Category Archives: NFL

Which Comeback Was Most Unlikely?

Last night was ridiculous. Only Brady and Belichick can go down twenty-five points in the Super Bowl without their fans losing hope. And to score thirty-one straight points to do it? And to win the first ever overtime Super Bowl? And to do it while setting records left and right? Holy crap.

Then again, the Patriots are really just continuing a trend that’s been going on since June. This past year has been filled with tremendous comebacks and upsets. Like the Pats, the Cubs, Clemson, Trump, and the Cavs have all accrued huge deficits and overcome them. But which comeback of these five was the most improbable of all? Let’s figure it out.

5. Donald Trump (28.2% chance of winning)

This was a difficult probability to determine. While we can rely on Vegas and ESPN’s win expectancy for sports, there are a number of conflicting predictions out there. Accordingly, I went with Nate Silver and FiveThirtyEight’s empirical prediction model, rather than predictions from places such as the New York Times (15%) and the Princeton Election Consortium (1%) that were less based on statistics.

Although the Trump’s victory seemed unbelievable and shocking at the time, statistically, it wasn’t all that unlikely. Accordingly, all four sports comebacks were more miraculous than the election results.

4. Chicago Cubs (21.7%)

The Cubs’ comeback seems unlikely, and it was, but they were favored in each of the fateful final three games, by significant margins. Vegas expected Chicago to win each of the individual final three games, so the fact that it did isn’t too surprising.

Based on the money line, Chicago was expected to win 67.7% of the time in Game 5, about 59% in Game 6, and 54.5% in Game 7. Combined, they had barely better than a one in five chance of winning three straight games and emerging victorious at their nadir, which, while improbable, isn’t quite the act of God that it appeared to be.

3. Clemson Tigers (9.6%)

Clemson’s low point came after a Calvin Ridley reception gave Alabama a first down. At that point, in the middle of the second quarter, Clemson was down two touchdowns against a dominant defense. Despite this, the Tigers came roaring back to draw within three.

Later on, O.J. Howard’s 68 yard touchdown catch pushed their odds of winning back down to 14%. Once more, Clemson fought hard and eventually took the lead for the first time, gave it up, and finally regained it with one second remaining.

This was an impressive victory against the Evil Empire of college football, and the situation for Clemson was dire for large chunks of the game. However, the deficit was only fourteen points at its largest, so it doesn’t seem like the Tigers had a major comeback, even though their win expectancy reveals they did.

2. Cleveland Cavaliers (5.9%)

Both the Cavs and Cubs had to win three straight games, including two on the road, to break a mammoth championship drought. So why was Chicago nearly four times as likely as Cleveland to end its season with champagne?

The main reason is that the Cubs were superior to the Indians, while the Cavaliers were inferior to the record-setting Warriors. Furthermore, basketball is much less reliant on luck than baseball, so Cleveland needed to legitimately beat a dominant, otherworldly Golden State squad three straight times–a lucky, fluky win wasn’t possible.

Because of this, the Cavs’ championship comeback was the most unlikely of all until…

1. New England Patriots (0.03%)

The Pats had a huge comeback over the Falcons last night: after going down 28-3, New England roared back to score thirty-one points unanswered. The Falcons had a win probability of over 92% for thirty straight minutes, including a twelve minute streak when it never dipped below 98%.

The situation was bleakest after an incomplete pass from Julian Edelman left the Pats facing a 3rd and 3 from their own 46 with just over six minutes left in the third quarter while down 28-3. ESPN gave the Patriots a 0.03% chance of coming back to win the game at that point.

Even if you want to use the least optimistic Trump prediction, which gave him a 1% chance of winning the election, last night’s Patriots victory was three times as unlikely as Trump’s triumph.

Any way you slice it, this Super Bowl comeback was nearly impossible, and was the most improbable of the last year’s five unlikely moments. Of course, that’s just from a quantitative standpoint, which can’t capture the in-the-moment insanity that accompanies any spectacular event, along with all the other qualitative factors that go into making a memorable moment. Which one do you think was the craziest comeback of all? Let us know in the poll and in the comments!

Sources: FiveThirtyEight for Trump, ESPN for Clemson and New England, and oddsshark.com for Chicago and Cleveland. 

The Ravens Are Off to the Least Inspiring 3-0 Start This Century

The Baltimore Ravens are off to a good start. They currently sit in first place in the AFC North with a pristine 3-0 record. Although its season has gone perfectly so far, Baltimore can’t exactly be ecstatic about how it has earned an undefeated start. Is there any reason for Baltimore to feel good about the future based off its 3-0 start?

Let’s first examine this question from a qualitative standpoint. Over the course of the 21st century, there have been 79 teams to begin the season 3-0. After examining every single instance, I’ve identified three of the least inspiring 3-0 starts since 2000.

The 2004 Jaguars got off to a 3-0 start with a point differential of 7. That’s not 7 per game; that’s 7 total. At least the teams it beat were fairly good. The Jags defeated a solid Bills team that finished 9-7 in Buffalo by 3, the Broncos who earned a wild card with a record of 10-6 by 1, and a lousy Tennessee team that finished 5-11 by 3.

Last year’s Falcons beat the 7-9 Eagles at home by 2, a 6-10 Giants team that couldn’t finish games by 4, and a 4-12 Cowboys team by 11. Sure, the eleven point victory looks good, but that game took place in Week 3 and Dallas had already lost its best two players: Dez Bryant in Week 1 and Tony Romo in Week 2.

Baltimore has managed to beat three bad teams–the Bills, Browns, and Jaguars–like the 2015 Falcons while also barely scraping out victories, with edges of 6, 5, and 2, like the 2004 Jaguars. That’s not exactly the combination that the Ravens are looking for.        

Now let’s take a look at this problem from a quantitative perspective. Per FiveThirtyEightteams that get off to a 3-0 start on average finish with 10.7 wins. However, based off their point differential, the Ravens’ Pythagorean expectation for wins is 1.9. If Baltimore was slightly lucky rather than incredibly fortunate, it would be 2-1. According to FiveThirtyEight, teams that start off 2-1 finish with, on average, 9 victories.

Now, take into account that the Ravens are on pace for nine victories without strength of schedule taken into account. The first three teams Baltimore faced this year are three of the worst teams in the NFL. If the Ravens can only capable of earning a meager edge in point differential against some of the worst teams in the league, imagine what will happen once they start playing teams that are actually good.

Sure, the Ravens already have three wins in the bank and they can’t be taken away. However, no matter how you slice it, Baltimore’s start to the season has been inordinately uninspiring and provides little hope for how the rest of the year will progress.

How Good is Denver’s Defense?

The Broncos’ defense is incredible. It’s otherworldly. It’s unstoppable.

While those three superlatives are perfectly reasonable things to say, they don’t say why the Broncos’ defense is so good. Instead of continuing on with opinion and conjecture, let’s take a look at a few statistics to determine just how good Denver’s defense is.

Simple measures of defensive performance are useful but limited. For instance, the Broncos allowed 18.5 points per game this season, an elite rate. However, that only ranks fourth in the NFL, behind Seattle, Cincinnati, and Kansas City. Clearly, this statistic is unable to show how great Denver’s defense is.

Instead, let’s use DVOA (check out this article for an in-depth explanation of DVOA). The Broncos’ defensive DVOA is an impressive -25.8% (remember that positive numbers represent more points so a negative DVOA is good for defenses). How good is that? First let’s take a look at a graph.

Screen Shot 2016-02-14 at 3.24.23 PM

This graph shows the distribution of defensive DVOA in the NFL in the 2015 season. Thirty of the thirty-two teams in the NFL are in the same cluster in the middle. The square in the upper right corner is the laughably bad Saints defense. The star all the way in the bottom left corner? That’s the Broncos.

Just look at the gap between them and second place. Although it’s not quite as large as the one between New Orleans and 31st-ranked Chicago, it’s still a fairly sizable gap. That’s how much better Denver’s defense is than everyone else’s.

Here’s another way to measure how dominant Denver’s defense has been: z-score. If you’re unfamiliar with the term, z-score measures how many standard deviations a given data point is over or under the average of all the data points in the sample.

Denver’s z-score is -2.31. As a comparison, here are eight other notable defenses in recent years and their z-scores based on defensive DVOA. Remember, a negative z-score is good for defenses, as it means that they are successful at preventing points (or producing negative points, if you will).

Screen Shot 2016-01-14 at 7.11.38 PM

As you can tell, this year’s Broncos defense stacks up very well against some of the best defenses in recent memory. A couple of non-Broncos notes:

  1. The Buccaneers were crazy-good back in 2002. Given a normal distribution, the chances of a defense being better than Tampa Bay’s in 2002 are just over .01%. Wow.
  2. To many, the 2000 Ravens own the title of best defense of the 21st century. However, on this list, they rank last. Why is that? Well, if you look one slot above them, you’ll find the 2000 Titans, who (obviously) played in the same season as them. Standard deviation, and, as a result, z-score, can be affected by the presence of a team of a similar caliber. Accordingly, the Ravens’ z-score is dragged down by the presence of the Titans.

There’s one more way to show how great Denver’s defense is this season. Given a normal distribution, the chances of a defense being better than the Broncos’ defense are 1%. That means that you’d expect a defense as great as the Broncos’ to show up roughly once every 3.125 seasons. So, with these Broncos, you’re seeing something that only occurs about three times a decade.

Although we’ve answered the titular question pretty thoroughly, there remains one, far more important question that has yet to be answered: Will this spectacular defense result in a championship?

Not having a star quarterback is a significant roadblock to winning the Lombardi Trophy, but not an insurmountable one. In recent years, pedestrian quarterbacks such as Joe Flacco, Brad Johnson, and Trent Dilfer have won Super Bowls. Flacco was helped along by an extraordinary hot streak, but both Johnson and Dilfer were accompanied by excellent defenses, both of which appear in the chart above.

Now, since we’ve already established that Denver’s defense is superb, it would stand to reason that they would have a genuine shot at the championship despite their lackluster quarterback situation. That’d be true if it were Brock Osweiler who were starting for the Broncos at quarterback in the playoffs as he’s been competent enough to allow them to win games. However, rather than Osweiler, it’ll be Peyton Manning starting under center for Denver. Manning, unlike Osweiler, has proven himself to be totally incompetent this season, throwing an interception of 5.1% of his attempts, an absurd rate.

While great defenses can carry mediocre signal callers to Super Bowl victories, even the best defenses can’t overcome abysmal quarterbacking. And, even if Manning were playing like he was in 2006, I certainly wouldn’t want to entrust my Super Bowl hopes to a guy who looks like he’s plotting to murder John Elway.

640pey1

DVOA FAQ

What is DVOA?

DVOA is a statistic created by Football Outsiders and stands for Defense-adjusted Value Over Average. Translation: DVOA shows how good (or bad) you are compared to an average team. DVOA is a percentage, with a league-average DVOA being 0%.

How is DVOA calculated? 

Put simply, DVOA is calculated by comparing every single play in the season to a league average baseline. It’s then adjusted for strength of opponent, game situation, and field position. Additionally, DVOA values successful plays more than other plays. A successful play is considered to be a play that gains 45% of the yards needed for a first down on first down, 60% of the yards needed on second down, and 100% of the yards needed on third and fourth downs. In other words, if you gain eleven yards on a 3rd-and-16, it matters a lot less than gaining six yards on a 3rd-and-4. DVOA is able to take this difference into account, unlike simpler statistics.

How can I interpret DVOA?

On offense, a positive DVOA means that you’re better than average while on defense, a better than average team has a negative DVOA.

What’s the reason for this discrepancy?

DVOA is a measure of how many more points you create than an average team. On offense, the best teams create more points than an average team, so their DVOA is positive. However, on defense, the best teams prevent more points than an average team, so their DVOA is negative, because they create “negative points”.

Why is DVOA better than simpler statistics like points scored and allowed?

DVOA is better than simple points scored and allowed because of its increased accuracy and predictive value.

How does DVOA gain its accuracy?

As mentioned before, DVOA is adjusted for a number of factors. In addition to the three mentioned above, DVOA is based on plays, rather than points. This is important in a couple of ways.

One, if the offense throws a pick-six, that shows up as points allowed for the defense. However, it was obviously the offense’s fault, not the defense’s. Simple statistics attribute those points to the defense, but more advanced metrics like DVOA correctly assign blame to the offense.

Two, while points allowed only records drives that end in scores, DVOA takes note of every play on every drive. For instance, let’s say that Team A gives up fourteen points and 250 yards per game, while Team B gives up fourteen points and 350 yards per game. By simple measures such as points allowed, Team A and Team B have equal defenses. However, by DVOA, it’s clear that Team’s defense is better than Team B’s on a play-by-play basis.

Are there any other details I should know?

There’s one important factor we’ve only mentioned so far, and that’s DVOA’s adjustment for field position. If your quarterback throws an interception and your opponent starts their position on your one yard-line, then it’s very likely, even if your defense is spectacular, that you’ll give up a touchdown. The opposite is true as well. DVOA gives your offense more credit for long drives than for possessions that begin in your opponent’s red zone.

Simple statistics don’t measure that. To them, a one yard drive for a touchdown is equal to a methodical fourteen play, ninety-two yard drive for a score.

There are various other permutations of DVOA that we won’t discuss in this particular article. One notable one, however, is DYAR, which is DVOA but in a cumulative statistic, rather than a rate one.

In future football articles, we’ll be using DVOA a lot more, and I hope you find this FAQ useful so you’ll be able to be a more discerning football fan. And, even if you still don’t understand it, it’s a fancy sounding statistic with an acronym; just using it in a sentence makes you sound about three times as smart.

Should the Giants Sit Their Stars?

The Giants are finally nearing the end of yet another lost season. Only one more game remains: a matchup against the Eagles at 1:00 this Sunday.

While they get all the attention, teams that are locked into their playoff seeds aren’t the only teams who have to think about benching their stars. This week, the Giants face an intriguing decision: should they sit their studs to keep them safe or should they play them for pride?

The main reason the Giants should consider sitting players like Eli Manning and Odell Beckham Jr. is that it makes no sense to risk your star players over a meaningless regular season game. Football is an inherently dangerous sport, and the more one plays it, the more likely it is that one gets hurt.

Those who think that teams should play their stars, even in meaningless games, like to point out that Week 17 is no more dangerous to players than any other game, as shown in the graph below.

In this graph, (found here, at Football Outsiders) the orange line shows that the risk of sustaining a new injury is stable throughout the season (Week 1 looks like a significant outlier, but, as the article notes, “anybody who suffers an injury in the offseason, training camp, or preseason will first appear” in Week 1’s stats). Accordingly, the logic goes, teams should play their stars because the risk of sustaining a new injury in Week 17 is no greater than the risk in any other game.

There’s an easy retort to this argument. While it’s true that a player is not more likely to get injured in Week 17 than in any other week, it’s also true that a player is more likely to get injured playing in a Week 17 game than sitting it out. Why should the Giants increase the risk of an injury to Manning or Beckham Jr. even slightly by playing them in a meaningless game?

It’s important to note that this graph does not indicate what kind of injuries were sustained. Obviously, there’s a significant difference in importance between a mild hamstring strain and a torn ACL, and that difference needs to be taken into account when determining the risks of playing one’s stars in a meaningless game. Logically, there’s no reason why the rate of serious injuries in Week 17 should be different than in any other game, but if anyone has any statistics on the matter, please email me at sushionsports@gmail.com.

Many will say that Eli Manning’s streak of 150 consecutive games started should not be broken, and that he needs to play to keep the streak alive. However, there’s a simple solution to this problem: play Manning for the first series of the game, and then sit him down for the rest of the afternoon.

The biggest reason the Giants should play their stars is out of loyalty to their fans. This game is being played in the Meadowlands, the Giants’ home stadium. The cheapest seat in the stadium, according to StubHub, is $50. That’s a significant amount of money to shell out for a team that’s not trying to win.

The Sixers, the preeminent tanking team, have realized that no one wants to pay a lot of money for a team that intends to lose. Accordingly, for their next home game, against the Timberwolves, the cheapest ticket available on StubHub is for a grand total of $9. That’s all. The price for a seat right behind the basket is $44. That’s cheaper than the cheapest seat in MetLife Stadium!

The Giants owe it to their fans, who are paying a lot of money to come watch the team play, to do their best to win the game. If it was on the road, that’d be another story, but it’s not. People come to Giants games to see Beckham Jr. and Manning, not Sidney York and Ryan Nassib.

And, just to prove my point, you definitely didn’t realize that Sidney York is the name of an electro-pop band, not a football player. Rather, it’s Elliott Brood who’s the current WR6 on the Giants.

So what do I think the Giants should do? It’s a tough decision, and I’m glad I don’t have to make it. Still, if I had to choose, I’d lean towards playing the stars for the sake of the fans. If I’m the Giants, I definitely don’t want to be compared to the Sixers.

And besides, who wants to see guys like Elliott Brood play football? Well, actually, it might be pretty awesome as, although you didn’t notice, Elliott Brood isn’t a wide receiver; it’s a Canadian alt-country band. And unless you want to see these guys playing for the Giants, you’d better hope the starters are playing on Sunday.

This article can also be found at Jock Journal.