The explosion of Internet is partially due to the fact that information can be inexpensively offered on the Web, as well as conveniently retrieved. The accessibility creates an enormous amount of raw data. In the future, successful information providers on the Web will be distingushed by bundling raw data with tools for data refining, data analysis and presentation of complex multi-dimensional data sets. This model of information delivery on the Web is derived from the object oriented model of computation, where data and methods which operate upon the data are bundled together.
It seems natural that CME would eventually offer a one-stop full-service Investment Analysis Resource Center. One of the most valuable components of the service will be sophisticated quantitative analysis capability. At present, retail investor is forced to use primitive techniques for quantitative analysis, due to prohibitively expensive software and hardware. In addition to cost, investors must learn how to install, use and maintain proprietary software and frequently must purchase different software for different investment tasks.
The idea behind the Investment Analysis Resource Center is to enable the most sophisticated analysis of investment data to a retail customer at the single initial expense of purchasing a Web browser. Through a carefully designed Web front end, an investor will be able to specify the kind of information he/she needs, the type of analysis requested and how the results of the analysis should be presented. A combination of existing Web technologies such as HTML forms and Java or VRML-based interfaces could enable a very high-level description of the analysis request. For example, using HTML forms alone, one could request "List 3 option contracts with biggest discrepancy between historical volatility and the implied volatility matrix among all Feb 96 through May 96 pork belly puts and sort in increasing implied volatility order".
A request is shipped from the investor's Web browser to the CME IARC Web server. The Web front-end is linked to an array of powerful back-end servers. Typically, IARC server would request services of historical database servers to obtain historical data or market servers for current pricing information. These data would be forwarded to a compute server for numerically intensive tasks like estimation of stochastic processes, option valuation, time series analysis, forecasting or application of technical analysis methods. The result of this Simulation-on-Demand engine would be forwarded to a visualization server, which would format the results according to investor's specification, prepare the results to be displayed on the final Web page and ship them to the IARC server, which would format and forward the Web page to the client's Web browser. The overall schema is similar to the one described for the Investor Education Center, but performance requirements of this system are orders of magnitude higher.
Back-end servers must be able to process a large number of concurrent users with computational tasks of varying degree of complexity. The keywords here are performance, scalability and fault tolerance, which implies that both the Simulation-on-Demand engine and the database servers would be most likely parallel processors. From business perspective, the size of the initial investment in a system of this functionality will be a barrier for all but the biggest competitors.
The collection of database servers must offer a data warehouse-like functionality to the Web front-end. The important performance issue will be processing of complex ad hoc queries instead of high transaction volume, so parallel query mechanism is important. Task parallelism will be the dominant computational model on the Simulation-on-Demand engine, which implies that both tightly coupled processors and distributed platforms may be appropriate. Sophisticated resource allocation and scheduling mechanisms must be put in place, since both the number of users and the complexity of tasks are likely to be very volatile.
The most important aspect of this system is that it radically changes the way investment decisions are made in the retail segment. Retail investors would finally be able to purchase the most sophisticated analysis available on the market without an upfront purchase of an array of parallel compute servers with installed million dollar simulation modules. Investors would have to pay-as-they-go only for resources they consume. If they decide to try trading in another instrument, they do not need new software. Convenience, cost structure and intrinsic value would turn this site into one of the busiest on the Web within days. A rudimentary prototype of this kind of service is offered on the Web by Olsen and Associates . They offer forex trading recommendations generated by their proprietary technical trading system based on intrinsic time volatilty analysis.