Introduction is here.
“All we need are enough inputs. The genetic algorithm will do the rest. We’ll make millions”, Tom told a skeptical Dan.
Tom Winchester had a fascination with genetic algorithms ever since he first learned about them in COMP203A, the legendary difficult programming course at the University of Waikato. He survived 203A, and later went on to complete a phD in electrical engineering at Auckland.
But now, after working as a VB6 programmer for a monolithic software house for the last six years, he was needing a new challenge.
And so over a beer at Galbraiths, a local pub, he was trying to convince his friend Dan to join him in his venture. Tom had known Daniel Alexander-Smith since his days as a post-grad.
He met Daniel whilst hunched over one of the mainframes at the Auckland engineering lab. At the time, Tom was at a loss. For the past three weeks he had been trying to find a bug in his program that was causing it to crash after five iterations. Dan, a newcomer to the lab, and keen to get his hands on the mainframe, volunteered to take a look. He rewrote the section of code that was causing the problem saving Tom months of headache.
They’d stayed in touch ever since.
“I’ll design it, you can write it”, suggested Tom.
“You’re talking about predicting the stock-market. That’s impossible. It’s a chaotic system”, argued Daniel. “Even if you could, the program’s very existence would cause it to implode in a logical fallacy”
“You’re right, it’ll never exactly predict, but there’s no reason why it can’t get close. Very close. Besides, our current stock market is largely driven by a bunch of 25 year old economics graduates on Wall Street. You know we can do better.”
“Yeah, but Tom, the 25 year olds aren’t just playing by a set of rules from 101 economics”, argued Dan. “They make decisions that a computer never could. An earthquake, the daily ramblings of the president, cash rates, the merger of a bank, it all affects stock prices. A program can’t predict the impact of those events, you need a brain for that.”
“Not anymore” stated Tom, as he finished his beer. “We now have all that information available online in the form of news reports, and other live statistics. And the real beauty is that information is already filtered in the form of blogs and comments. All we’d need to do is weed out the relevant stuff from the gibberish, and the genetic algorithm works out the rest.”
“Well if you’re so convinced why don’t you do it? Why do you need me?”
“I will if I have to”, replied Tom. “But it would be much easier with a coding guru on hand.”
“Whatever, I need another beer. What are you having?”
“Macs Gold”, replied Tom.
“Look at it this way”, reasoned Tom, as Dan returned with two more pints. “In the worst case we’ll release the source code on sourceforge as a cool experiment, and in the best case Fairfax will offer to buy our work for 100 million dollars and we can take early retirement.”
“And buy mahogany power-boats, and impress girls with flashy iphones”, Dan wishfully mused.
“Exactly. We can’t lose”.
“All right, what do you want me to do?”
“Well the first step is to write the framework for an evolutionary genetic algorithm…”
Part 2 is here.