Tuesday, February 14, 2012

"Can Moneyball statistics be used to beat Jeopardy!

From Barking up the Wrong Tree:
NPR covers the fascinating story of Roger Craig, a PhD in computer science, who used data-mining and statistics to make hundreds of thousands of dollars on Jeopardy.

Using data-mining and text-clustering techniques, Craig grouped questions by category to figure out which topics were statistically common — and which weren't.

"Obviously it's impossible to know everything," Jones says. "So he was trying to decide: What things did he need to know? He prepared himself in a way that I think is probably more rigorous than any other contestant."

Once he'd calculated the odds a category would come up, Craig quizzed himself on a variety of questions to find the gaps in his own knowledge.

But Craig says it's a fallacy for game-show candidates to think they need to know everything.
"They want to learn every capital of every county in the world. And you really don't need to," he says. "Instead, you need to know the 80 percent people have heard of."

That's because the show, he says, isn't written for the contestants. It's written for the people playing along at home, who need to feel they have a fighting chance to keep up.
Which of course reminds me of one of the great moments in Jeopardy! history.
From Feb. 2010's "Three Stocks that Didn't Go Down Today: Tesla, Solyndra, Codexis (CDXS; SOLY; TSLA) and a Money Management Story":

...As I was re-writing the headline I couldn't help thinking of Cheers' Cliff Clavin.*

*When Cliff makes it to the Final Jeopardy question with an insurmountable lead, Alex Trebek asks the contestants to identify Archibald Leach, Bernard Schwartz and Lucille Lesueur and says
"Cliff, unless you've done something incredibly stupid like wagering everything, you've won"

Cliff bet all his money.
His answer was "Who are three people who have never been in my kitchen".

The categories for the game were:
Cliff's "Dream Board".