PaineWebber Plans Ihas Pilot: Intelligent Hedging Assistant

PaineWebber is developing an intelligent hedging assistant known as IHAS. Currently in pilot mode, IHAS is intended to assist block desk traders in pricing and developing risk management strategies for large blocks of stock, sources say.

The IHAS pilot manages several analytical tools in combination with a proprietary knowledge base to tell traders what's going on in operational terms in real time, without requiring them to become mathematicians.

Large inventories of stock are accumulated by block traders in the course of market-making activities, and unless these positions are adequately hedged, profits can be wiped out by small price changes.

Without realistic estimates of relative beta or hedge ratios, block traders can easily under-hedge or over-hedge. Unfortunately, these errors don't cancel each other out and the result is either over- spending or over-exposure.

When institutional clients contact a block desk to solicit bids or offers, traders can't afford to dither. IHAS is configured to provide the information needed to price a block and select an appropriate hedge for the position in a matter of seconds.

IHAS is characteristic of a new wave of expert systems development for trading applications. Instead of attempting to swallow the world whole and spit out the pits, more recent projects (TST #6) bite off manageable subsets of the trader's job and offer solutions in a supporting role.

To access IHAS, a trader simply hits a function key and enters a ticker symbol. The pricing function instantly delivers the following security-specific information to the user's windowed screen:

fundamental data

dividend due date

projected earnings

price history, volume history

current and historic liquidity ratio

beta and volatility measures

current price

regression coefficient of the stock against the market

price performance versus industry group

correlation coefficients for a number of synthetic indices

IHAS doesn't deliver raw data fresh from the price box. Information presented by the pricing function has been cleaned, pressed, and packaged. Pricing information and all correlates residing on the database are updated every five minutes.

Having settled on a price, traders hit one of two function keys indicating sale or purchase, and input the size and the price of the transaction.

IHAS then presents several specific suggestions for both short and long hedges. Each solution is a 'perfect' hedge insofar as it covers 100% of the position. The trader chooses what proportion of the hedge he wishes to use.

Each trader will have a proprietary rule base that helps determine which hedges IHAS suggests and how they are ranked. Hedging suggestions are ranked by 'goodness' as well as by trader preference.

Traders input preference profiles to keep IHAS up to date on individual prejudices for correlated stocks and options. Several different criteria can be applied as hedge ranking attributes including price, volatility, and time value.

Any number of exceptions to these preferences may be included. A trader might say, for example, 'I'll only hedge IBM up to 50,000 shares with options.'

IHAS consists of five integrated components:

A market data feed supplies tick-by-tick price data for equities and options,

An online database offers rapid access to historical data,

An expert system for generating and evaluating hedging strategies,

A high-speed numerical computing processor for applying quantitative analytic tools, and

A color workstation for presentation of hedging advice and interaction with the trader.

All components of the IHAS system are to be connected by Ethernet. The gateway server is a Compaq 386, while both the database and the expert system servers are Sun 3/280s. Trader workstations will also be provided by Sun. Network communications are via Sun's Network File System (NFS) and its Remote Procedure Call (RPC) protocol.

Lack of connectivity was one of the big reasons many early expert systems projects on Wall St. didn't work out. PaineWebber has recognized the need for one coherent network environment from the trader's screen through the entire system.

Price histories and various correlations sit on the database server and are managed by Sybase's relational database management system. All Sybase transactions are handled via DB Library, part of the RDBMS' toolset.

The Compaq 386 price server handles the data feed from CMQ Communications and updates the five-minute, hourly and end-of-day tables in the database.

The expert system sits in the middle and talks to both the database and the price server. From the database, IHAS extracts expert systems facts, fundamental and technical data, and historical security prices. Via remote procedure calls (RPC) and attendant external data representations (XDR) IHAS retrieves current prices from the price box.

Once correlations or hedging strategies are prepared, a packet of data goes out through RPCs to the workstation which picks it up, displays it, and interacts with the user.