Commentary

Jared Dillian
Traders Magazine Online News

Was it Worth It?

In this piece from 10th Man, author Jared Dillian discusses how the ETF revolution is less about ETFs and more about indexing; about how people have come to view stocks less as stocks and more as blobs of stocks.

Traders Poll

Would you feel better if the Chicago Stock Exchange were purchased by U.S. firm or consortium rather than a foreign one?

Yes

73%

No

4%

Doesn't matter to me

23%

Free Site Registration

November 8, 2005

Can Algorithms Be Tailored to Portfolio Needs?

By Gregory Bresiger

Also in this article

  • Can Algorithms Be Tailored to Portfolio Needs?

Traders are increasingly worried about market volatility. Move a basket of stocks quickly and market impact can run up costs. Trade the basket slowly and opportunity costs can exact a hefty price.

Yet algorithms, designed to mitigate these market volatility issues, are often missing the mark because many were not designed for trading large numbers of issues. Indeed, many of these mathematical marvels will miss the VWAP benchmark by half a penny or so, says a firm selling its own brand of algorithmic solutions. Susquehanna International Group (SIG) claims it has a better idea. The firm has released a new portfolio algorithm, Tempo (Targeted

Traders are increasingly worried about market volatility. Move a basket of stocks quickly and market impact can run up costs. Trade the basket slowly and opportunity costs can exact a hefty price.

Yet algorithms, designed to mitigate these market volatility issues, are often missing the mark because many were not designed for trading large numbers of issues. Indeed, many of these mathematical marvels will miss the VWAP benchmark by half a penny or so, says a firm selling its own brand of algorithmic solutions. Susquehanna International Group (SIG) claims it has a better idea. The firm has released a new portfolio algorithm, Tempo (Targeted

Execution Model for Performance Optimization), that is especially designed for portfolios. Tempo, which has a patent pending, is designed to solve the problems of the cost conscious trader. This typical trader commonly must move entire portfolios instead of individual stocks.

"Right now there a lot of algo offerings for single stocks," said David Margulies, head of algorithm sales at SIG.

"But what we found is that people are more interested in executing on a portfolio level." But the average buyside trader-who in an era of microscopic spreads and single-digit market returns must find more ways to cut costs-sometimes finds that depending on electronic trading and benchmarks doesn't solve the persistent cost problem.

VWAP Failures

Use an all-day VWAP to trade large-cap stocks and the trader gets one kind of execution. Trade small-cap stocks and get another, neither of which is satisfactory, say SIG officials. And there's another aspect to the cost-efficient trading puzzle to be considered: almost all equity stocks-no matter how seemingly different they are-have some kind of correlation.

Beyond three or more stocks, it can be difficult to measure the correlation because there are more factors that affect the mix such as sector, industry and the size of a stock. That mix, SIG officials say, can have unexpected results when two or three names are traded at the same time.

Put these different kinds of stocks into one portfolio and, again, large trading costs can result. There must be some way to link these trades in a way that will lower the predictably high costs of different stocks.

"Whether you're using a TWAP or a VWAP," notes Kai Huang, the SIG executive in charge of program trading, "it often takes too long to do a basket of trades and this is exposing you to big market impact risks," So Tempo is designed to use a big picture approach.