There is a new contender in the data visualization business: Big Blue is joining Visible Decisions, NEOVISION and Advanced Visual Systems in the quickly developing market for visually-oriented software. IBM's Diamond API (Application Program Interface), is a cross-platform set of function calls that allows programmers to integrate interactive visualization into OS/2, UNIX, Windows 95 and NT applications.

According to Bernice Rogowitz, who manages the project at IBM Research, Diamond is particularly appropriate for risk managers because it allows people to dynamically explore interactions between many variables. The picture above is a "screen shot" of a sample application in which the user can explore how risk factors interact. In this case, the user has selected the period of time in May 1995 when the price of the yen went through the roof. The graph in the middle of the screen shows an animated view of the volatility of the CAC-40 as a function of the volatility of the FTSE. With a press of a button, the user can see how the relationship between these two volatilities evolved during the yen crisis.

A stand-alone version of the underlying visualization software used in this approach is available commercially from SPSS Inc. at http://www.spss.com. The API version is available from IBM. For more information, contact Keith Bear, keith_bear@uk.ibm.com.

The end of 1996 was marked by a flurry of initiatives designed to reduce the high cost of PC ownership for corporations and financial institutions alike. On a per-user basis, the annual cost of PC ownership is estimated at $5,000 to $12,000. Much of these costs stem from the necessity of constantly upgrading PC-based software and managing the distribution of new soft-ware to numerous, decentralized user groups. Technology gurus see the intranet as a likely solution: Basically, most enterprise-wide software would reside on powerful servers, and users could access this software through low-maintenace terminals equipped with Web browsers.

Sun Microsystems is now shipping two products that should help this trend along: the Netra j server and the Java Station. The former is a scaleable, Java-based server and the latter is a bare-bones terminal, with up-front costs of about $700 and estimated annual maintenance around $2,000. IBM is also getting into the game; the firm will soon release its "Network station" which, like the Java station, will be able to run Java applications and connect to a wide variety of servers.

In anticipation of a movement toward this "thin client/fat server" configuration, Oracle is offering word processing and spreadsheet packages designed for the Java environment as well as a network computing architecture, a toolkit which will allow users to blend diverse hardware environments-including UNIX servers, NT-based machines, PCs and network computers.

Microsoft, which has a vested interest in the continued presence of the PC on desktops everywhere, also has a plan to make PCs-as-usual easier to administrate, with a new version of NT designed for the "zero administration PC," which will make upgrades easier and less costly.

Despite initial skepticism, it appears that Integral Development's prime directive-the access of pricing and risk management data through a coroporate intranet-is gaining acceptance. According to some industry experts, CIBC Wood Gundy will use Intgeral's intranet connectivitity to provide widespread access to risk reporting via the net.

More companies than you'd think are quietly using neural networks and genetic algorithms to help make tough hedging decisions

What if you could supplement your staff of market analysts with powerful, "virtual analysts" capable of predicting markets with a 70 or even 80 percent accuracy? Could virtual analysts help you make risk management decisions as well as trading decisions? According to Steve Ward, president and technical director of the Maryland-based Ward Systems Group, the answer is yes. "A number of large companies are using neural nets and genetic algorithms to help make internal investment decisions," he says. "Others are using this same methodology to track commodity markets, including oil, soybeans and so on, and to use their predictions to make better hedging decisions."

The goal of corporate hedging, some would argue, is simply to filter out the impact of market fluctuations-both positive and negative-on earnings. However, in reality, many companies only hedge a portion of their exposures. This portion of hedged transactions may fluctuate, based on market predictions. Now, these predictions may come from a neural network or a genetic algorithm.

Neural networks are learning programs that, when trained on relevant data for a certain period of time, can yield market predictions or equations which, in turn, model market movements. Likewise, genetic algorithms-a technique borrowed from biology which has lately enjoyed a period of vogue in the financial markets-mate "mother" and "father" predictive equations to produce offspring. Only those offspring whose solutions come closest to a designated benchmark are allowed to survive into the next generation.

Not many companies that use neural networks are willing to talk about it. To a certain extent this is because of the hype that the technique initially received as a trading panacea. Many embraced neural networks as the next and grteatest get-rich-quick scheme-and were disappointed. The problem, explains Ward, is that many users do not feed their programs the appropriate data. "You cannot just give your program high, low and closing data and expect it to provide accurate predictions," she says. Instead, it is imperative to experiment with a wide variety of market indicators to provide your neural network with the appropriate "food for thought." Thus, neural networks, like most other tools, work best in the hands of experienced finance professionals.

At least one large oil and gas trading company has been quietly using neural networks and genetic algorithms in its pursuit of the perfect hedge. The technical data produced by both the neural net and the genetic algorithm is considered particularly important by managers in cases when they are likely to change their hedge ratio. "When the neural network and the genetic algorithm agree on a particular market direction, we might consider adding or subtracting hedges," reveals a company analyst.

For more information see:
http://www.profittaker.com
http://www.promland.com
http://www. jurikres.com

Java, the hyped free Internet software from Sun Microsystems, is at last finding some real-world economic applications among the Chicago exchanges. Heretofore it has been something of a toy for nerds. The Chicago Board Options Exchange (CBOE) has a team working on enhancements to its educational program that will have interactive worksheets for different "what if" scenarios. Meanwhile the Chicago Board of Trade's (CBOT) Project A plans to allow traders access to the system through their PCs with a dial-up service. The ability to access this electronic after-hours system from homes, or on the road via laptops, might silence criticism from CBOT traders that Project A is difficult to use, has poor functionality and high costs. "We're moving toward a Java interface, so that we'll be available on PCs and Macs in 1997," says a spokeswoman for CBOT.

The CBOE hopes that the interactivity that Java provides will expand its aggressive outreach program for new options investors. "This will allow an investor to chose different strategies based on his portfolio and view of the market," says CBOE assistant vice president of investor services Carolyn Mitchell. By springtime, visitors to www.cboe.com will be able to enter such data as stock price, strike price, expirations and so on, and the system will compute the consequences. "You'll be able to graph out profit and loss and the break-even points of different inputs," Mitchell says, which will be a great leap forward from the present arrays of static "brochure ware" at this site. Further down the road the CBOE will unveil an on-line classroom with many more Java applications. "Our goal is to create an interactive educational area where people can come and learn about options instead of going to a seminar. We hope to attract a new target group of investors," says Mitchell.

Connecticut-based Greenwich Associates, a research and consulting firm, recently completed a survey on the use of derivatives by corporates and financial institutions that has yielded a surprising statistic: 35 percent of derivatives users surveyed do not use risk management software. The report also shows that, while average transaction volume per user jumped to $2 million from $1.5 million during the past year, one of every four respondents who used derivatives in 1995 did not use derivatives in 1996. Many attribute both the reluctance to invest in software and the decline in the number of derivatives users to uncertainty surrounding the FASB's latest proposal on derivatives accounting. This means that risk management software, for better or worse, also remains on the back burner at some institutions.

On February 4 at the New York City Marriott Marquis, the Securities Industry Associaton brought together the usual industry suspects to consider the future role of the Internet in securities dealing and processing. Speakers at the conference emphasized the importance of complying with regulations in order to combat burgeoning net-based securities fraud. For more information, see http://www.sia.com.

The Toronto-based risk management specialist has recently launched RiskBrowser, a Web-based risk information management system, which facilitates the access of risk-related information and analytics on a company-wide level through standard Web browsers, such as Netscape. The RiskBrowser will be offered in conjunction with an updated version of RiskWatch, Algorithmics' signature product.

Risk management may not be the sexiest of topics, but it is getting more than a fair spotlight on the Net recently. Here is a quick tour of sites that have budded in the last few months:

http://www.netcom.com/~bschact/varbiblio.html

Barry Schachter, who works for the Comptroller of the Currency, has assembled a massive VAR bibliography, with hot links to reprints of some 40 magazine articles. Sources include Financial Analysts Journal, Wall Street & Technology, Risk, Derivatives Strategy, and Treasury Management International. Recent books are also featured, as are drafts of about 30 working papers.

http://www.contingencyanalysis.com

Though consultant Glyn Holton only launched this site last November, it seems to have a strong following in its discussion groups. There are three different subject areas: general, theory, risk selection. In the theory segment we came across discussions of VARdelta, intraday limits, volatility of volatility, jump diffusions and corellations, principal component analysis. Heavy stuff! There is also a glossary of 200 risk management terms, which have hotwires to in-depth articles and explanations. The site also offers its very own search engine for the Net.