Sunday, January 17, 2010

Making Sports Out Of Nothing At All: Things Sports Analysts Rely On Which Don't Mean Anything (Nonsportsmanlike Conduct!)

Today, I'll talk about the fact that sports "analysis" has very little to do with analysis. But first, just so you'll spend the day humming the same song I've been humming since I thought up this topic while I was sitting in the Babies!' room at 2:30 a.m. last night, trying to calm down Mr Bunches (who had woken up by a nightmare) and also trying to calm down Mr F (who was woken up by Mr Bunches), here's a video to watch:

The beauty of that video, and song? Even if you didn't play it, right now, you're thinking... And I don't know how you do it, makin' love out of nothin' at all. And then, in your mind, you hear that choir echoing: Makin' loooove.

That's probably true even if you've never actually heard the song -- something that's only possible if you were only just born, just this minute. And even then I probably ruined it for you as you clicked on that video and now you've heard it, too.

The other beauty of that song is how surprisingly good it is. Air Supply is one of those bands that is easy to make fun of, mostly because, they're Air Supply, but that song holds up every bit as well as, say Every Rose Has Its Thorn. Which means that we might not be very far from "VH1 Presents Russell Hitchcock's Supply Of Love," a reality show in which Russell Hitchcock, former lead singer of Air Supply, travels around the country by small plane, meeting random women picked out of a phone book to see if he can... make love out of nothing at all.

Then again, maybe we won't see that show.

The other other beauty of that song, and the group that made it popular, is how well it fits in with today's subject, which is, of course, an examination of the meaningless things that sports "analysts" rely on to provide us discerning sports fans with "analysis," by which I mean predictions.

Have you noticed how much of what passes for "news" these days is, in fact, opinions and predictions? Every time I watch CNN, which I mostly do upon returning to bed after getting the twins settle down by letting them watch a Sponge Bob DVD in the middle of the night -- don't you judge me! It was the third time he'd woken up in three hours -- every time I watch CNN, it seems like mostly what CNN is reporting is what people think about what CNN should be reporting. It works like this:

A major news event happens -- let's say, a giant monster invades Manhattan.

No, that's not quite right. Let me try again. A giant monster invades Manhattan.

Yes, there we go. This giant monster starts ripping down skyscrapers, flame-broiling helicopters, not advancing President Obama's health-care agenda, all kinds of mayhem-wreaking.

CNN then leaps into action. Wolf Blitzer fires up the diagram board to show areas of Manhattan where the monster might attack, noting that some of those areas "are believed to have been populated." Other reporters begin filing video reports reminding people that CNN can be followed on Twitter, and Robin Meade tells us, breathlessly, that the latest CNN poll is shaping up like this:

Should the government offer a bailout to the Manhattan Monster, or nuke it?
Bail It Out; It's Too Big To Fail: 51%.
Nuke It; It Might Have Explosive Underwear: 48%
Why Isn't Wolf Blitzer Out In The Field? 1%

Then we get a smattering of what you think, from the CNN Twitter page: "I just thnk the MonSTer is misundastood. Holler!" Says Deannapril, from Cold Springs.

If you were expecting to see the monster, or hear what it's done, you'd then have to wait until the next day, when viewers upload grainy, shaky cell phone camera footage to CNN.

When they're not reporting what we think about things we don't know about because they haven't actually reported on them, news shows tend to predict things -- analysts gather 'round the fake desk and say whether health care reform will pass, or whether this person or that person will get elected, or otherwise tell us what's going to happen in the next few hours, or days or weeks -- never bothering to cover what happend in the last few days or weeks. (That's what IReports are for, I assume.)

Nowhere is that problem more prevalent than in sports analysis, where virtually every single article, show, interview, or sideline-cheerleader-cam is devoted to telling you what's definitely going to maybe happen today (we hope), predictions that not only don't do anything to help you understand the game you're going to watch, or the game you've just seen; no, they don't just fail that way. They also fail because they're based on "facts" and "statistics" that are entirely meaningless.

Here's an example. Yesterday, when I breathlessly sat down to watch the big game of the day, I was a little early, and so I got to witness former Steelers/future Bills coach Bill Cowher analysizing the upcoming Ravens/Colts game. Cowher was going over what the Ravens need to do to win the game, and doing so in a very superficial manner that included, as the highlight of the "analysis," this line: pass early to run late.

I stopped listening and instead read my Malcolm Gladwell book until the game came on, because Cowher wasn't analyzing whether the Ravens could "pass early to run late," which I assumed meant pass the ball a lot to score quickly to get a lead that they could then hold by running the ball and eating up the clock... a strategy that isn't analysis at all -- it's a strategy, and it's one that would be perfectly valid for any football team ever to play any game. Instead, Cowher was just giving bland aphorisms about what might work for the Ravens. He didn't say something like "The only way to beat the Colts is to get your passing game in rhythm, but I doubt that Joe Flacco can do that because he's off and on as a quarterback and nobody in America, not even the Ravens' coordinator, can name a starting Ravens' wide receiver, so Flacco's best bet is going to be to dump it off a lot to running back Ray Rice and hope that Rice can turn that into positive yardage."

That would be analysis. Instead, we got pass early to run late.

It's not entirely the "analysts" fault... oh, wait. It is. It's entirely their fault, because in their quest to fill up all that air time without actually saying anything meaningful, they pump this gargantuan air supply...

... see what I did there?

... onto the networks, and they're, at heart, dumb guys who have no understanding of basic concepts like cause and effect, and statistics, so they simply try to rattle off things that seem to make sense (and try at the same time to not offend anyone they might have to later interview, like when Tony Dungy predicted that the Cowboys would lose to the Saints and then got ripped by the Cowboys, post-game). Being dumb guys who majored in blocking in college and having to talk a lot and being flooded with information, these "analysts" then fill you full of misinformation that you use to bet on your team, putting your mortgage payment down because you're going with Mike & Mike's Stone Cold Lead Pipe Locks, only to watch your team lose a game in which the legality of the final play will be endlessly debated, even though the team that suffered that fluky final play probably shouldn't have been in the playoffs anyway and certainly wouldn't have been in overtime at all if it hadn't been for a fluky field goal miss moments before, and wouldn't have been in overtime, either, if their quarterback hadn't begun the game with an interception that should never have been thrown.

Here's something for you Packer fans who are still stupidly mad at Brett Favre: If you are downgrading Brett because he threw an overtime interception against the Giants in the NFC Championship in the Packers' last playoff game, are you now mad at Aaron Rodgers because he threw a first-play interception in the Packers' next playoff game? How do you manage to contain that hobgoblin of inconsistency?

Here's an interesting fact for the rest of us: The Packers ended their last playoff series with an interception on offense, and began their next playoff series with an interception on offense, making them the only team ever to throw consecutive opening-and-closing interceptions in playoff games.

Back to Air Supply and how it fits today's theme. The lyrics of that song, which I've now listened to three times as I write this, have Russell Hitchcock -- remember him, from way back in this post? --

Singing about the things he knows:

And I know the roads to riches
And I know the ways to fame
I know all the rules
And I know how to break 'em
And I always know the name of the game

Including even some SPORTS things:

I can make the runner stumble
I can make the final block
And I can make every tackle at the sound of the whistle
I can make all the stadiums rock

Before getting to the big things that he doesn't know:

But I don't know how to leave you
And I'll never let you fall
And I don't know how you do it
Making love out of nothing at all

That song is the perfect template for sports analysts and their predictions for games -- they know all kinds of things, like hot to make every tackle at the sound of the whistle, which is, I'll note, technically incorrect: Tackles aren't made at the whistle; tackles are made, then the whistle blows the play dead. If you tackle someone at the whistle, you'll likely be flagged for unsportsmanlike conduct...

Sportscasters know all kinds of things. But they don't know what they don't know -- they don't know that the things they rely on don't mean anything, and so they go on making sports out of nothing at all.

I have now stretched that metaphor as far as it can go, so let me just get to the point. Here are the things that sportscasters routinely rely on to "analyze" and "predict" games, and because of the flaws in these measures, they mean nothing, making the "analysis" that you'll hear -- everywhere but this blog -- meaningless, and making sportscasters predictions random. Entirely random and pure luck.

One thing sports casters love to rely on in predicting and "analyzing" games, is Yesterday.

By "yesterday" I mean whatever just happened in the sports world. "Yesterday" is huge in the world of sports; whatever just happened is deemed to be what must happen next.

That's why everyone -- except me -- was so surprised when the Packers lost to the Cardinals last week. The Packers, you see, had won YESTERDAY; they had just beaten the Cardinals, the week before, handily. Not only that, but the Packers had won a lot of yesterdays - -they were 7-1 in the last 8 games of the season, we were reminded going into the Packers-Cardinals game. 7-1! And one of those wins was over the Cardinals! Last week!

Yesterday only matters, though, if there's no reason to think that today will be different. I like to make fun of weatherman who provide me the forecast for the day by pointing out how easy their job is, generally: Today is going to be like yesterday, only a little more so, I like to say weathermen could say, and for about 80% of the days of the year, that prediction would be right. Today, January 17, is going to be a lot like January 16 was, only a little more so.

Unless something changes, that is. Unless, say, overnight a cold front blew in and there's going to be a major snowstorm. Or unless I've traveled, overnight, to a different hemisphere. Things can change - -and they do change, and Yesterday is only helpful as a guide if you know whether things are the same today as they were yesterday.

In the Packers-Cardinals game, that was obviously not the case. The Packers were 7-1 in their last 8 games -- but they'd given up the one loss to the Steelers, a team that had thrown for 500 yards on them and scored more than 30 points because the Packers' defense had no answer for a veteran quarterback with good receivers. That's a yesterday that analysts conveniently forgot while they focused on the meaningless yesterday of the season-ending Packers-Cardinals' game, when the Cardinals, with nothing to play for, pulled their starters and didn't try, while the Packers kept their starters in for most of the game, going all out.

It didn't matter, in the end, what had happened yesterday, though, because too much had changed by the next day and nobody was paying any attention to a host of new factors: Aaron Rodgers was making his first playoff start, ever -- and he came out keyed up and full of bad decisions before settling down in the second half. The Cardinals had that veteran quarterback and excellent wide receivers. They had a reason to play in the playoff game, whereas the week before it hadn't mattered.

Sometimes, though, sports "analysts" like to look a little deeper, seem a little more important and authoritative. They go beyond yesterday to do that, and look at Ancient History.

Examining Ancient History gives the sports analysts, they hope, "gravitas." It doesn't work; putting a patina of historic importance on someone silly talking about something meaningless just makes the historic importance seem less so. Like this:

Not only does it not work, it also DOESN'T MATTER. Ancient History matters less than yesterday, but it's constantly brought up by sportscasters who want to look like they know what they're talking about.

So yesterday, we were told by the annoying-voiced announcers of the Saints-Cardinals game that Kurt Warner is "0-2 lifetime" in the Superdome in the playoffs. Those two losses came when Warner was with the Rams and lost to the Saints in the playoffs, and when Warner was with the Rams and lost to the Patriots* in the Superbowl played in the Superdome.

In 2000 and 2001, respectively. When Warner was nearly 10 years younger, playing for a different team against different teams.

Earlier in the week, one "analyst" crooned that Brett Favre, who's Brett Favre's Minnesota Vikings face the Cowboys today, has never beaten the Cowboys in the playoffs!

Did you know that! It's true! Brett Favre is 0-3 against the Cowboys in the playoffs! 0-3! Lifetime!

That 0-3 record came in losses in 1994, 1995, and 1996. Each of them was on the road -- playing at the Cowboys' home field. Each of them was against that great Dallas team that won 3 of 4 Super Bowls in the 1990s. Each of them came when Brett was with the Packers, playing for Mike Holmgren, early on in his career.

But never mind that! Brett Favre is 0-3 lifetime against the Cowboys in the playoffs! So he's probably going to lose today! Because of that history, you see! The history!

You know what I'm surprised they missed? Here's an Ancient History fact that I dug up and which nobody else will point out but which I'm sure has all kinds of meaning and portent:

The last time Brett Favre played against the Cowboys in the playoffs was in the NFC Championship on January 14, 1996, a 38-27 loss. You know who the Cowboys beat in the Wild Card round that year to get to that championship? Minnesota! Dallas beat Minnesota 40-15 on December 28, 1996!

AND, that same year, 1996, Dallas beat Philadelphia in the playoffs, 30-11, on January 7, 1996! Just like this year!

Ancient history seems to say that any year in which Dallas plays Philadelphia, then plays either the Vikings or Brett Favre, they're going to win -- in fact, Dallas did go on to win the Super Bowl that year, beating the Steelers, so bet it all, now on the Cowboys!

Or, um, remember that none of those games had any bearing... whatsover... on today's game or teams. Tony Romo was 16 and still hanging out and pitching pennies at the strip malls in Burlington, Wisconsin, when those games were played. Wade Philips wasn't a head coach; Brad Childress was coaching at the University of Wisconsin-Madison. Darrell Bevell, the Vikings' offensive coordinator, was a year out of college and coaching the Westmar University Something-Or-Others -- a college that would close in 1997.

But sportscasters jump all over historical facts like that, without bothering to say things like "You know, this is interesting but really has no bearing on anything at all whatsoever," because if they said that, they wouldn't be "experts."

Of course, they also wouldn't be "experts" without a little math thrown in -- and throw in the math they do, because anyone can look up historical facts and put them on a blog; I just did it. How are "analysts" supposed to keep their high-paid jobs spouting things that people can look up on Wikipedia (a source, it should be noted, that has been ruled by courts to be unreliable.)

"Analysts" don't want you to feel like you could get up on TV and say stuff like If the Ravens score more points than the Colts they'll win, because if you do that, they'll be out of a job and what could they do then? Westmar University closed, remember. So they throw some math at you, in the form of things like Quarterback Ratings and Point Spreads.

I don't have a song for Quarterback Rating, specifically, but in the spirit of why "analysts" tell you about Quarterback Ratings, here's They Might Be Giants' "Never Go To Work."

Quarterback Rating is thrown out over and over and over by "analysts" who want to sound like they know what they're talking about. "He's got a great quarterback rating," they'll say, or "His quarterback rating is really low."

Here's a couple of questions for you to ponder when they say that -- questions I'd ask Bill Cowher, or Terry Bradshaw, or that really annoying Rich Eisen. I'd say:

1. Do you know what quarterback rating is supposed to measure? and
2. Do you know how it's calculated?

Everyone knows that the higher the QB rating, the better, right? Or so we think. But if you don't know what it's measuring, or how it's calculated, you might as well rate quarterbacks based on how American-Manly their name is, as I do.

Quarterback rating measures, in a word, efficiency. It doesn't measure how far the ball is thrown, arm strength, or anything like that. It measures how efficient the quarterback is, and to understand what's meant by efficient, you have to look at the formula. The formula is:

(a + b + c + d)/.06. = QB Rating.

Those variables mean:

a = completions per attempt, with some math, so a = (((Comp/Att) * 100) -30) / 20

b = touchdowns per pass attempted, so b = ((TDs/Att) * 100) / 5. Note that it's not touchdowns per touchdown pass attempted; if your QB throws a dumpoff pass from his own 1 yard line, to get room to punt, that is an "attempt" which factors into touchdowns per attempt. For some reason.

c = interceptions per attempted pass, with more math, so c = (9.5 - ((Int/Att) * 100)) / 4

and d= yards gained per attempted pass, with more math thrown in, so d = ((Yards/Att) - 3) / 4

As an added bonus, there's this rule: a, b, c and d can not be greater than 2.375 or less than zero. What happens if they are? Who knows? Maybe some kind of wrinkle in the universe makes us all look like Russell Hitchcock.

Would that be so bad?

I have no idea, frankly, why the various bits of math are thrown in there -- why, for example, "Int/Att" is subtracted from 9.5. I suspect that those are "constants" thrown in to the formula to make it work out to something, the way the ancients used to use the made-up concept of retrograde motion to explain the orbits of the planets, or the way Einstein made up the "cosmological constant" to justify his belief that the universe was static, or the way scientists now use the made up concept of "dark matter" to explain why they can't explain anything.

But I don't need to know those things to know that the formula doesn't really accurately measure anything. Take yards per attempt. If, today, Brett Favre throws a pass to a running back, dumping it off behind the line of scrimmage, and that running back runs the length of the field, breaking tackles to do it, and that's the only pass Favre throws all day, his yards per attempt will be 100, which in the formula would give him a value, for d of -.7475, which can't be used. So nobody can measure how efficient Favre was or what his passer rating would be, because the entirely-fictitious formula can't account for that kind of anomaly. And the more Favre has success that day, the worse it gets. If his next pass goes for an 80-yard touchdown, his d value in the QB rating is (2/100 - 3)/ 4, or -.745 -- a little higher, but not measurable.

To make matters worse, there's more than one system for the quarterback rating. The NCAA doesn't use the NFL's formula at all. College says that a passer rating is:

Passer Rating = {(8.4 \times YDS) + (330 \times TD) + (100 \times COMP) - (200 \times INT) \over ATT}.

In the NCAA, a passer rating can range from -731.6 to a high of 1,261.6. In the NFL, it ranges from 0 to 158.3.

But it doesn't matter, either, because passer rating doesn't tell you who's going to win, and having a low passer rating doesn't mean your team will lose. The Ravens won against the Patriots* last week even though Ravens' QB Joe Flacco had a rating of 10.0. At least two quarterbacks have won games despite having a zero passer rating, which is presumably as inefficient as you can get.

Equally important, the passer rating doesn't tell you what happens after the pass is thrown. Last week, Aaron Rodgers threw a pass that should have been caught, but James Jones turned his head and started thinking about running instead of catching. That incompletion counts against Rodgers' efficiency, even though Rodgers did everything a quarterback can do: he delivered a perfectly thrown pass to a wide-open receiver, who then simply dropped it. Brett Favre, on a Monday night game, once heaved a pass skyward, throwing it behind the receiver, Antonio Freeman, a bit too much. The defender (a Viking!) got a hand on the ball, Freeman stumbled and bobbled the ball and fell to the ground, and the ball bounced into Freeman's hands:

Leading to a touchdown and overtime win. Favre got credit for that pass in his quarterback rating, even though it was a horribly thrown pass that should have been intercepted.

Quarterback ratings don't necessarily predict success. Of the top 6 rated quarterbacks at the end of the regular season, only five made the playoffs and only four are still playing -- with two of those four, Favre and Philip Rivers, yet to play their first playoff game. Mark Sanchez ranked 28th in QB rating, and he's still playing, too. His regular season passer rating of 63 is barely over 1/2 of Drew Brees' 109.6.

Here's an interesting fact, too: Mark Sanchez has the highest passer rating in the postseason: 139.4, fourteen points higher than Brees and fifty-two points higher than Peyton Manning. But who would you rather have starting for your team today: Sanchez, or Peyton Manning?

So, when "analysts" tell you that someone has something or other as a quarterback rating, do what I do: laugh, and try not to spill your Ramen noodles.

More so than quarterback rating, Point Spreads might be the most misunderstood and misused statistic or number in football, or sports. Again, I've got no song to go with this, so let's plug in another They Might Be Giants song: "Science Is Real."

They Might Be Giants obviously don't know any modern-day "scientists," or sportscasters, because if they did they'd know that "science" today is anything but real. In a world that thinks the Point Spread has any meaning, how could science be real?

The Point Spread is used by "analysts," and bettors, and, probably, you, to help determine who people think is going to win, because people assume that an expert, somewhere, has looked at a game and analyzed it and determined how likely it is one team will win and by how much.

But it doesn't work that way. At all.

The point spread is set by a few large Las Vegas sports books, and it has one purpose and one purpose only: To get an equal number of people betting on each team. Las Vegas sports books make money off of people betting on games, but not off of people betting on games and losing; they make money off the number of bets placed on games, taking a cut of each bet. An ideal system for a Las Vegas sports book (or any bookie) is to have an equal number of people betting on each team. When that happens, 1/2 the people lose, 1/2 the people win, and the payout from the book is equal to what they took in, less the book's cut (usually 10%, called the vig. Yep, they really call it that.)

So in a typical game, like today's Cowboys vs. Brett Favre's Minnesota Vikings matchup, the books hope that half of all bettors take the Cowboys, while half of all bettors take Brett Favre's Minnesota Vikings. That way, when Brett Favre's Minnesota Vikings win, the books take all the money Cowboys' fans threw away, and pay it to all the people who like Brett Favre (me, and Deanna Favre), and the books take 10% off the top, coming out like bandits.

To get half of you to bet on the Cowboys, the books have to offer a point spread -- they have to give points to one side or the other. Today's point spread for the Cowboys-vs-Brett-Favre's-Minnesota-Vikings game is 2.5 to 3, and it's been that way all week. What that means it that the books didn't think that people needed much incentive to bet on the Cowboys.

The key to the point spread is it's not based on experts, or "analysts" or statistics, or anything. It's based on what you think.


You're the hidden expert behind the point spread. The books set the point spreads to get half of you to bet on a given team, and how many points they give depends on what you think -- that is, if you aren't likely to take the Cowboys, no matter what, the point spread will be huge. If you're likely to take the Cowboys, the point spread will be small.

Another key point in the point spread? More or less half the people in the world think you're wrong.

No matter which team you pick.

If you take the Cowboys, and the three points, then half of everyone who bet on that game thinks you're wrong. It's set up that way. So you take Brett Favre's Minnesota Vikings, instead... and half of the world is betting against you. It's set up that way.

That's something the "analysts" don't mention, do they? I'm not sure they know it. But they never say something like "I'm going with the points and the Cowboys, even though that means that half the world disagrees with me."

Because if they said that, they'd have to explain why they were saying that, and why they felt they could disagree with half of everyone who exists.

There's two NFL games today, and two next week, and then the Super Bowl. Over the course of the next month, you're going to hear, over and over and over, about what happened Yesterday and about Ancient History, and you'll be told who has a better passer rating and what the favored team is...

and it'll all mean nothing. Nothing at all. Watch and see how much, now, actual analysis you get over the course of that month, and how much, instead of analysis and thought, what you get is a bunch of sports made out of nothing at all.

No comments: