Page 396 -

Q9-6 How Do Organizations Use BigData Applications?

So BI for Securities Trading? 395
What?

Since the 1970s there have been rumors of large computers
buried in nondescript offices near Wall Street, cranking out
analyses for smart stock trading. Do they work? Who knows?
If you found a correlation between, say, a decrease in the
dollar-to-euro exchange rate that influenced the price of
3M stock, would you publish it? No, you’d trade on it and hope
that no one else noticed that correlation. Or if your hedge fund
developed a model that failed miserably, would you publish that
failure? No. So, due to a lack of data, no controlled study of the
success and failure of model-based trading has been done (nor is
likely to be done).
Still, it is known that traders such as Alexander Migdal,
a former Soviet physicist, made millions of dollars in a high-
11
frequency trading firm that he started. The firm and others
like it earn small gains on hundreds of thousands of automated Source: tonsnoei/Fotolia
12
transactions. Unfortunately, such high-frequency trading places
severe stresses on the market and was responsible for the near 4. Analyze risks
13
melt-downs in 2007 and 2008. Still such trading continues, if
with a bit more control. 5. Place trades 15
Critics say there is far too much noise in the market for any Does it work? Two Sigma and other firms claim it does. We will see.
reliable-over-time predictive analysis to work. Consider, for example, We can, however, make one important observation: It
the factors that influence the price of 3M stock: global exchange has never been easy, some would say even possible, for regular
rates, oil prices, the overall stock market, recent patent filings, investors to time the market by buying just before an upturn or
patents that are about to expire, employee and customer tweets, selling just before a downturn. But today, if you try that, you’re
product failures—the list goes on and on. No model can account for not only trying to beat the market, you’re competing with Two
such complexity. Sigma, with its hundreds of PhDs and massive computing power,
Or can it? and with a slew of similar companies. For most of us, John Bogle,
Today a new class of quantitative applications is using founder of Vanguard, had it right. Buy an index fund, take your 6
BigData and business intelligence to analyze immense amounts percent, and be happy. And, over 30 years, that 6 percent will net
of data over a broad spectrum of sources. These applications a near sixfold increase.
both build and evaluate investment strategies. Two Sigma
(www.twosigma.com) is in the forefront of this new style of Questions
quantitative analysis. According to the firm, it analyzes vast
amounts of data, including corporate financial statements, 1. Consider two publicly traded companies: Apple and Alaska
developing news, Twitter activity, weather reports, and other Airlines. List 10 factors that you believe influence the price
data sources. From those analyses, it develops and tests of those two stocks. The factors may be different.
14
investment strategies. They could, in theory, model all of the 2. Pick one of the two companies in your answer to question 1.
factors that influence stocks like 3M. Briefly explain how each of the 10 factors influences the price
Two Sigma uses a five-step process:
of the stock.
1. Acquire data
3. For the factors in your answer to question 2, list sources of data
2. Create models for measuring each of the 10 factors. What role would BigData
3. Evaluate models play in processing that data?