Thread: ESPN Football Power Index - The Broncos are no longer a good football team

Fair prediction I think...

---

Today, FPI is going all-in.

Normally, our Football Power Index delivers information in degrees: a percentage chance of this and the likelihood of that. That's the responsible approach to take with a model like this one. But on the precipice of the NFL season, we're throwing caution to the wind and letting FPI deliver predictions with stone-cold certainty.

It's never advisable for an NFL head coach to preside over a bad year, but some coaches would likely get a mulligan if it happens in 2017. The data says these guys, on the other hand, almost certainly wouldn't.
Of course, keep in mind that in order to do this exercise, we're taking the information that FPI is giving us and then making a leap toward the black and white. FPI does not feel 100 percent certain about anything -- there's a 3.1 percent chance the Patriots will miss the playoffs, after all -- but we'll pretend, for a minute, that it does.

For what it's worth, the 50 percent marker is not special here, either. Some of these predictions are less than 50 percent likely to occur, but that's because they may be the most likely scenarios of possible options or simply because this is, after all, a story about bold predictions. Anyone can guess that the Browns will be bad.

----

The Broncos are no longer a good football team

While Denver missed out on the postseason last year, the Broncos were still widely considered a very solid squad. Those days are over.

FPI projects Denver to win just 7.7 games in 2017, thanks to a lackluster-at-best offense.

While the Broncos' 2015 team proved it could win despite poor quarterback play, it is still an exceptionally difficult task to accomplish. With Trevor Siemian under center again, the defense will have to be just about perfect, and FPI isn't quite as optimistic about the Denver defense this year -- ranking the unit seventh-best in the league."

http://www.espn.com/nfl/story/_/id/2...s-win-afc-west

You got me interested in the merit of FPI, so I wanted to find out more about it. Here is an excerpt from interview with analytics team than built the model.

What was something that you unexpectedly found mattered (or mattered more than anticipated) in the preseason ratings or game projections?

ZB: As I expected, performance from past seasons is a strong indicator of performance in subsequent seasons. However, to my surprise, offensive performance tends to be noticeably more consistent from season-to-season than defensive performance. The NFL is a QB-driven league, which contributes to this trend.

Good chance they are right. I see us right around an 8-8 team, much like this model projects.

Originally Posted by BroncoWave

Good chance they are right. I see us right around an 8-8 team, much like this model projects.

Big wildcard is offense. If the offense isn't significantly better, than with the tougher schedule, it's going to be tough to have a winning record.

they could easily be right about our offense. . . this ain't the 7th best defense, though, so i'm going to discredit them in my mind. . . eff off, initials!

Originally Posted by Tned

Big wildcard is offense. If the offense isn't significantly better, than with the tougher schedule, it's going to be tough to have a winning record.

I think the defense is more of a wild card than people want to admit as well. As the guys who do these analytics said, defense is harder to maintain year to year than offense. We don't really know yet what the impact will be of losing Wade's coaching or Ware's leadership. And with rumors of Ward being shopped, his potential replacement would be another wildcard. I still think it will be a really good unit, but it wasn't as good in 16 as 15, and if it's the same as 16 or worse this year (which is very possible), then it will be tough to win more than 8 even if the offense is better than we think.

7.7 wins.

The single most important factor in that prediction is 2016 offensive production. The greatest uncertainty is how much of jump the QB will make in year 2.

So the QB portion appears to be a unique aspect of NFL FPI. Can you go into more detail how you determine how much a QB is worth, compared to his backup?

ZB: Quarterback impact is measured using a similar approach to Total QBR, with a few differences. The biggest difference from QBR is the incorporation of an aging curve. Young quarterbacks tend to improve with time and the performance of older quarterbacks starts to decline. Given a quarterback's age in the current season and age/efficiency in past seasons, we can estimate a quarterback's expected impact. To obtain a more predictive than descriptive quarterback metric, the effect of extremely lucky and unlucky plays is limited, and there is no down-weighting for when the game is less close as there is in QBR. Players without previous experience are set at replacement level.

If it's a jump as suggested from his 2016 performance and the performance curves based on every QB snap taken since 2006, then the data suggests 7.7 wins.

The part that seems fuzzy to me is this bolded part at the end:

In the preseason, FPI uses a number of predictive factors to project future team strength. The main component of preseason FPI is Vegas expectations; the expected win totals and money lines for each team are an accurate representation of predicted team strength and provide a strong baseline for teams entering the season.

But relying solely on Vegas has its flaws, and more information is needed to determine what percentage of a team's projected win total can be attributed to its offense, defense and special teams units -- the components that make up FPI.

To gather more information on each unit, ESPN polled a panel of NFL experts regarding the expected offensive, defensive and overall performances of teams for the upcoming season.

How much of this is data-driven, and how much is 4 guys in a room making power rankings?

Originally Posted by Hawgdriver

How much of this is data-driven, and how much is 4 guys in a room making power rankings?

The bolded part makes it seem that quite a bit of it is. If they're asking a panel of experts their opinion regarding overall performance and then using that to effect their model then this metric is trash.

Originally Posted by Davii

The bolded part makes it seem that quite a bit of it is. If they're asking a panel of experts their opinion regarding overall performance and then using that to effect their model then this metric is trash.

I'd prefer they just stuck with Vegas and gave no weight to the opinions of a "panel."

Originally Posted by Hawgdriver

How much of this is data-driven, and how much is 4 guys in a room making power rankings?

Well, when they started talking about how the QB portion (after as much as saying that's synonymous with the entire OFFENSIVE portion) is closely related to QBR, that pretty much tells me four staffers spitballing over the water cooler is about an order of magnitude more "important" here than actual onfield performance.

Plus: "As I expected, performance from past seasons is a strong indicator of performance in subsequent seasons"? What league has HE been watching? The extreme WEAKNESS of correlation between consecutive NFL records is PRECISELY why I quit trying to make preseason predictions in the first place. As they note, the obvious ones are safe: Cleveland and Jax will remain victims of their FO until the owner dies or sells the team; NE* will always cheat and usually get away with it. Everyone else though, from the guys NEAR (but not IN) the cellar or penthouse to those who just barely missed/made the playoffs? Good luck matching this seasons performance to the last ones, because even a few key injuries and/or FA losses are enough to transform a 10-6 team into a 6-10 team. Didn't last season LOOK pretty good at the start?

We opened by winning our SB re-match, then beat Luck by a pair of TDs, then went on the road for Siemian to torch a Bengals team that had finished in a three-way tie with us and NE* for best in the AFC. Except it turns out the defending 15-1 NFC Champs finished 6-10 last year, the Colts were .500 again and the defending 12-4 AFCN Champs finished 6-9-1. Beating all those teams that were great in 2015 didn't mean squat, because they all sucked in 2016: We were just a little better than suck.

When will people learn the Rasmussens' "statistical predictions" only marginally beat combing through a dead chickens entrails? There's a long statistical history of THAT.

W​pi

19-0

Actually, what the "Football Power Index" says of its OWN predictive history is itself telling:

Granted, FPI thought Carolina would walk into the playoffs last year, but then again, who didn't?

Nearly everyone who's ever heard of the SB losers slump, which is nearly everyone who's ever heard of the SB. Guess the FPI didn't factor in things like that, or the "intangible" crisis of confidence that can go with getting so close to being the best only to be brutally and repeatedly beaten down as Camolina was by our D, or the far more tangible crisis that results from a teams best FAs getting the offers that come with reaching the SB without the incentive to stick with a successful organization that comes with WINNING one.

So what ARE the FPIs factors? Well, much like QBR, ESPN doesn't explicitly or exhaustively tell us nor anyone, only selectively site individual metrics as they become relevant and/or convincingly fit data to the predetermined curve. Special teams are measured, which we're told to explain why KC will win the AFCW: Because they've had top five STs the last four years, #1 in two of them! Which makes one wonder why they only won the division ONCE in any of THOSE years, and missed the playoffs entirely in one of them.

FPI doesn't specifically factor Elliott's suspension into its calculation, but that is a factor in the Vegas win total (and the cut taken by the sportsbook, also known as the vig), which is an input for FPI.

So they don't factor in his suspension at all, except where they do, sort of, though HOW is completely opaque. Alrighty then.

5. The Jets will draft another quarterback from USC

THAT'S NOT EVEN A PREDICTION ABOUT THEIR RECORD! So what's it got to do with predicting their record?

FPI believes the Jets are the heavy favorite (32.8 percent) to land the No. 1 overall pick, and fortunately for Gang Green, a quarterback is atop Todd McShay's recently released list of the top 32 players in the 2018 draft. Yes, USC's Sam Darnold could be the next savior in New York. Get those "Broadway Sam" headlines ready.

Ah, right: It relates to the bones some other ESPN writer cast to predict next years #1 overall pick (which itself has a lot to do with which team finishes dead last; I wonder if the two authors discussed both matters much together? Still, if he IS the #1 overall pick, the Jets are the leading candidate to have that shot at him, because FPI is confident they're terrible. INCREASINGLY confident, in fact:

Believe it or not, FPI actually thinks the Jets are even worse now than last week, when we noted that the model thought they were particularly terrible.

Okay, FPI, I'll bite: What STATISTICALLY happened within the last WEEK that made the Jets even worse than they already were? I'm OK with math; you can just give me the raw numbers and I'll work out the rest on my own. Because the only POSSIBLE stats that come to mind in that brief time span are from Saturdays crosstown preseason game with the Giants, which the Jets narrowly lost by a single point when their 2PAT failed. All we've been hearing the last month, just as every August, is how utterly useless meaningless games between camp fodder are for predicting the best teams once starters are playing games that COUNT. Apparently this is news to ESPN; odd, since sports news is kind of their raison d'etre.

The best part (of course) was the precisely stated 49.6% chance (basically a coin flip) NE* will return to the SB and the equally precise 32.3% chance they'll WIN it again, alongside alongside predicting they'll only win a modest (by dynasty standards) 11.8 games because "There's just too much variance in the NFL. Not too much randomness to give the Eternal Champions <3:1 odds to win the SB (an order of magnitude better than a purely random chance) but too much to predict they win >12 games. Because even "That's more than a whole win more than any team has been projected by FPI to earn in the last three preseasons."

In other words, NEW ENGLAND IS THE FIRST AND ONLY TEAM FPI HAS PROJECTED TO WIN >11 GAMES IN FOUR YEARS OF TRYING.

Given that OVER A DOZEN teams (nearly 15%) have done just that, including all but one SB team and every Champion, I'm starting to think that maybe preseason games where the guys playing half the downs won't even be on the team Opening Day are a poor predictor of regular season performance. Again, I believe pretty much everyone who knows anything about pro football has always known that—so where does that leave ESPN...?

Holy Hell, is it time for REAL football yet?! I don't know how much more filler and manufactured narratives I can take....

I am definitely in the minority but I think it's ridiculous to be paying a linebacker 20 million a year, even more so for a team that most likely won't make the playoffs.

With the additions of Bolles, Leary, Barbre, and Watson, the return to health of Paradis, and
experience of McGovern, the O-line is far better than last year. Perhaps it is not yet top ten,
but it is improved and deeper.

The RB corps is far deeper than it was, and injury to C.J. or another will not be as devastating
as it was last year. The receivers are deeper. And Siemian is one year removed from his
rookie year now.

This is a much better offense than last year's. It won't live up to Emmanuel Sanders'
prediction (#1 in the league), but it will be improved.

The defense? So a much deeper interior line and edge amount to a decline when the rest of
the defense (ILB and DBs) remain the same?

I understand the team might be challenged to reach a 10-6 record (my prediction), but that
will be because of the brutal schedule, not a decline in talent. I just think that ESPN is way off.

Originally Posted by topscribe

With the additions of Bolles, Leary, Barbre, and Watson, the return to health of Paradis, and
experience of McGovern, the O-line is far better than last year. Perhaps it is not yet top ten,
but it is improved and deeper.

The RB corps is far deeper than it was, and injury to C.J. or another will not be as devastating
as it was last year. The receivers are deeper. And Siemian is one year removed from his
rookie year now.

This is a much better offense than last year's. It won't live up to Emmanuel Sanders'
prediction (#1 in the league), but it will be improved.

The defense? So a much deeper interior line and edge amount to a decline when the rest of
the defense (ILB and DBs) remain the same?

I understand the team might be challenged to reach a 10-6 record (my prediction), but that
will be because of the brutal schedule, not a decline in talent. I just think that ESPN is way off.

Hope you are right top, and wrong.

I hope you are right about the offensive line. That is the motor for the whole offense.

I hope you are wrong about the record.