Why?

Jonathan Willis
January 24 2009 10:35AM

Statistics and mathematical modeling have changed the way society functions.

Meteorologists use new methods of gathering information, especially from the atmosphere, to create a far grander database than any that human society has used for that purpose before. They feed these data points into supercomputers, and create complex models that involve a multitude of variables -- often variables that are in a constant state of flux or that are difficult to get firm information on.

Edward Lorenz, who served as an army weather forecaster during World War II, and then studied and taught meteorology, coined the “butterfly effect” as shorthand for how tiny variations could affect weather models: “Does the flap of a butterfly’s wings in Brazil set off a tornado in Texas?” He pioneered the field of chaos theory largely in an effort to better predict meteorological changes. Scientists continue to use long-reaching models to predict damage caused to the earth by global warming and human impact, while similar (albeit less complex) models are used all over the Earth for long-range forecasting.

The world economy, tracking the financial dealings of seven-billion plus through a dizzying array of legal jurisdictions, tax laws and the like, is tracked through a series of macroeconomic statistics; the ones produced at Berkeley use GDP, CPI, unemployment rate, corporate profits, change in business inventories, housing starts, interest rates, exports, and personal savings rate, among others. It’s on the basis of similar economic models that men like the Governor of the Bank of Canada or the Chairman of the Federal Reserve issue projections and make their decisions -- decisions that affect millions.

Even the advertising industry has made extensive use of mathematical modeling. Taking statistics compiled by government and the private sector, they break the population into demographics, and target their ads to different sectors of the population. By doing this they hope to better market their products. In 2001, The Coca-Cola Company alone spent nearly $2 billion on advertising -- money that was largely distributed on the basis of demographic information compiled by market research companies.

Even sports teams have turned more to mathematical models in recent years. The book Moneyball, based on the success of Oakland Athletics’ GM Billy Beane, is generally considered to be the work that brought statistical work into the mainstream in baseball. Theo Epstein, originally a PR man with the San Diego Padres, worked his way up the ladder and in 2002 was hired by the Boston Red Sox as General Manager. He was 28; the youngest GM in the history of Major League Baseball, and a man who’d never played the sport at even the high school level. Using sabremetric principles (math pushed by the Society for American Baseball Research), his team has won two World Series championships in his six years at the helm.

These are just a few examples; in virtually every scientific field, whether “hard” science or the social sciences, mathematical modeling is a primary tool for extrapolating all kinds of factors. It’s used in other sports, it’s used in scenarios that are for less predictable than what occurs in the carefully regulated confines of the arena, and often it’s used with data far less complete than we have available for free from the stats page and play-by-play reports at NHL.com.

In most of these other fields, suggesting that your powers of observation and gut feeling were a better method for prediction than mathematical models using a vast database would get you laughed out of a job.

In hockey, it’s the only commonly accepted practice -- and I have only one question. Why?

74b7cedc5d8bfbe88cf071309e98d2c3
Jonathan Willis is a freelance writer. He currently works for Oilers Nation, Sportsnet, the Edmonton Journal and Bleacher Report. He's co-written three books and worked for myriad websites, including Grantland, ESPN, The Score, and Hockey Prospectus. He was previously the founder and managing editor of Copper & Blue.
Avatar
#51 Ender
January 26 2009, 12:49PM
Trash it!
0
trashes
Cheers
0
cheers

Mike wrote:

Ender wrote: But for every Briere there are a dozen Hejdas who come out of nowhere (I say this, ccertain that you’ll pull numbers from his euro league days, but I still think it’s beside the point). Is that really beside the point? I would say a gem toiling in a Euro league is EXACTLY where the “seen him good” approach will fail, simply for lack of resources. It’s both easier and cheaper to pay an intern to put all the numbers in a spreadsheet and flag the guys who float to the top. Hejda only “came out of nowhere” if you didn’t know where to look.

I even said " I’m not saying stats can’t help that," if you would have followed the quote for another sentence or two. All I'm arguing is that stats are not necessary. I'm not saying they're not useful. That said, there are so many mitigating factors that anyone who bets that kind of money only on stats will always be crazy. The house always wins, as they say. However, A scenario exists in which stats are not necessary, and a better decision can be made than using all the stats in the world, and that's in-person scouting.

Sorry, but you're arguing against a straw man, and it is beside the point.

Avatar
#52 speeds
January 26 2009, 05:53PM
Trash it!
0
trashes
Cheers
0
cheers

Hejda is also a strange example because the guys who watched him and related to him all year (Oilers management) didn't think him worth re-signing (or, at least, the didn't re-sign him), while CLB did, to a 1 year, 1 mil contract.

Meanwhile all the stats guys in the oilers blogs were crying, hoping and praying that EDM would re-sign Hejda as the stats they had compiled indicated that he was an underrated defencemen. And they were right.

Avatar
#53 Ender
January 26 2009, 06:37PM
Trash it!
0
trashes
Cheers
0
cheers

@speeds

Q - "In hockey, [Watching a player is] the only commonly accepted practice — and I have only one question. Why?"

A - " I’m not saying stats can’t help that. I’m saying that stats *cannot* tell the whole story. However, between “seen him good” and access to people around the player, not using stats *can* tell the whole story. Not saying it definitely will or *should* but I still think it *can*.

Whether or not people use the tools at their disposal for the best result is another story altogether."

I'm not going to argue these straw man cases. I always end up trying to explain and re-explain myself in these threads, and I'm done with that. Here's another Q.

Q - "When people use non-stats arguments, why do the stats people argue "but stats show that!" when it has absolutely nothing to do with what the person is even saying?"

Avatar
#54 speeds
January 27 2009, 09:14AM
Trash it!
0
trashes
Cheers
0
cheers

What is a straw man about Hejda?

Your contention is that by watching him play the Oilers should have known he was worth keeping? Admittedly everyone makes mistakes, but who saw him more than the Oilers?

Avatar
#55 Ender
January 27 2009, 02:44PM
Trash it!
0
trashes
Cheers
0
cheers

It's not a matter of whether or not you think he was worth keeping. It's a matter of whether or not he fit into their long-term plans. Until you can tell me with any level of certainty exactly what they were thinking when they let him go, it's a straw man argument.

Avatar
#56 Doogie2K
January 28 2009, 04:09PM
Trash it!
0
trashes
Cheers
0
cheers

The problem with "saw him good" is that, as the British commercial goes, it does exactly what it says on the tin. You see him good one night, and don't see him again for another couple of months, what do you know? If you watch a guy every night, that's fine, but even then, it's easy to miss things. The stats are a tool to help fill in the blank. Using only stats or only saw-him-good are, I think, fool's errands. Using both things in their proper contexts, as well as things like talking to a player, his coaches, etc. (if you can) seems like the best way to get the whole story on a guy.

Put another way, the problem with stats has less to do with whether they can tell you everything -- rare and mistaken is the person who tells you so -- and more to do with whether they tell you what they think they're telling you, and that's a usage and sample-size problem. The problem with saw-him-good is that it purports to do exactly that, and I don't believe that it can. Your eye can catch certain things the stats can't, just as the stats can catch certain things your eye and memory can't.

As for Hejda, I was under the impression he refused to come back because of how he was treated (benched for most of the first half of the season). As for the stats, well, I saw him good all the way back in the pre-season that year, so make of that what you will.

Comments are closed for this article.