A new feature Bloomberg Tradebook has added to its algo-assisted trading software for U.S. securities lets traders use the algo the way airplane pilots use auto pilot mode — when they wish, traders can let the B-Smart algorithm decide when to aggressively seek liquidity and when to wait for the market to come to an order price. When traders become uncomfortable with letting software make such decisions for them, they can turn off automatic pilot and resume the controls.
Bloomberg Tradebook's B-smart algorithm was created two years ago "to solve the problem of where do you quote in a fragmented marketplace," says Gary Stone, director of development, trading research and strategy. "Our clients were telling us that in the fragmented market structure, you could have a bid in Nasdaq and the bid would be getting hit in Arca. The market, even though it's connected, is still very transitory. We came up with a quantitative model for smart order quoting that looks at the world and says, where's the most likely place my order should be in order to be in the path of the person who wants to sell." The model can look across traditional exchanges, ATSs, ECNs or dark pools. Typical clients include hedge funds, institutional investors and brokerages.
The algorithm helps traders place orders at various levels of aggressiveness. "There are some times when you're going to be passive and some times when you're going to be reaching to get liquidity," Stone says. B-Smart has four modes: aggressive (grabbing liquidity), normal (passively quoting across market centers), passive (sitting at one price level until someone hits that price) and reactive (participating in active markets).
The new feature, Auto, lets traders let the B-smart algorithm decide which of the four modes it should be in. "As the markets have evolved, clients have become busy to the point where they can't drive the aggression levels or they may be faced with trading a name they're not familiar with and may not know how to drive it," says Sabrina Rovelli, manager, U.S. Equity Product and creator of B-smart. "Where the previous version was for the client that liked to micromanage and was a high-touch trader, the second version is a low-touch strategy; B-smart understands intraday where liquidity has been and where it continues to be, predicts what the stock is going to do dynamically, and based on that prediction automatically decides how aggressive to be." As in its regular price-prediction activity, B-smart takes into account historical data and intra-day data, bid/offer spreads (all from Bloomberg data feeds), and order size.