Thu May 31, 2007 1:07PM EDT NEW YORK (Reuters) - Wall Street equity analysts have had their share of problems in recent years, from regulatory heat to budget cuts to a loss of star power and prestige. Now, with the growth of quantitative analysis, some are facing a worse possible fate: being supplanted by computers. At an increasing number of Wall Street investment banks, hedge funds and elsewhere, computers are churning out investment analyses culled from enormous pools of data. And some say they're more objective than human-based reports in determining which stocks, bonds or currencies to buy or sell. Increasingly, such programs are used to turbocharge traditional research. "Given the same set of factors, it will always produce the same result," said quant industry veteran Tanya Beder of quantitative analysis. "Its signals are pure and systematic." Beder, who built the quant trading division of top-performing hedge fund Caxton Associates LLC and was chief executive of Citigroup's (C.N: Quote, Profile, Research Tribeca Global Management in 2006, estimates that quantitative analysis and trading "drives one-third of the market" on any given day. The trend has made quantitative specialists, or quants, among the most highly valued on Wall Street, with top players commanding seven-figure salaries, similar to stock analysts a decade ago. "There's great demand for these rocket scientists," said Sandy Gross, managing partner at Pinetum Partners, a Greenwich, Connecticut-based executive search firm specializing in hedge funds. "They don't necessarily come out of financial services, but have a strong background in mathematics and statistics that can be applied to all asset classes." More: http://www.reuters.com/article/reutersEdge/idUSN3041392820070531?pageNumber=1
Here is the performance of Caxton Global Investments Limited for the last few years. It is hardly mind-bendingly exciting. Date NAV per share ($) % change (YTD) * MAY 30,2007 597.78 6.11% DEC 31,2006 563.36 13.18% DEC 31,2005 497.75 8.04% DEC 31,2004 460.72 9.97% DEC 31,2003 418.96 8.10% DEC 31,2002 387.56 26.45%
"Given the same set of factors, it will always produce the same result," (same mediocre result) "Its signals are pure and systematic."(pure and mostly) "There's great demand for these rocket scientists," (everybody wants to get in on the latest thing, even if they don't know why)
I can't wait until huge volatility comes back to the financial markets and those quants get blown out of the water. Then we will see who really can survive in this business.
Reuters is selling their program for like $1 million bucks. It makes decisions to buy or sell based on reading the latest news available instantly. It can read whatever comes out of Bernankes mouth and buy or sell. But the thing these people dont realize is you still need risk management for the thing, and just because you can create a target list of keywords doesnt mean they know anything about optimising the entry, exit, and position size for the various language based signals. Something that generates 100s of quality trades a year using 1000s of words... doesnt hold water without knowing the risk in the trade. And, that said, a system that knows the risk, and sizing often doesnt even care what is happening fundamentally about a stock or future.
If that firms Sharp Ratio or a similar measure is close to three, along with say no worse than 5% draw downs, then given those numbers you could probably raise a billion dollars to manage on 2/20. nitro
Wow you are more of an idiot than I thought if you think that a damn computer can trade better. THEY ARE MACHINES! They have no thought or feeling about the market they simply do what they are told. Sure in low volatility situations like we are having now they can probably churn out some profits but you get volatility back they don't stand a chance. Thats when you have to make quick decisions you have to actually think on your feet not just leave the trade to the computer or break out the complex calculus. Welcome to the world of commodities. And thats what separates humans from machines