<
>

# Behind The Bets

Newsflash: Understanding risk is what separates everyday squares from the big-money wiseguys who live off their gambling wits.

Now, here's a piece of candy: Since 2000 the New England Patriots are 50-16-3 vs. the spread when they've won at least 50% of their games and the over/under line is 42 or less. It's true. The tip comes from a math genius with a data-mining program that can apparently find gold. Again: 50-16-3! What's this mean? Well when the Pats face against Oakland on Dec. 14 in a contest whose over/under total is 40, you should probably lay the seven points most bookies are offering.

That said, this gift-wrapped pick comes with a lesson: On the surface, data like this is no more valuable than a pile of iron pyrite (fool's gold if you are not geologically hip). How you handle it, even if you don't want to be a wiseguy, will be the difference between betting success and failure. If you're serious about the former, then understanding risk is key. Knowing that a sure thing still involves risk is a start.

The man who provided that generous tip above is named Mike Orkin (you can check out his software at www.snoopdata.com, although the numbers only go through 2006. My man Scooch hooked me up with the Pats stats for the past two seasons). Orkin's got a Ph.D. in Stats from Cal-Berkley, is the Professor of Statistics Emeritus at Cal-State East Bay and runs the Business, Math and Sciences depo at Laney College in Oakland. He's also written a couple of books about the relationship between math and gambling, and even taught a course explaining how the theory of probability was discovered because a degenerate, 17th-century dice player needed an edge. Imagine if your math teachers had used craps to teach you about adding, subtracting and statistics (think Prezbo in The Wire). Would American kids' math scores today be so much worse than the rest of the world's? Anyway, that's another column for another website.

But there's serious danger in fancy-looking stats. As Orkin explains it, "One of the big mistakes people make is looking at empirical data and seeing trends, sports bettors love that. But if you let software programs run they will find things, end patterns that are actually just random noise. If you generate coin-tossing data, instead of wins and losses, and construct an identical database you will still find trends like this Pats example. You have to be careful."

Nobody likes to think that a stat leading to an apparent mortal lock is actually random and unreliable. We count on those gimmes to balance out the truly impossible picks, like the home Ravens minus-1.5 over the Steelers this weekend. But, Orkin makes the case that taking the stat at face value leads to irresponsibleâ€”i.e., badâ€”betting. Even if you're right on Sunday, in the long run you'll lose big. That's because, all together now, you don't have an understanding of risk.

Don't feel bad. Part of this is inherent. Not even the smartest kids in the class excel if they don't have good instincts. Success requires an imagination that considers the unlikely event and then plans for it in whatever formula you create. Just ask the math wizards who wrote all those algorithms for trading sub-prime mortgages. They couldn't know that greed from both bankers and buyers would combine to create a financial black hole. What looks better, a wonderfully low interest rate, or 50-16-3? Greed (and hope) says they both look good. But understanding risk forces us to ask how it could look so good in the first place.

"People who don't understand what risk is wind up exposing themselves to enormous losses, and you have to have a good intuitive feeling about that," says Orkin. "In sports betting there may be examples where you have long sequences of a situation that has been winning for years, but you still have to account for the unexpected and unknown."

Like the bubble eventually bursting. Or a ref stealing a touchdown from the Steelers.

If you're serious about making a trend work for you, you need to figure out why it's happening, not just know that it is happening. The good news is that this knowledge gap can be filled, and it will take as much time as you already spend researching matchups, a comparably useless endeavor. When you've uncovered a trend, test each element individually. In the example above, how do the Pats do against the spread when they have won at least 50% of their games on the season? How do they do when the over/under is less than 42? In the games they lost, what happened? This might help you account for the impact of unlikely events. Then you'll not only be betting like a pro, but thinking like one, too.

I am totally going to do this. Very, very soon. But, for now, I'm choosing blind ignorance. And the Pats.