Stats Gone Crazy: How Much Information is Too Much?

image

AP photo/Keith Srakocic

Yesterday morning I was reading ESPN’s game recap for Game Two of the Stanley Cup finals. The winning goal, for those of you not into hockey, was scored by rookie wingman Conor Sheary. Which is awesome. How awesome? The public needs to know:

With the goal, [Sheary] became the first rookie to score an OT goal in a Stanley Cup finals since Brian Skrudland in 1986 for the Montreal Canadiens, according to Elias research.

But there’s more. Of course there is:

Elias research also notes that Sheary is the first player to score a winning goal in OT in a Stanley Cup finals after his team lost the lead in the last five minutes of the game since Patrick Kane in Game 6 of 2010 against the Flyers.

Thanks for sharing. I’m a better person for knowing that, amIright?

What is it about stats? I used to hate statistics, much like, I gather, most people. Perhaps “hate” is not the mot juste for something as dry as statistics. It’s probably better to say I was uninterested in them, and deeply cynical as to their use.

My eldest son, who graduated with a degree in vocal performance from Carnegie Mellon University, was hired as the Systems Administrator for the Statistics Department. Go figure. He had the full panoply of music and vocal instruction and nary a computing class outside of the mandated “Introduction to How Not to Screw Up the Machines in the Computer Clusters” class. They called it something more politically correct than that, of course. Which just goes to show that most kids should have to work at McDonald’s for a few years before being allowed to attend a university.

All my persistently-held stereotypes about statisticians were confirmed when he reported an incident which occurred when he went around to introduce himself to all the professors. One of the first he met shook his finger at my son and said “I’d better not hear you singing in the hallways!” He was quite serious.

But something  funny happened when I became a Steelers fan. I began to find the compilation of data and the charting of it to be quite fascinating. I quickly discovered it was more interesting to me than to almost anyone else, but I couldn’t help myself. And as I discovered the appeal of other Pittsburgh sports, particularly Major League Baseball, I was inevitably sucked even further into the dark side.

Let me be clear—I’m not one of those people who fills out the score card in the program at baseball games. I assumed only old guys did that, although when I laughingly noted this to my young piano tuner (he’s still in his 20s) he looked abashed and said “I do that too.”

And in the past few years I’ve noticed more and more bits of statistical trivia being dropped into football broadcasts. Football is not a game which lends itself to the same sort of minute statistical examinations as baseball because the sample sizes are so much smaller, but the guys in the NFL broadcast booths are certainly trying.

Okay, I don’t suppose it is the broadcasters—they’re just reading the stuff on their feed. But when you start transmitting such things you get situations like Antonio Brown’s streak of “5 receptions for 50 yards.” The sensible reaction to this is “um, cool, I guess.”

But that wasn’t how AB felt about it, and as a result there were perhaps times when the ball was forced to him to attempt to keep his streak going. I can bet Mike Tomlin didn’t care about this streak. I’m surprised he didn’t tell AB that he “walks past six Lombardis every day, not six commendations for the longest receptions/yards streak.” Maybe he did. But I’m guessing it was worth it to Ben Roethlisberger to make AB happy if he could. Receivers are all divos of one type or another, not excepting the working man’s receiver of choice, Hines Ward, and they have to be treated as if they are special if you want them to go risk their lives and livelihoods trying to catch a ball you may or may not have thrown well.

But the things about statistics is, properly used, they can be predictive. This is why the Pirates have been on the cutting edge of statistically-informed methods which go far beyond the Moneyball stuff, although they grew out of it. Defensive shifts are perhaps the first and the most copied. But as a small-market team they have to be constantly searching for ways to pick up minor, hopefully cumulative advantages, because their payroll is substantially smaller than over half the league.

The differences are staggering. The team with the smallest payroll—this season, the Milwaukee Brewers—has a payroll that is just over a quarter the size of the team with the largest payroll—this season, and just about every season, the LA Dodgers. The Pirates, who are usually right around the bottom of the league, are currently at No. 19 out of 30.

Because of the enormous amount of data collected in baseball, statistics make sense. They track absolutely everything in baseball, and can use, for instance, a given batter’s tendencies as to where on the field he hits the ball to align the defense more effectively. It works, at least enough of the time to make it worth doing.

With football, there’s still stuff you can collect, although for most things the sample sizes are a great deal smaller. And teams are doing so, including the Steelers, who hired a stats guru last summer. But the guy they hired, Karim Kassim, is a stats guru with a difference. I’m making an educated guess here, but his extensive background in social psychology and the neural basises for thought means he’s actually presumably working on what Kevin Colbert calls “hearts and smarts.” And, not surprisingly, he’s working in the scouting department.

So really, what looks to me like a fantastic draft is going to be something of a test of his effectiveness. There are other components, of course. You can only draft who’s available. But I’m guessing that whether Artie Burns was a ‘reach’ is ultimately going to come down to how effectively Kassim assessed his psychological profile. The physical tools were there for everyone to see, as was the rawness and inexperience.

We have a recent example of a guy who appeared to have all the physical tools, to the extent that the Steelers gave him a nice contract extension. The only problem was, he fell apart. I’m speaking of Cortez Allen, naturally. I don’t know whether Kassim could have predicted this or not. But, given Allen’s background at the Citadel and so on it was certainly a surprise, and I’m guessing that if anyone could have predicted it it would have to have been someone like Kassim.

How many members of this draft class actually manage to make it on the field and make an impact in the short term, and how many stick in the long term, will be an interesting test. A draft class can be extravagantly praised at the time, only to look bad in retrospect. The 2008 draft is certainly an example of this. The three players who stayed in the league the longest, OT Tony Hills, S Ryan Mundy, and QB Dennis Dixon, are all currently “free agents.” None of them, or, for that matter, the other guys chosen, were playing for the Steelers by 2013, and most had departed long before. Their first round pick, Rashard Mendenhall, never lived up to his promise.

Could someone like Kassim have prevented this? I’m guessing he would have at least nixed the Limas Sweed pick. But who knows?

The point is, statistics can be useful and predictive, even, to a limited extent, in football. But a great many of them are just silly. To return to the original example, does it really make Sheary’s goal on Wednesday night any more special because he was the first guy to score in OT after the team had lost the lead during the last five minutes since, uh, 2010? Nope.

What does make it a bit more exciting is that Sheary is not only a rookie but an undrafted one, one who has spent much of the 2015-16 season shuttling back and forth between the Penguins’ AHL team in Wilkes-Barre and the NHL team before finally being called up for good. This is a human interest story. Another is the fact that captain Sidney Crosby drew the play up just before the face off. That’s another piece of human interest. The bit about the “first guy” etc. is a water cooler factoid.

What I love about the Steelers’ stat guru is that he is focusing on the human part of the equation. Ultimately, for me football is about the guys who play it, and I am becoming more and more excited as I write the profiles of the incoming players.

Especially in the later rounds, the Steelers seem to have focused on the guys who have found a way around their physical limitations and just got it done in college. They also seem to be exceedingly solid citizens. Those two things might well be related.

I look forward to collecting the data and charting their progress, of course. I said I can’t help myself. But more importantly, I look forward to recounting their progression from raw recruits to seasoned professionals, and from promising rookies to men of character.

One comment

  • Stats certainly do have a place. It does seem lately that seemingly meaningless stats are being used to trump up players. Sorry, but I don’t think that AB’s 5 for 50 made him a better player. It was a pretty cool stat, but also pretty much out there. I do recognize that it gives a lot of credence to his production, and it was certainly a very good achievement. Maybe I’m stuck in the realm of only being impressed by the overly impressive due to the information junky society we’re in today. I just don’t think that I care about how a QB’s 84% completion percentage when a receiver runs a slant on 3rd and 1 against zone coverage in the 3rd quarter of tied games played in the Pacific time zone after Halloween make him a great QB.

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s