By Kevin Merz, Managing Director and CEO of Enterprise Technology Corporation, a New York-based technology management consulting firm.
One of the issues confronting many buy-side institutions, including the larger hedge funds, is the cost of maintaining a professional trading desk
One of the issues confronting many buy-side institutions, including the larger hedge funds, is the cost of maintaining a professional trading desk. In fact, in a recent survey our firm did among buy-side institutions, trader productivity was listed as their number one concern. Although most will not argue the value of having seasoned traders on staff, nobody has an appetite for paying trader's salaries for clerical work. At the same time, we all recognize the value of capturing trade data at the source (i.e. the trader) and sending it immediately through a process, untouched further by human hands, that currently goes by the label 'straight-thru processing (STP)'.
Nothing focuses the minds in a (progressive) firm so much as when the firm, with a certain number of traders, is faced with the prospect of adding one more trader (salary plus the attendant costs and organizational friction), or thinking of a better way. This article discusses the route one prominent hedge fund firm has taken to 'work smarter, not harder'.
The goal of our project was to increase trader productivity by at least 15% per year, while at the same time reducing trading errors from 5% of transactions to 0.25% of transactions. We also wanted to provide real-time execution data and firm-wide P&L information to the portfolio managers and implement real-time point-of-order and execution compliance.
There are several different ways a buy-side firm can approach its trading activity. Many firms use more than one and our hedge fund uses the first three of those outlined below:
1. Portfolio manager makes decisions, and also trades and allocates. Very common with fixed-income and derivative products.
2. Portfolio manager gives orders to trader. Trader trades, and is responsible for input into system.
3. Portfolio manager, trader, and trader/portfolio assistant shares responsibilities.
4. Computer-generated model trading with or without traders in the organization.
5. Program trading and proprietary trading.
Our hedge fund has 6 front office staff (portfolio managers, etc.), 4 traders, and 8 back-office people. There are generally around 400 different securities in the portfolios at any time, with 50% or greater turnover per month. Trading volume varies from 150-1000 trade tickets per day, with about 350 brokers and counterparties. The only major instrument not traded is mortgage-backed securities. The firm does end-of-day trade allocations based on average price. During the day, the traders maintain order and execution information on handwritten blotters, and they key it into the firm's on-line trade order management system when time permits. One can easily see that a key component of optimizing trader productivity is the opportunity to increase the value of telephone calls by reducing or eliminating untimely or unnecessary information and interruptions. We also must deal with the age-old "real-estate" conundrum: screen space for market data display and monitoring must be maximized.
To achieve our objectives we have adopted an approach encompassing the following major elements:
- FIX messaging to receive automated indications and execution data (FIX Protocol is a language which defines specific kinds of electronic messages for communicating securities transactions between two parties. It is an open standard for communicating financial information in a secure but non proprietary way).
- Customizable visual and audio event notifications through sounds and/or icons or pop-up boxes on the user's screen which notify the trader that there is an important message for him/her without unnecessarily disrupting him.
- Elimination of manual trade blotter
- Automated calculation of average prices
- Electronic end-of-day recap with counterparties
- Straight through processing for all security types including foreign exchange
- Pre- and post-trade compliance checking to see that trades do not violate client, firm or prospectus guidelines (a) when the order is entered and (b) at the end of the day, respectively.
- Real time net exposure calculations
To make this all work, we needed to realize and accept several immutable facts about the people we are working with. Traders usually don't have two hands to free for typing, andduring peak workloads, data entry into the system just isn't timely. In other words, at the times that we need the most timely information available in the organization, we have the least. Another interesting observation about traders is that they write faster than they type, and they talk faster than they write.
In fashioning technology-based solutions to business problems, one must fit the technology to one's business, not the business to the technology. Then, every major technology expenditure should be considered as an investment: if it doesn't make or save money, why do it? The payback should be less than 50% of the time to obsolescence (often, that means less than two years).
Thus far, we have developed two pieces of technology to implement our strategy. The first is a flat-screen tablet to replace the hand-written blotter. The tablet recognizes the trader's handwriting, and communicates the written order and execution data directly into the trade order management system. In a few months, the tablet will automatically display executions received from brokers via FIX and other system interfaces. Traders are able to view and modify other traders' blotters as necessary, to cover vacations, rest stops, etc. The tablet is portable, to facilitate visits to portfolio managers' offices while allocating trades. We also did a working prototype of a voice recognition system. Then we realized that writing is actually better, because the trader cannot be orally telling his computer what he just discussed with the last broker, while simultaneously talking to the next broker. He can, however, be writing broker #1's information while talking to broker #2.
The second technology project was the development of an application that addresses the need to communicate trade information to portfolio managers without disturbing conversations. This implies a foolproof method to verify that the messages are received. Important messages must remain on the recipient's screen until he/she responds, and this tool had to be extremely easy to use, and very fast both in terms of system response and user time required to generate a message. A permanent log of messages and responses must be maintained.
Although this might sound like a good use for E-mail, it is not. Email products are too slow, do not have the 'burn-through' feature, and they require too much effort to make a response. E-mail products also could not work on all of the firm's platforms (Novell 4.11, Windows NT, and Win95).
The resulting messaging system is quick (average time to send message is 1 second), reliable (2 levels of verification that give sender and receiver status of message), easy to use (click on receiver, write message, click on send; to respond, click on button or send message).
The results thus far are already impressive. The number of missed trades is down 60 percent. The average size of trades has increased by 20 percent, due to hitting larger blocks more consistently, and the number of buys on the bid and sells at the offer are up 20 percent. The number of trading room / portfolio manager communication errors is down over 50 percent; the number of trading room errors, down 25 percent. And the firm hasn't had to hire that fifth trader.