David W. Graves, an independent consultant specializing in electronic trading systems, explains how to evaluate new trading ventures.

In 1984, when I worked for Intex, the Bermuda gold exchange, it was the only electronic exchange in the world. Now, with new trading ventures being announced almost weekly, there are more than 70 electronic exchanges.

What's the best way to evaluate them? I believe there are eight variables that can help us assess the services a trading system provides to its traders.

Cost. Total cost is the net amount a trader expends to trade on a system. Some traders prefer to divide total cost by the number of tickets traded to get a per-ticket cost. Either way, lower cost is better. Electronic systems tend to be less expensive to operate than open-outcry markets. Continual investment in system upgrades, however, can make electronic trading systems expensive.

Liquidity. The most important attribute of any market is liquidity. While there is no agreed-upon method to measure liquidity, I measure it as the time required to liquidate a position that requires trading against the weaker side of the market. Consequently, liquidity can be described as the number of contracts per unit of time that the weaker side of the market can absorb. If most traders are short, it may take a long time in an illiquid market to find sufficient buyers to liquidate a position.

Price Discovery. Another key component of a trading system is price discovery, the interplay between bids and offers to determine the fair value of an item. It's important to determine if a particular market is the primary venue for price discovery or if it is a parasitic system. The New York Stock Exchange, for example, is the primary price discovery venue for large-cap stocks. As an indication of the value of its price discovery function, it receives in excess of $100 million annually for the market data feed of these prices. Crossing Network, a subsidiary of Reuters, by contrast, permits blind crossing of trades after the NYSE close at minimal transaction cost, using the NYSE closing price. Crossing Network is thus a parasitic price discovery exchange. It receives no market data revenue.

Transparency. This is the availability of data on the system regarding a potential transactional counterparty. Most exchange limit-order books, such as Access, Globex and Eurex, disclose the quantity and price of orders but not the counterparty. Bulletin board systems, such as Bloomberg Powermatch, disclose the bids and offers from approved counterparties only. The details of the counterparty for a given trade are disclosed at the time of trade. The critical measure of transparency is when a trader learns the identity of his counterparty.

Liquidity is finite. If a new system gets it, some older system must lose it.

Transactional Efficiency. This is a measure of how a system processes trades. It is composed of several elements. The cost element is made of round-turn slippage (the difference between the prices necessary to execute positions and their fair value), plus any direct fees charged to execute the transaction. The speed element is the latency time between sending an order and the receipt of an irrevocable trade confirmation. Shorter times reduce potential error expense. The risk element covers error management: Where can errors occur? How can they occur? Who pays for them? When Intex was about to open, many traders felt they would not be exposed to the errors that typically occurred on an exchange floor. But if you intend to type a bid for $200 and mistakenly fat-finger it for $220 and it's executed, isn't that an error? Different exchanges have different rules for dealing with these types of problems.

Clearing Support. This consists of two parts, mutual offset and credit guarantee. A clearinghouse accepts matched trade records for position management. Any clearing member with an opening long position can use a short trade with any other member to liquidate the long through mutual offset. A clearinghouse can also contractually stand in opposition to members on all trades. By taking the contractual place of members, the clearinghouse guarantees the trades. This relieves the traders of their risk management function. Without a clearinghouse, traders must perform risk management on a principal-to-principal basis.

Settlement Support. This describes the process of delivering a traded item against payment. A settlement system that eliminates a significant amount of market-participant paperwork is preferred. Prior to 1987, for example, all equity trades were settled by delivery of endorsed certificates against payment by certified checks. There were hundreds of runners executing deliveries throughout lower Manhattan. Subsequent to the formation of the Depository Trust Company, all NYSE settlements have been done by book entry at DTC against wire transfer.

Color Commentary. Traders with open telephone connections to open-outcry exchanges get both continual private descriptions of activities in the marketplace and a private, oral market data feed from their ring. Traders who trade based on algorithms from a digital market data feed do not need this type of commentary.

Once these eight variables have been laid out, how does one make a decision? To me, cost and liquidity are by far the most important variables. Liquidity itself deserves at least a 50 percent weighting. Without it, no system could ever succeed. Price discovery, transparency and transactional efficiency come next as a unit. Although they can help create a niche, they can't overcome a shortage of liquidity.

When I analyze a new trading system of any kind, I lay out the variables and look at where the liquidity for the new system will come from. Liquidity is finite. If the new system gets it, some older system must lose it. To create liquidity, an exchange's traders must have risk capital in the same order of magnitude as its competitors. The exchange order flow must represent a variety of trading interests, including market-makers and market-takers.

Once you look at the other variables, it's time to make a judgment. What market niche will the electronic exchange focus on? Which of the eight factors will it master to control the niche? Does the new venture have the capital to finance a years-long fight for survival? Instinet, for example, took three years to become a success. If you cannot spell out exactly why a new electronic trading system's strategy will be successful, it probably will not.

Of the 70 systems in development, most will be out of business in three years. They will be victims of consolidation or failure. In my experience, each market niche that trades products that are in essence fungible will wind up with three companies. The first will be powerful and profitable. It will control the segment. The second will be profitable but must follow the strategy lead of the first company. The third company will struggle continuously and barely survive.