Deutsche Bank Algos Get More Nimble

The best way for a bulge broker to corral more electronic trading business is to be able to modify its algorithms “on the fly” or quick. Otherwise, the buyside will take its orders and business elsewhere.

That’s the philosophy of Jay Fraser, head of Autobahn equity execution and Autobahn equity sales in the Americas at Deutsche Bank.  The exec recently sat down exclusively with Traders Magazine and talked about how the bank is responding to institutional demand for fewer and more responsive algorithms.

“From a customer perspective, they’re telling us that they have too many algos and there is too much customization in those algos,” Fraser said. “There’s been a hyper-tailoring in algorithms for the buyside and now they want a return to fewer and simpler algorithm that can be tailored, if need be.”

This desire to return to fewer algorithmic offerings has been confirmed by Traders Magazine in its conversations with the buyside. One buysider recently said that of the 30 algorithms his brokers offer him, he uses only a handful.

Fraser added that while still offering its clients several types of algorithms under its umbrella Stealth offering, the focus has shifted from creating wholly new algos to constantly updating and tweaking its core offerings. Updating its algos, “on the fly” or as either the market structure changes or a buyside trader’s trading style changes, is what the bank is looking to achieve. 

“Algo tweaking benefits the customer base as a whole rather than working on individual customizations that don’t cross pollinate the platform as a whole,” Fraser said. “By the time the customer realizes their customization no longer fits the change in spreads, volume patterns or volatility he may have given up tens of basis points in performance. Our focus on the performance of Stealth insists we stay disciplined day in and day out to keep it fresh. We follow the performance of individual customer configurations with increased intensity.”

For example, Deutsche Bank’s algos have picked up the tilt of ‘to the close’ volume early. A single customer customization may lag a market shift by an entire month or even a quarter, he added.

Stealth uses high-frequency alpha models to source liquidity in both bright and dark venues. The algorithm adapts to intraday market conditions and attempts to reduce the ability of HFTs to step ahead of large orders.

The desk also offers clients an algorithm designed to work specifically at the close, dubbed “TargetClose.” It is designed to project each day’s closing volume based on real-time and historical market data. It continuously adapts to market conditions in order to minimize the trade’s deviation from the closing price on the date of the trade.