Hola and welcome to the divisional round volume of our guide to upsettery in the NFL in 2017. As regular followers will no doubt have noted, last week was an interesting one for Upset Watch.

Sometimes, your eyes don't believe what the raw data is telling you, and I made the classic mistake of not playing the probabilities correctly. We still had a good week, but it should have been a great week, and that is on me. Our raw data told us to pick the Titans, Bills and Falcons to cover. They promptly did, but thanks to my conservative (small c) decision to go rogue and hedge, we ended up with only 2 of those games right when it could have been 3.

This morning one of the experts we feature on out Against the Spread page, Case Keefer of the Las Vegas Sun, had this to say about last week's slate being covered entirely by the underdogs in comparison with last year's Wild Card round, where favorites covered without exception:

"It should have served as a reminder that there’s no tangible connection between two separate playoff fields. Any suggestion to the contrary is nothing more than a mutation of the gambler’s fallacy.

Handicapping should almost always be focused on the current teams and the data associated with them, not loosely-correlated situations that had their own sets of variables."

Now, I both agree and disagree, and here's why...

First off, Case is correct in one sense. Last year's WC games in isolation have very little to do with this year's. We shouldn't assume that 4-0 to favorites is enough to go 4-0 towards favorites this season.

But here's where I think he's wrong: When NFL teams play in the same situations, we can see some correlation between what has happened in the past, and what happens in the future. Why? Because NFL teams inevitably replicate the mistakes and successes of each other, and players themselves react similarly to corresponding situations they have faced before.

Coaches are the primary reason for this. The coaching carousel in the NFL is very much closed to anything too wild or risque. You've got a good idea? At the first sign of negativity, it'll be dropped like a stone, and if - like Chip Kelly for example - you're a coach who becomes attached to a new idea that fails, you can expect a relatively short shelf life in the NFL.

So new styles of play, coaching techniques and situational playbooks have a tendency to look remarkably similar among teams that are -in name - extremely different. The players who come into any team, at any position, are inevitably coached in the same ways, regardless of their team. They are taught not to improvise and to retain discipline in following their coach's schemes. It's not necessarily about whether a team is calling the exact same play as another team, it's about whether the players respond in the same way.

But it's more than that.

This is about situations. This isn't about teams. NFL players follow incredibly generic paths to the NFL. Rich or poor, good or great, they go through the same college football teams, the same coaching, and guess what? They're usually successful on an individual level. Very few NFL players come into the league having been bad in college.

And then BANG. You get drafted by a bad team.

But for the vast majority of players, the NFL isn't continuous winning and losing (unless you get drafted by Cleveland...), but a cycle of wins, losses, comebacks, chokes, long road trips, passionate home crowds, hot weather, cold weather... you name it, NFL players experience it.

And then they leave and go elsewhere.

So that experience, it's wasted right? Well, not exactly. Players are encouraged to be generic, and because the coaches largely lack imagination, the players have a tendency to find themselves playing similar situational football. When they're heading into a decisive game, the chances are that many of the players on the team have experienced something similar before, be it for the same prize or not. It's not based on position, height, speed, race or any other individual factor, it's about the collective nature of the NFL. Don't be an individual, be part of a team.

So when I started the project that is gathering pace at Pickwatch HQ, it was with this in mind. How have teams reacted when they've been in this situation before? And if they've never been in this situation before, how do other teams react in similar scenarios? Which factors are most important? How do we know when to discount scenarios or teams from our comparison?

The results have been illuminating, as followers will know. What I can tell you about our project, is that we have been working on the biggest change to the site in the last 5 years, and we are looking to give you the tools to work with this information on a week-by-week, game-by-game basis in 2018, as well as the expert data you know and love. I'll have a lot more on this as the offseason progresses.

Season Performance:

Last week we went 2-2 thanks to the aforementioned rogue audible to not take the Titans. Breaking even saw us remain at +74 units on the season, going 99-72-3 (59%) on the season.

This week's Divisional Round picks:

Notes

First off, nothing can prepare us for the unpredictability of the playoffs. As we saw last week, there were almost upsets in all four games, and in both of the Saturday matchups, the underdog won outright. This week may be no exception, but as always, we play the odds in our predictions. That means ascertaining how many upsets are likely, and where those situations occur most frequently. We then apply it to the divisional slate and select the most likely candidate(s).

The first stat that jumped out this week was probably the most basic, in that favorites are a whopping 16-4 straight up in the divisional round over the last 5 years. That essentially puts us in a situation where we can realistically expect no more than one outright upset this week.

But when we got into ATS territory, such a simplified rationale won't do. Favorites are 10-10 over the same span, and so, as you'll read below, each game needed to be assessed across a variety of factors...

As with last week, the games are in order of the schedule, rather than likelihood of an upset.

Atlanta (-3) at Philadelphia

What makes this game a particularly big anomaly is that this is the first occasion in six years that a backup QB will lead his team into the Divisional round. The only prior statistical precursor was that Brock Osweiler returned from the bench last year to lead the Texans against the Patriots, where they were soundly beaten 34-16 (a cover, but we'll go into New England's propensity to cover the spread later...)

So Nick Foles, particularly as the QB of the no.1 seed? Yeah, it's fair to say that's unusual.

That said, Atlanta are the pick for a number of reasons. First of all, teams that played in the previous season's Super Bowl are 4-1 in the last 5 years. This includes anomalies such as the 2013 49ers on the road as favorites by 2 points against the Panthers, the only other instance of a road favorite in the divisional round over this period.

In addition, every single one of those four underdog victories over the last five years came when the underdog had actually won a playoff game the previous season. I often talk about psychology in this page because it's incredibly relevant to NFL players. Often, at this stage of the season, the primary difference is not talent, it is how players react to the dynamic that surrounds the game.

The Falcons went to the Super Bowl last year and the Eagles are not a team with playoff experience. As a franchise they have not won a playoff game in ten years, and very few of their roster are holdovers from the last loss in the 2013 wild card round (Foles is ironically one of them). This may sound over-simplified, but there is no statistical reason to suggest the Eagles are destined to win on Saturday. If they do, it'll be an extraordinarily unpredictable moment moment.

Tennessee at New England (-13.5)

Some teams play as favorites so often that rather than looking for wider trends and analysis, like Case Keefer pointed out in the above quote, we can actually look to them alone as a guide to how they perform in certain situations.

Tennessee really have no playoff history, but underdogs who have battled the odds to make the playoffs in week 17 and then win as underdogs in the wild card round do, and it should be noted that those three teams all lost, but they did so while covering points spreads of 7 points or more...

So right now, we're leaning Titans, however there are some reasons to think this may not be as good a predictive tool as in the wild card round. First of all, while all of those teams had different scenarios to the Titans, with three veteran QBs who all had playoff experience beyond the divisional round, and two with Super Bowl wins (Brees and Flacco). The Titans head to Gilette with a QB in his first playoff run, and while that's not necessarily a negative, it puts the previous stat into a different context. This is what I mean about digging as much information as possible about each scenario.

Perhaps more importantly, the Patriots are still the Patriots, and their roster turnover is not great enough to override the fact that they have won every game at this stage for the last 5 years, and gone on to 3 Super Bowls, winning two.

In that time they've covered 4 out of 5 times, each spread over 7 points.

The only question is the size of the spread. What can we take from a spread so unusually high at this stage? Well, the 2016 Patriots knocked out the aforementioned Osweilered Texans and a 16.5pt spread. In fact, the 2014 Seahawks covered a 13.5pt spread against the Panthers who arguably had more playoff experience than this Titans team.

The Patriots are an extraordinary team and they defy comparison. I have often found that the best statistical precedent for events in the Tom Brady era is in fact, the Tom Brady era. Until they walk into the unknown and become the underdogs for the first time in nearly two decades, their own performances as favorites are our best tool.

Jacksonville (+7.5) at Pittsburgh

AH HAH! Upset I have ye. I'm not sure on the Jags to win, but I think it's possible and our friend Neil Greenberg of the Washington Post agrees. He's one of the very few picking the Jags this week in Pittsburgh.

You know what I think? There are often factors that override previous meetings between two teams in the same season. For example, if the first meeting of these two teams occurred in Jacksonville, I'd factor that in accordingly. If there were key injuries or events, those go in too, but in this case, there are enough similarities to make it an interesting statistical tool.

The Jags struggled to beat the Bills as favorites last week, which they actually did in week 4 against the Jets too. That week they lost, and then responded with arguably their most defining game of the 2017 season, a 30-9 upset of a +7.5 point spread in the Steelers' back yard.

Now, I'm not suggesting that will be the case again this week, because the stakes are far greater and the Steelers much improved, but I think the Jags have a good shot at beating this spread. First of all, don't read anything into Blake Bortles's performance last week. When the Jags won in October, they did so behind Leonard Fournette, with Bortles totalling only 95yds passing.

Secondly, favorites are unusually bad at covering spreads in this 7-9pt range (2-5) over the last 5 years, despite winning 6 of the 7 games.

Then there's the Steelers and their ATS record this season. They closed out the season 1-5 vs the spread despite losing only once - their only cover coming in week 16 against a derelict Houston team that could barely function under the signal calling of Tom Savage.

As much as this Jags team is reliant on good defensive play, the Steelers are unusually adept at the narrow win. It's possible that they learned enough in week 5 to enact a crushing revenge win on the Jags, but I'm not convinced that is the case unless the Jags turn in a complete Bortles classic...

New Orleans at Minnesota (-4)

The hardest game to pick this week came down to this. For me this is the best matchup of the playoffs so far, and like the above game, has some statistical backdrop from the regular season. These two teams met on Monday Night football in week one, where Sam Bradford produced a career best performance to utterly destroy the Saints at home.

Of course, Bradford is o longer the starter (although interestingly, he is healthy enough to be in contention to back up Case Keenum...) and so we can't read an awful lot into that. Perhaps more importantly, the Saints improved dramatically over the intervening period.

So what this came down to for me was that the teams' individual ATS records over the last few months ahve gone in complete opposite directions. The Vikings are 9-1-1 in their last 11 games, often as slight favorites. Meanwhile, the Saints are 4-7 over their last 11. I think more importantly, they aren't so good as underdogs, going 1-4 this season when they aren't favored.

The two teams in this game are pretty evenly matched. There's a case that can be made that it's a 50/50 call on the Vikings beating the spread yet still winning, which is unusual in itself, but ultimately, these two teams and their 2017 form is as good a guide as we need. We only have room for one upset this week I'm afraid...