjustments take into account how well a team did and how their opponents did—not just against them, but their opponents—and iteratively, how those guys did against their opponents and so on. In this metric, the Vikings rank 27th.

According to that metric, the Vikings had the weakest schedule thus far, and the strongest upcoming schedule. But that's not the only metric we can use to evaluate a team's quality after accounting for their schedule. Fivethirtyeight's Elo system has a natural strength-of-schedule adjustment to it, and there the Vikings rank ninth.

There are a lot of systems to this end, and they have different conclusions for the Vikings. TeamRankings' model has the Vikings 11th, while Prediction Machine ranks them 17th (which is the same as how opponent-adjusted point differential ranks them). Numberfire has the Vikings in between, in fifteenth.

The wide range of outcomes for the Vikings isn't because some of the models are bunk—after all, all of them are doing extremely well when tracking their picks. The Vikings are just in a unique situation for some of the models—the Vikings and the Raiders had the most variance in the models, while the Patriots had the least. The average of those models is here:

The Vikings rank 16th, though move up to 14th if you drop every team's highest and lowest ranking.

All of this assumes, of course, that the San Francisco game has a lot of meaning when it comes to determining the Vikings' quality as a unit. The research Football Outsiders has done indicates that it is meaningful (or rather, that early-season games, even outliers, improve prediction accuracy), but I can believe any number of arguments that would mark that game as an exception, particularly because the Vikings are going to suffer from brutal SOS implications once Blaine Gabbert plays for the 49ers.

Incidentally, five-win team in a six-game schedule (assuming the Vikings who played San Francisco have no bearing on the Vikings now) is very likely to be good despite the opponent adjustments—with a 54.2 percent chance of being in the top two tiers.

One model I like a lot, in part because it does a very good job against Vegas in the final five weeks of the season, is Game Script. I talked about it two years ago in the Blind Power Rankings I was producing here. Quoted:

One more thing that would be fun to add, which I've stolen from Football Perspective's Game Scripts concept is the average margin of victory throughout the game. More specifically, we know that if a team is winning by 30 points for three quarters of a game and then gives up garbage time touchdowns to bring the score within 9 points, that the "true" quality of the game is not measured by point differential.

Instead, using the point differential for each individual second in the game and averaging it out might allow us to use a popular piece of data (points) while providing for significantly more points of data with which to draw conclusions. If a team has a last-second come-from-behind victory, they may have a negative average margin of victory, which would best encapsulate the strength of that team's play over the course of the game and do a better job predicting future quality.

The wording isn't as clear as I'd like it to be, so to reiterate: it measures the average point differential a team has over every second of the game. I like eliminating/reducing garbage time in this, even though it's prone to weight early defensive or special teams touchdowns too heavily. Those rankings are below, and they include the iterative adjustments (your opponents quality, your opponents' opponents quality, your opponents' opponents' opponents quality and so on) that I mentioned earlier.

If you remove the Minnesota-San Francisco game, they jump up to 9th. This also says, unlike Football Outsiders and PFR's opponent-adjusted point differential metric, that the Vikings rank 23rd in strength of schedule so far. That fits because Detroit actually kept close to Seattle and Denver (the Broncos broke out ten points in the fourth quarter), while Chicago kept it very close to Green Bay in Week 1 and has generally had one of the toughest game script schedules.

These are all better, in my opinion, than opponent record if only because accounting for how well you played is much more important than if you played a good team or a bad team when figuring out if a team has quality. While they unfortunately played it pretty poorly against Chicago recently, they took care of Detroit and controlled Kansas City while hanging in with Denver—all things that game script and several other models take into account. Good teams beat bad teams handily, and the Vikings beat a few of them handily.

The Vikings are in a situation where their subjective strength of schedule is pretty easy to determine as well. They have not played a team that suffered from a serious injury in a way that would affect past or future performance (the Giants' win over Dallas doesn't look as good in some models as it should, because of Romo's injury and the Cassel/Weeden circus in there to replace him—not to mention Dez Bryant).

Not to mention the Vikings look like a cohesive team who can contend and be a Wild Card hopeful. A stout defense is spearheaded by a good defensive line, athletic and smart linebackers, the top safety in the game and a solid secondary. It's playing at the moment like a top ten defense regardless of the metrics and flashes top five ability. The offense is functional and often takes the points its given (seriously, they rank 18th in points per drive despite everyone's complaining about Bridgewater—and that includes Adrian Peterson's liabilities and the offensive line's issues).

The offense is performing better with Stefon Diggs in the lineup and the defense is performing better with Eric Kendricks receiving more snaps. Possibly, this team is not very much at all like the one that laid in egg in the bay.

It's not a world-beating team, but it has the parts to beat good teams and consistently put away bad ones. They demonstrated that in Denver and against San Diego, as well as their recent Detroit and Kansas City games. Inconsistency is a problem—it almost always will be for a young team—but talent, technique, instinct and smarts are giving the team a chance in every game.

If that's not at least an average team, if not better, I don't know.