Trading entire portfolios with algorithms is not high on the buyside’s to-do list, but that isn’t stopping brokers and vendors from developing technology to make it possible. Agency brokerage Instinet and technology firm FlexTrade Systems recently joined a growing list of broker-dealers deploying algorithms to analyze and trade hundreds of stocks at a time. “More buyside firms have taken active control of their portfolios, and more trading decisions are moving in-house,” says Vijay Kedia, president of FlexTrade. “A buyside firm now doesn’t have to expose an entire portfolio to a sellside firm to leverage a portfolio-level algorithm.”
Almost 80 percent of buyside firms use algos, but only 13 percent of those firms use them to trade large baskets of stocks, according to the TABB Group. Most algorithmic trading is of the single-stock variety.
However, the research firm points out that portfolio trading algos are becoming more advanced. A recent report notes that algos from several providers now incorporate “correlations among the names in the portfolio to formulate an execution strategy that helps traders make intelligent trade-offs about managing exposure and reducing market impact.”
Portfolio Strategies
The new portfolio algos from Instinet and FlexTrade join those from ITG, JPMorgan, Goldman Sachs, Susquehanna, Miletus Trading and a few other firms. Instinet’s Wizard PRO (Portfolio Risk Optimizer) and FlexTrade’s Flex PTS (Portfolio Trading System) are geared toward large portfolios of several hundred to a couple thousand names.
Both Instinet’s and FlexTrade’s algos use implementation shortfall as the trading benchmark. Both are easy-to-use, “plug and play” models that allow customers to adjust parameters such as their level of urgency and short-term alpha expectations for particular stocks.
Both algos also rely on something their competitors don’t have-the smarts of Northfield Information Services. Northfield, a respected technology firm that creates short-term risk-based models, developed the 20-factor model that forms the backbone of Instinet’s Wizard PRO algorithm. Flex PTS also incorporates Northfield’s risk model into its algorithm.
Instinet worked with Northfield for the last two years to develop Wizard PRO. Instinet uses Northfield’s risk model and optimizer, and will soon also include the technology firm’s market-impact model. Flex PTS uses Northfield’s risk model. FlexTrade uses another firm’s optimizer and can use its own or a customer’s market-impact model.
The Northfield U.S. Short-Term Equity Risk Model analyzes portfolios and determines the risk/return characteristics of stocks and the correlations between them, including stocks in different sectors. This enables the model to identify natural hedges for illiquid and hard-to-trade stocks within a portfolio to reduce various risks as the portfolio is traded. The optimizer determines the most efficient trading schedule and the pace of trading for the portfolio’s stocks, based on the risk model and market impact model.
Trading Schedule
Instinet’s and FlexTrade’s algos execute orders according to their optimized trading schedule. In both cases, the schedule can be re-optimized during the day if, for example, a trader executes a block of stock on a crossing system or if market conditions change.
So far the largest portfolio ushered through Wizard PRO was $1 billion, says Thorsten Schmidt, first vice president for algorithmic trading at Instinet. Instinet beta-tested Wizard PRO internally and with several clients over the last several months.
“Wizard PRO adds the most value for portfolios that are not quite cash-balanced and portfolios with a fair amount of nonsystematic risk,” Schmidt says, referring to the presence of stocks whose movements are not correlated with the broad market. An example, he says, could be an institution buying $150 million in Russell 2000 names and selling $100 million in mid-caps.
Schmidt adds that Wizard PRO is ideal for lists of 50 to 150 stocks valued at $200 million to $1 billion, but can also be deployed for larger baskets. The cost for using the algo is 3 cents per share and works its way down, based on the client.
As for FlexTrade, Flex PTS isn’t the firm’s debut in the portfolio-trading arena. FlexTrade had earlier portfolio algorithms that kept a portfolio dollar-balanced and/or cash-balanced across sectors, but it did not use an external risk model and did not incorporate the correlations between names into its analysis and trading schedule.
FlexTrade’s Kedia says an advantage of Flex PTS is that buyside users can trade portfolios through any broker or combination of brokers on the FlexTrade platform, instead of through only one broker. He expects Flex PTS to be used by buyside firms as well sellside firms that don’t have in-house portfolio trading algorithms.
FlexTrade clients must license Northfield’s risk model, either through the firm itself or through FlexTrade. Clients can also use an internal risk model or a model from another risk technology vendor instead of the Northfield offering, Kedia says.
