Storm Copestand
Traders Magazine Online News

Conquering Fear in Trading

In this exclusive to Traders Magazine, therapist Storm Copestand examines how traders can manage expectations and conquer their fear during the entire execution process.

Traders Poll

Amid changes in builder, do you think the CAT project will be completed by 2020?

Free Site Registration

September 30, 2004

Building the Better Mousetrap?

By Peter Chapman

Upstairs, traders use knowledge of their clients' orders to make markets. Downstairs, brokers use knowledge of other brokers' orders to benefit their own clients. A Big Board floor broker, for example, will often impart that information to his clients in the form of floor looks.

"If you could centralize the interest of all these algorithms," Keith says, "you would have the largest source of liquidity in the known world. An order of magnitude greater than the New York Stock Exchange."

PDQ is pitching to both the sellside and buyside. Likely users are of two types, according to PDQ: quantitative traders such as ATD, Susquehanna and TradeBot Systems, and brokers who manage orders for institutions using algorithms. Most of the larger investment banks now do this.

To make the concept palatable to these sophisticated shops, the inventors had to overcome two hurdles. First, the system had to preserve the anonymity of the algorithms and their masters. Second, the system had to eliminate the risk of information leakage.

To prevent leakage, PDQ creates a replica of the client algorithm. It works strictly with this daughter' algorithm and not with the original which resides on the customer's server. The daughter algorithm represents the interest of the mother algorithm.

The daughter algorithm imparts no information to the mother algorithm unless there is an execution. In addition, the mother algorithm is never made aware of the identity of a potential contra order unless an execution occurs.

The mother algorithm doesn't even know an order has been created and sent to a market center unless an execution is achieved. "The daughter algorithm can't tell mama," Keith explains. "That is a fundamental part of our proposition. There is absolutely no risk."

As for anonymity, when PDQ approaches an algorithm in representation of its customer algorithm, it does not reveal its customer's identity. In addition, if the algorithm does not have the other side of the trade, it doesn't even know that it was asked.

"If you can represent the interests of the algorithmic trading desk with a central platform designed for information containment," Shapiro says, "then you can create tremendous efficiencies in matching the interests."

PDQ is strictly an order router. It does not arbitrate between orders. It does not match two orders. It does not guarantee that its client's order will match once it reaches the destination. The arbitration process is done by the market center. So, the client algorithm order may still have to compete with other orders for the right to match once it reaches its destination.

Although limited to order routing, PDQ is not passive. If it finds a contra order for its client, but determines that the client needs to adjust its pricing strategy, for example, it will tell the client so.

"If you want to get a piece of a particular [contra] order," explains Shapiro, "we'll tell you what you need to do. For example, you're going to need to price at the midpoint." PDQ will then generate an order reflecting the circumstances.