Trading has become increasingly automated with the result that Wall Street has a lot fewer traders than it did a few years ago even while volumes are rising. But traders aren’t going away.

Richard Korhammer, CEO of Lava Trading, estimates the street has lost a third of its sell-side traders, yet volumes keep rising because trading has become so automated. Two factors have contributed to this shift, he added.

Decimalization reduced spreads, which meant that traders were less profitable for their firms. Compression of commissions means that buy-side customers are paying less per share.

Where have they gone? Some have left for the buy-side, which can use tools like Lava to reach liquidity without going through the sell-side, or with just a brief stop at a broker to conduct an electronic trade. Lava, like the direct market access firm UNX, is entirely built on Microsoft.

Other traders have left to start hedge funds, which often use algorithms to send thousands of trades into the marketplace looking for a good execution that won’t impact prices.

At TradeTech USA, Joseph Wald, the CEO of EdgeTrade, sought to provoke the audience by asking whether traders, algorithms or direct market access were on the border of obsolescence. He saw life for all of them. Direct market access, before algorithmic trading engines, required a lot of work by the traders to locate pools of liquidity and execute orders. Many of the ECNs have been acquired in an industry-wide consolidation drive. But Reg NMS is about to launch the cycle all over again. That should be good news for Microsoft because all of the latest ECNs were built on Microsoft technology.

Algorithms, said Wald, are tools for traders, not replacements for traders.

“The best algorithms do what computers do well and they don’t attempt to do what people do well. They can look at massive amounts of data and respond in milliseconds…Traders should be offered toolkits of algorithms to meet certain trading needs, and they should get an education in how algorithms go about their business. But algorithms cannot exercise judgment or respond to a scenario that hasn’t been defined. Traders and algorithms have complementary skills.”

The growth of hedge funds using high-speed algorithmic trading has led to a sharp increase in the sales of order management systems (See WFS Spring ’06) and in execution management systems, like EdgeTrade.

Looking to meet the demands of the hedge fund business which is growing fast even while margins are shrinking, at least two major investment banks expect to build prime broker platforms using as much Microsoft technology as they can. Kenny McBride, capital markets industry manager for Microsoft, expects that by the SIA Technology Conference in New York in June, at least one prime broker will announce a hedge fund platform built on Microsoft.

Also at SIA, Microsoft and Sapient will have more detailed information on how the new release of Microsoft’s Excel spreadsheet, Excel 12, will provide important new capabilities for capital markets. One key change – Excel will be available on a server so the spreadsheet or models built in a spreadsheet can be made available to the firm and to customers while remaining under the control of its originator.

A senior executive at a London-based bank who had been briefed on Excel saw immediate advantages to running it from a server.

“Our traders may develop a model and offer it to clients to conduct a trade. The more you have on one system the easier it is,” he said.

The most widely used tool in capital markets, Excel in its new release will offer more power, more columns, and far more control for the financial enterprise.

But participants disagreed on where algorithms work best, and where they don’t work well at all. Some use them for small and mid cap stocks, others like algorithms when markets are highly liquid but resort to traders to finish off a trade, while some said algorithms can be tuned to work even in times of low liquidity.

“There is no set-it-and-forget-it,” said Doug Coté, senior portfolio manager at ING Investments.

At Susquehanna International Group, traders and sales staff know their clients’ sensitivity, so when they run into difficult trading they know which clients want a call before traders should pursue an outlier, said David Margulies, director of program trading at the group.

“We will call the customer and change strategies to meet their goal or stop all together.” If certain names don’t have enough historical data for the algorithm they will trade those issues manually.

Chetan Joglekar, senior trader at TIAA-CREF, said the pension fund built its own algorithms.

“When we started, they were terrible, but we tuned them over time.” The costs of trading go up for the last 50 names in a 500-name basket, he said. “We rarely see a trader put a stock on an algorithm from the start of the day to the end. As you go lower in the capitalization scale, the algorithms tend to behave erratically because the market conditions are not the same as the historical. Most active traders use algorithms as a tool and they have different strategies at different points in time.”

Greg Komansky, vice president at Legg Mason Capital Management, said he won’t use an algorithm he doesn’t understand, and he won’t use an algorithm from a sell-side firm if the firm won’t explain how it works.

“I want to see most of the steps. Firms are becoming a little more forthcoming because then know they will get the lion’s share of the business. If they think we are going to trade against them, they shouldn’t enter into a partnership with us.”