Even as traders drown in volumes of market data that would have stunned their predecessors back in the 1980s and '90s, they want the market data spigot turned up, not off. "People are trying to rationalize and make sense of the amount of data they need, but depending on the type of fund you operate, the more data you have, the more effectively you can use that data," says Scott DePetris, COO of Portware.
And hedge fund managers like Olympian Capital Management's Arun Kaul are looking beyond the surface of today's data, digging deep into yesterday's data to uncover an edge. "I like to dig as deep as possible," says Kaul. "Doing due diligence in terms of the completeness of the data is key."In terms of a look-back period, the rule of thumb is, the longer, the better.
One market data analyst tells Advanced Trading about an investor that wanted historical data that went back to the Civil War. While he might not want to venture back to the 1860s, Kaul says he prefers a long view. "For fundamental data, several years or decades is useful," he relates. "For price data, it depends on the duration of the alpha signal. If it is short-term, we can use a shorter data set; but if we are doing longer-term strategies, then I would require a larger data set." Kaul adds that a long-term data set allows a researcher to separate secular changes from cyclical changes and lets him isolate a trend or alpha decay versus spurious relationships, which typically are of a shorter duration. This is often where complex event processing, or CEP, comes into play.
OneMarketData, for example, announced a recent deal with Galium Capital, a start-up high-frequency hedge fund based in San Francisco that trades multiple asset classes. The tick data provider's VP of global sales and marketing, Richard Chmiel, says Galium wanted to back-test its strategies, use historical analysis on its tick data and deepen its research. The fund "also wanted to perform real-time signal generation in the CEP engine," according to Chmiel.
For an industry that seems doomed to repeat its mistakes (see: the dot-com and housing bubbles), buy-side firms have a passion for historical tick data -- say, the price of orange juice after a sharp frost in an election year when a Democrat is looking to return to the White House. But traders need the numbers to crunch. "Access to deep history and the ability to store data going forward is becoming more and more important to back-testing strategies," says Bill Ruvo, global business manager, real-time solutions, for Thomson-Reuters. "A cloud-based solution," he adds, "is ideal for buy-side firms that have little desire to maintain terabytes of data on-site."