Imagine Magazine - Johns Hopkins - November/December 2013 - (Page 10)

A Safe Bet The world of finance is, at its core, one of uncertainty-prices go up and down based on everything from supply and demand to company rumors and the weather. All those factors play a role in determining whether buying or selling-and how much-will generate the most money. In order to remove as many unknowns from the equation as possible, financial mathematicians create complex models of economic markets to reveal where the best investments lie. Today, many of those mathematicians partner with-or are themselves-computer scientists, writing code that executes trades automatically based on the models they've developed. Experts have estimated that 50-60 percent of all stock trading in the United States is now automated. Such trading programs use algorithms that analyze current data about stock prices, how much is being bought and sold, and how quickly it is moving to figure out the most profitable trade to make. At their fastest, these programs can do that analysis and make a trade in microseconds, before a human even sees the data. This type of trading is often referred to as high-frequency trading. Financial mathematicians initially design trading algorithms based on how humans make similar decisions. For example, one of the first algorithms, introduced in 1969 by programmer-mathematician Thomas Peterffy and trader-psychiatrist Dr. Henry Jarecki, included conditions such as "sell gold on Fridays, except when gold prices were down the previous four days." According to journalist Christopher Steiner's account in his book Automate This: How Algorithms Took Over Our Markets, Our Jobs, and the World, Peterffy and Jarecki started with that rule because it was what the human traders in their firm were already doing. Today, trading algorithms that start with one financial model can improve themselves over time in a process called machine learning, in which the algorithm compares the results of all the trades it makes and tweaks its rules based on which ones led to the most profits. Although a hurricane affects only a portion of the world at a time, the most accurate way to predict one is to model atmospheric conditions around the globe. That requires solving equations for factors including temperature, humidity, and wind speed and direction at points all over the world, problems that can take several hours for the most powerful supercomputers to solve. Because weather depends on so many dynamic factors, the physics equations used to describe it are never perfect. Nor do meteorologists have a complete set of data on the conditions everywhere on Earth at any given time. They must therefore rely on a diverse set of mathematical models to make predictions. Some of those models attempt to calculate global conditions while others are confined to one region, and some analyze historical weather patterns while others use only recent data. The National Hurricane Center lists more than 40 commonly used models for tracking hurricane trajectories and intensities. Meteorologists then combine the top three or four predictions into one "consensus" forecast in order to say when, where, and how hard a storm is likely to hit an area. Models can also help scientists studying climate change. By looking at decades of data about atmospheric conditions and temperatures, for example, they can see patterns in how those variables have changed over the past century or more and can predict how future changes to the atmosphere, such as the addition of greenhouse gasses, will affect global temperatures. 10 imagine Nov/Dec 2013 SHUTTERSTOCK-ALL IMAGES ON THIS PAGE Storm Watch

Table of Contents for the Digital Edition of Imagine Magazine - Johns Hopkins - November/December 2013

In My Own Words
Interested in Econ
The World in Numbers
Reckoning with Randomness
Elliptic Curves
A League of Our Own
More Than Math
Developing Your Numbersense
Where Math Meets Imagination
Selected Opportunities & Resources
Dancing in the Footsteps of My Ancestors
Off the Shelf
Word Wise
Exploring Career Options
Planning Ahead for College
Students Review
Mark Your Calendar
Knossos Game

Imagine Magazine - Johns Hopkins - November/December 2013