Satu S. Parikh
Finance Department
The Wharton School of the University of Pennsylvania
1300 Steinberg Hall, Philadelphia, PA 19104-6366
LITZ2@finance.wharton.upenn.edu
(215) 382-7749
Gerald L. Lohse
Operations and Information Management Department
The Wharton School of the University of Pennsylvania
1326 Steinberg Hall, Philadelphia, PA 19104-6366
Lohse@wharton.upenn.edu
(215) 898-8541
The CME currently has some of the most crowded futures
pits in the world. Imagine you are a trader in such a
futures pit, and your task is to sell 100 soybean contracts at
a given price. You scream in as loud a voice as possible,
"100 for [price]" while simultaneously touching a closed
fist to your forehead with palm facing outwards. Within
seconds a person wishing to buy a hundred soybean
contracts
indicates a desire to buy but wishes to lower the price.
Some negotiation takes place, and once an acceptable price
is found, the trade is executed. The average trader on the
CME executes 60 trades an hour.
There are a number of existing attempts to automate some
of the most liquid financial and commodity markets in the
world, and the results have been less than desirable.
GLOBEX, a joint venture between Reuters Holdings PLC
and the CME, is one example. Two years after its
introduction, the system has been described as restrictive,
slow, and inefficient relative to the open outcry method of
trading, and there has been a great deal of debate about
why this is so, or rather why this should be so [14]. In
spite of the many reasons why a floor-based system is
"better" than its screen-based counterpart, from an
interface standpoint the superiority of the trading pit
structure is readily apparent.
The trading pits are where all of the dealing in a particular
futures instrument is legally required to take place [2, 10].
More specifically the Commodity Futures Trading
Commission's Regulation 1.38 states the following [6]:
All purchases and sales of any commodity for future
delivery, and of any commodity option . . . shall be
executed openly and competitively by open outcry or
posting of bids and offers by other equally open and
competitive methods in the trading pit or ring or similar
place provided by the contract market.
In these pits, the traders' shouts and hand signals form a
complex network of communication that may seem
uncontrolled and chaotic, but which is, in fact, remarkably
orderly. One may postulate that with a computerized trade
execution system, communication between traders can be
provided by showing the best bids and offers in the market
for a specific contract at any particular moment, thus
eliminating any need for hand signals and/or verbal
communication. This is true, but what needs to be
considered and explained is that by removing the human
element in a trade, the efficiency of price matching may
increase but only at the expense of market liquidity, which
is defined as the ability to execute an order of any size
without a significant discrepancy to the market price [3].
Thus, the electronic automation of futures markets has
definite interface design issues, namely effectively
replicating the communications bandwidth of floor-based
futures trading in an electronic securities market. It is
important to note, however, that the intent of this research
is not to provide decision support for those traders engaged
in open outcry. Rather, it is directed towards addressing
interface design problems and issues in screen-based
futures trading.
To understand why the existing computerized trade
execution systems fail to substitute for an open outcry
market, one must first understand the basic market
principles as well as some details about how different
orders are executed. Thus, section 1 begins with an
overview of the futures markets. Section 2 delves into
aspects of screen-based futures trading systems currently in
operation with a primary focus on GLOBEX. While other
systems are currently in operation [12], GLOBEX is the
highest volume electronic market and is the most widely
used for after-hours trading [9].
Section 3 looks primarily at the interface considerations of
both the floor and screen-based trading systems. Here the
different problems are stated and explained, and possible
ways to increase the communications bandwidth in an
electronic futures market are explored.
Section 4 describes the components of a prototype we are
developing that represents the beginnings of a solution to
this very difficult interface design problem.
Hedging is used to offset the risk exposure from potential
future cash market transactions in the product. For
example, large multinational corporations may use
currency futures to hedge against a future payment in
foreign currency or lenders may use interest rate futures to
hedge against a future fall in rates.
Obviously futures are very useful in reducing market risk,
but they may also be used as speculative instruments.
Arbitrage is exploiting small discrepancies in different
markets. By having positions in both the cash market and
the futures market, one could earn a riskless profit if
market inefficiencies arise.
Market making is not specific to futures, but the way it is
implemented is unique to the futures markets. Making a
market involves taking both sides of a trade and profiting
off of the spread, which is the difference between the buy
(bid) and sell (ask) price. In most futures exchanges,
market makers, also called floor traders or speculators, are
those trading for their own account as opposed to filling an
order for a customer. Thus, it is the market maker who
supplies much of the capital that ensures a liquid futures
market.
Figure 1. How a trade is made [4]
Once the broker has indicated the buy order, other brokers
with sell orders or market makers wishing to create or
close out a position may choose to sell the required
contracts to the buyer-represented broker. If the broker
wishes to buy a large quantity of contracts, she may enter
into a number of trades with different brokers in order to
fill the order. At any rate, once the order has been filled,
the broker writes the seller's identification number and the
details of the trade such as the time, price, and quantity on
a trade ticket. Once the trade is executed, exchange
employees relay the price to the exchange's computerized
price reporting system. In relaying the information, the
exchange's employees use a variety of methods. The CME
uses mobile phones; the New York Cotton Exchange uses
hand-held computers; and the Coffee, Sugar and Cocoa
Exchange employees use hand signals to report prices to
other employees entering the price data into a computer.
This information is then transmitted to market observers
and participants worldwide, thereby allowing people
around the globe to trade with relative ease.
The last step in the execution process is confirming the
trade. A runner returns the filled order to the member
firm's communications desk, where it is once again time-
stamped. Confirmation of the futures purchase is then sent
to the customer.
In understanding the different interface implications in
automating the futures market, one must be familiar with
how and why such a market functions. The next section
explains the GLOBEX system and other automated
securities markets.
Primarily, two types of orders are given, market and limit
[11]. A market order instructs the floor broker to execute
the trade at the current market price or better with no
restrictions on price. While a limit order, restricts the
broker to a given price or better in her bidding or offering.
Many view GLOBEX to be restrictive in placing different
types of trades, since GLOBEX only permits limit orders to
be entered. A trader at a GLOBEX terminal chooses the
contract he wishes to trade in, and then enters an order,
which is ranked by a price/time priority algorithm and
stored in a limit order "book." The trade is executed when
the prices cross.
GLOBEX offers two trade options, hit and take, which are
akin to market bids and offers, but the GLOBEX
alternatives differ from market orders distinctly. Hit
orders, when entered, place a sell order at the best bid
price, and take orders place a buy order at the best offer.
Both are supposed to duplicate market orders. The
difference, however, lies in reconciling the quantity of
contracts specified in the orders. On the floor, market
orders are either filled in entirety or in part. If they are
only partially filled, the broker can continue to trade the
remaining contracts at the current market prices. On
GLOBEX, hit/take trades are filled to the quantity of the
standing order, and the remaining balance is automatically
canceled.
Another way a market order is constructed on GLOBEX, is
by entering an order at a better price than the best standing
order. Thus, the new order is immediately executed at the
best bid or offer. If the newly entered order is for a large
quantity, the system moves through the book, providing
fills at the best prices, until the order is filled or the limit
price is reached.
Figure 2. GLOBEX Screen
The trading window is where a trader finds the best bid
and offer available for a GLOBEX product. As it contains
the most information, it is the largest of all the windows,
comprising nearly a third of the entire screen. It contains
the current contracts being traded as well as their best bids
and offers. The quantity and last trade price are also
displayed along with the high, low, and volume for the
current trading session. The trader may modify a portion
of this window in order to show the total number of his
orders entered as well as his last bid/offer.
In addition to displaying the market for a given contract,
the trading window is where trades are entered and
modified. Assume the trader's intentions are to place an
order to sell a notional amount of a given product listed on
GLOBEX. First, he must depress the number key
representing the contract or use the mouse to select it.
Once this is done, a dialog box appears giving the trader a
number of alternatives. More specifically, he may bid/offer
the contract thus making a market, separately bid or offer
the contracts, hit/take the best bid/offer, change or cancel
an existing order, request for quote, or display the order
book which only shows the five best bids/offers. Assuming
this trader wishes to sell his contracts, he would use the
keyboard's "arrow" keys coupled with the "enter" key or
the mouse to select the "offer" button. The trader must
now grapple with another dialog box requesting the details
of the trade, such as the offer price and quantity. After all
information has been entered as desired, the trader clicks
the "OK" button, thus entering the trade into the limit
order book. The trade is subsequently ordered by
GLOBEX's price/time priority structure and is advanced in
the execution queue as trades are executed.
Trades take five seconds to enter for hit/take orders and ten
seconds for regular bid/offer limit orders. Use of the
mouse in trading generally speeds up the order entry
process. By allowing traders to quickly select choices
instead of typing, order entry using a mouse is 30% faster
than a keyboard.
The response or trading mailbox window is smaller than
the trading window. This window enables the trader to
receive messages from the central or host computer,
including order acknowledgments, executions, and
respective order numbers.
The ticker windows lies to the right of the trading mailbox.
It displays the most recent market movements providing an
updated message each time a trade occurs and each time
there is a change in the price or size of the best bids and
offers. The ticker employs a vertical scrolling technique in
updating market information. Each time there is relevant
information, it is displayed from the top of the windows
and each successive new piece of trade information is
updated below. Thus, the window does not actually scroll
but rather updates and overwrites old information.
GLOBEX also utilizes a monitor window, which contains
information from the cash (spot) markets. An option
allows the trader to show those futures' and/or options'
prices from those traded in the trading windows. The
trader may also view a number of indices and market
composites which are provided by GLOBEX as well.
The Request For Quote (RFQ) and Alert windows alerts
the trader when the price of a selected instrument has
reached a preset level. GLOBEX permits alerts in all of its
listed products. Another feature, which is appealing for
thinly traded contracts, is the RFQ. If a contract in which
a trader wishes to transact in is not being traded, the trader
may "ask the market" for a quote by clicking the
appropriate button on the particular contract's
aforementioned dialog box. This is broadcast to GLOBEX
screens around the world. Others wishing to trade or
perhaps make a market in the contract may then choose to
respond with a bid or offer in the requested market.
GLOBEX is efficient in clearing and matching trades
according to its price/time priority algorithm. However,
the system is inadequate in maintaining an active and
liquid market. This is partly due to the nature of after-
hours trading, but much more attributable to the system's
inability to recreate the human-to-human interaction
present in the trading floor environment. This interaction
is forgone (1) by GLOBEX's method of order entry and
matching, and (2) by the system's interface.
At 10:30 the best buy order is at the price of 1.55 while the
best sell order is for 1.64 per contract. Since order prices
do not cross, no trade execution is possible. Note that the
spread between the best bid and offer is .09. This simply
signifies that the cost of buying while simultaneously
selling a notional amount, would cost .09 per contract. At
10:35 the best offer is entered at 1.59 per contract, while
the best bid stands at 1.605. Since the best offer is less
than or equal to the best bid and, conversely, as the best bid
is greater than or equal to the best offer, the trade is
executed at the standing order's price of 1.605. Since the
quantities are not equal, only 50 of the 95 contracts are
executed at the bid price. The balance, at the trader's
request, may remain as a standing order or simply be
canceled. At 10:35, immediately after the trade has been
executed, the order book would look as follows:
A new best bid is entered at 10:36 at the price of 1.63.
Since this crosses with the best offer, another trade is
executed. However, as there are two offers at the price of
1.63, GLOBEX makes use of the time priority to
distinguish the orders. Since both orders occurred at
different times, the offer entered first is given priority.
Thus the new bid is executed at standing price of 1.63, and
since the existing offer is for 35 contracts, this is the
number of contracts traded. Assuming the trader entering
the order did not want the balance canceled, the remaining
10 contracts would be executed against the new best offer
of the same price for 70 contracts. Thus, the order book
would look as follows after the new order-entry at 10:36
and the two new trades occurring immediately after 10:36.
As new orders are entered, they are prioritized in the book
as described, and if matching is possible, the trade is
executed at the standing price. Reconciling the quantity
depends on the trader's instructions. If the balance is left
as a standing order, it remains in the order book and is
executed as illustrated above; otherwise it is simply
canceled.
Orders may be modified on GLOBEX. Traders may
change the price and/or quantity of an existing order if it
has not yet been executed. However, by modifying an
existing an order, time priority is lost. The system views
order modification as equivalent to canceling the existing
order and entering a new one. This may prove costly in a
high volume market, where many orders exist for a given
price.
Although GLOBEX allows market making it does not have
the communications bandwidth necessary to trade actively
in a liquid market. As an illustration, assume two markets,
GLOBEX and a floor based exchange. In addition assume
a market maker wishes to place a bid and an offer for a
highly liquid contract. On the floor this trader using the
open outcry method of trading can indicate her desire for
both orders in tandem. Using hand signals she can place
the order in a few seconds. On GLOBEX the trader would
have to grapple with the contract's dialog box which does
allow a "BID/ASK" option. After entering a specific bid,
offer, price and quantity, the order is entered into the limit
book where it is ranked according to the price/time priority
algorithm. Clearly, this would take much longer than
simply waving and shouting one's desires. Speed,
however, is not the only issue of import in distinguishing
between the two systems. Versatility is equally substantial.
If the market begins to move away from the trader, on the
floor she can simply modify her hand gestures and
shouting to account for this. On GLOBEX it is not as
easy. The trader can clearly change her order, but not
without relinquishing time priority in the queue. In a
liquid and volatile market, forfeiting an advantageous
place in the order queue may result in an unconsummated
order.
One also has to be aware of the aforementioned
communications bandwidth present on the floor of an
exchange. In a specific pit, a trader can view the whole
market in front of him. He can select an appropriate price
and trade immediately. In addition, it is not imperative
that he trade with someone in physical proximity, the
trader can choose someone across from him or anywhere in
the pits. This is because the system of hand gestures is so
efficient, traders can signal intentions and desires at an
instant to a large number of people.
Another example of the pit psychology is available in Leo
Melamed's Critique of Automation [14]:
"There is yet another type of liquidity which is
exclusively a product of the floor, or rather of the pits....
This liquidity is the product of pit psychology. For
example, pit trader A, a larger trader who trades solely for
his own account, is about to enter the pit. Pit trader A is
known to have �bought the market' earlier that day or the
previous day. The market has fallen during the last thirty
minutes. As pit trader A enters the pit, but before he has
indicated any bid or offer, he is seen by pit traders B, C, D,
and E, who also trade for their own accounts. Pit trader B,
a fairly large local trader, is short the market. His split
second reaction is that A will no doubt continue buying the
market and perhaps cause the rally. Thus, he immediately
buys to offset some of his shorts in an attempt to secure
some of his profit. Pit trader C, a fairly large local trader,
who is long the market and is presently with losses and
nervous about his position, reacts entirely differently. He
feels that trader A is about to sell and liquidate his entire
long position causing the market to fall further. Thus C, to
minimize his losses, immediately begins to sell the
available market bids. Pit trader D, a small local �scalper'
believes that A will not sell out his position, but rather that
he will buy more. Therefore, D tries to buy one or two
contracts for his own account, so that he can �scalp them a
few seconds later at the higher price he expects will soon
develop. Pit trader E, a �spreader' between two or more
different contract months, has just taken a short position in
the contract month where A is about to enter the pit. E
had intended to stay in an �unhooked' position for thirty
seconds or so before completing his spread by buying the
other contract month. Upon seeing A, however, he decides
quickly to �hook up' his spreads in the other contract
month so that he does not unnecessarily expose himself,
since he is unsure of what A will do.
Actually it may turn out that A entered the pit, looked
around, and left without trading; or he may have done
what some of the traders expected. However it really does
not matter what A did. All the bids, offers, and
transactions in reaction to A's entering the pit accounted
for additional market liquidity. In fact, they created new
market positions which, in turn, would have to be offset,
probably that day, thereby creating further market
liquidity.
The reasons for the trades are unimportant to this
discussion. It is important that there was a visual and
psychological interaction which cannot be duplicated by
computer. Without the pit psychology, this form of
liquidity is lost."
Currently, virtual reality (VR) systems are available for
viewing financial data. For example, VR applications exist
for showing multidimensional financial data to aid foreign
exchange currency option traders using head- mounted
display helmets, data gloves, and stereoscopic images[15].
Another example is a VR room with bobbing icons, each
with different visual attributes corresponding to different
financial data, through which a broker can accurately view
a client's portofolio holdings[8]. This technology in its
refined state could greatly facilitate the trade process, but
current technological limitations make these early systems
impractical for trading use. Other technologies with
application in the area include an eye tracking device for
object selection, voice recognition, multimedia, video
conferencing, and touch screen.
Design distinctions must be made between the
symbolization of information, defined as data relevant in
decision-making, and tasks (i.e., bidding, offering,
canceling, etc.). Information can be further classified as
either quantitative in nature (i.e., prices, quantities, spot
data, etc.) or qualitative or psychological in nature. The
latter characterization is illustrated by the pit psychology
example. It is difficult to transpose this information
asymmetry, present in psychological information, onto a
screen-based system, but a number of ways exist to
incorporate the available quantitative figures onto such a
system.
Generally the above relationships hold, but, since traders
are continuously gleaning profits and losses from the
market, movement between "rings" is common [1]. In
addition, there is not and should not be a definitive line on
the pits demarcating the different classes of traders.
Figure 3 Trading Pit Hierarchy
Using icons gives users symbols of actual human traders.
With the advent of video conferencing it may soon be
possible to show a bit mapped or even real-time image of
each participating trader. While this may be farfetched,
the idea should be transparent: to give the trader the
human element which exists on the floor. Using traders'
initials as the icon is more practicable, and coupled with
price and quantity data, the icon would combine
quantitative as well as qualitative aspects into one package.
It is also easy to view the icons superimposed onto a bit
mapped image of a trading pit. To account for the
hierarchical issues discussed above, icons can move to
different regions on the pit depending on some moving
average of that specific trader's total dollar volume of the
current day (initial positions can be determined from the
previous day's volume figures). Figure 3.1 shows an
example of a trader-populated electronic pit. Each icon
represents a trader. Those willing to buy are represented
by a white icon and those willing to sell are depicted by a
black icon. The icon for a small trader or scalper V, would
indicate an offer for 5 contracts at 1.53, while an
experienced, high-volume trader may be trader AE who is
located towards the center of the trading ring bidding 1.37
for 200 contracts.
Figure 3.1Electronic Pit Representation
London International Financial Futures Exchange's
(LIFFE) Automated Pit Trading (APT) system [13] uses a
10 second limit on all orders after which they must be
refreshed. This added feature forces traders to get more
involved in the trading process. On the floor, traders must
shout and gesticulate continuously in order to place a trade.
Thus, on a theoretical system as proposed above, such a
feature could be incorporated to keep the traders' orders
executable as long as their "breaths are still warm[13]."
This system must also allow orders to be changed easily.
Currently, GLOBEX can cancel an order and reenter it
with a time priority loss. A new such system would have
to allow order modifications as dynamically as possible,
without any such loss of time priority.
Abstract
The primary concern in designing an interface for an
electronic trading system is the impact on market liquidity
[9]. Current systems make use of efficient order-execution
algorithms but fail to capture elements of the trading floor
that contribute to an efficient market [9]. We briefly
describe tasks conducted in futures pit trading and current
off-hours electronic trading systems. Understanding the
tasks helps define key components to an interface for
electronic trading. These include visualization of the
market and its participants, a trading process which allows
active participation and price discovery as well as
concurrent interaction among each of the participants.
Keywords:
Futures trading, automated exchange,
trading pits, interface design, electronic markets
Introduction
In the futures pits of the Chicago Mercantile Exchange
(CME), tens of thousands of people crowd into 70,000 sq.
ft. and trade in excess of 550,000 contracts a day by using
their voices and hands. Even this volume is pale in
comparison to the total dollar volume of all futures
contracts worldwide, which are over $500 billion a day
[12]. In comparison, the electronic automated counterpart
to the Exchange, the GLOBEX Trading System, trades
only 6,000 contracts daily. "[GLOBEX] is just nickel-
and-dime small investor retail business so far," says a
portfolio manager at a large New York investment firm.
Another person familiar with exchanges and trading, Mr.
Dale Lorenzen, first vice-chairman of the Chicago Board
of Trade, readily admits, "I just don't think GLOBEX is
ready to take the volume that we can generate with open
outcry." Why an electronic system cannot effectively
replicate the human component of floor trading, highlights
some important interface design issues.
FUTURES MARKETS
A futures contract is simply an agreement to buy or sell a
given amount of a commodity or financial instrument at a
particular price on a stipulated future date. Common
contracts are crude oil, precious metals, agricultural goods,
interest rate instruments, and currencies. Each is traded
on an exchange, and some are traded on more than one.
Futures contracts are used for hedging, speculation, and
market making.
Order Execution
The actual trade is executed in a ring shaped pit. The
buyer, wishing to invest a given amount of money in a
particular contract, places an order with a member firm
(e.g., Lind-Waldock Futures Inc., Merrill Lynch, etc.). In
placing the order, the buyer may set certain restrictions on
the order such as price limit, market, or opening only [16].
The member firm communicates the order via telephone or
computerized order-entry to its clerks on the exchange
floor. The order is time-stamped for order-clearing
purposes, and then a runner delivers it to the firm's floor
broker in the trading pit. Using open outcry, the floor
broker shouts the order into the pit. As there are many
brokers and traders shouting in these pits, shouting alone
may not accurately convey intentions to buy or sell. Thus,
hand signals are an integral way in which a pit trader or
broker communicates with the rest of the pit participants.
Order Clearing
After an order is filled it must be cleared by the respective
exchange's clearing agent. The CME's Clearing House is
the guarantor of performance for each futures contract
traded. This means that the timely delivery by the seller of
the exact quantity and quality of the commodity or
financial instrument and the full and timely payment by
the buyer, are fully ensured by the Clearing House. In
addition to monitoring the physical delivery process, the
clearing agent oversees the daily settlement of all accounts.
This means that if an unliquidated (open) position exists at
the end of the day, it must be marked-to-market. Marking
to market involves recalculating the value of the position
using that day's settlement price. Any loss in value is
treated as a debt to the clearing agent which must be
repaid, and any increase in value is credited to the
customer's account.
AUTOMATED SECURITIES EXCHANGES
Rapid development in technology has permitted the move
from floor-based markets to screen-based systems, and one
such system currently in operation is the Global Exchange
(GLOBEX) Trading System. More specifically, GLOBEX
is an international, automated order entry and matching
system that is operated by Reuters Holdings PLC and the
CME. The GLOBEX Network extends to ten financial
centers, including New York, Chicago, London and Tokyo.
At present, more than 325 trade terminals are in operation.
The system operates after the close of the regular trading
hours and closes just prior to market open [7, 9].
Types of Orders Available
PROBLEM DEFINITION
The problem in electronic securities exchanges is the fact
that by automating the trade process, the human element is
forgone. While the efficiency of trade clearing and order
matching may increase, the added liquidity, provided by
those willing to trade on the human element (i.e., noise
traders), is sacrificed. The question at hand is how to
improve the human interaction in these automated trade
systems.
Current GLOBEX Interface
The current interface for GLOBEX (Figure 2) uses a
windowing technique which subdivides the area of the
screen into windows: trading window, response window,
full ticker window, request for quote (RFQ) and alert
window, monitor window, and the limit order book. While
this setup may be adequate in displaying one window of
information, it is difficult to view information across
multiple windows. Further, the interface slows the entry of
trades relative to open outcry and it would be difficult to
enter, let alone execute, 60 trades per hour using
GLOBEX.
Order Execution Algorithm
The rules of an automated trade-execution process are
referred to as an order-execution algorithm [11]. For
example, GLOBEX uses an automated continuous double
auction, in which transactions take place when limit prices
cross. For the continuous double auction process, a
standing price/time priority structure is used. That is, bids
(offers) are assigned priority in the limit order "book" from
the highest (lowest) price to the lowest (highest) price.
Orders of equal prices are given priority according to the
time of entry. When order prices cross, the trade occurs at
the lower of the two quantities specified in the orders.
The following brief example illustrates this process.
Assume the following chronologically successive bids and
offers for a specific futures contract:
Discussion of Order Entry and Matching
An understanding into why GLOBEX hinders an efficient
and liquid market thus requires discussion into some of the
previously published ideas about the inadequacy of trade
automation. The two primary reasons explored are the
importance of traders or "locals" and the psychology of a
futures pit[14].
Locals and Traders
The trader is integral to a properly functioning market.
Trading for their own accounts, they have a great deal of
flexibility in deciding how their money is allocated. Some
choose to assume no risk by taking both sides of the market
and profiting from the spread (i.e., scalper or market
maker). Others incur some risk by playing natural market
movements (i.e., spreader, arbitrageur, or position trader).
Whether some or a great deal of risk is borne by traders,
they are infusing the markets with a large supply of liquid
reserves, and it is these liquid funds which result in
commitments to a newly listed futures contract or option as
well as the long term viability of an established market.
Pit Psychology
Less quantifiable than a high volume and liquid market are
the psychological issues unique to the trading pit and how
these are to be replicated on a screen-based trade system.
Many traders, experienced with both the pit structure as
well as GLOBEX, assert that on the floor there is
information meaningful to the negotiation process which
cannot be found on GLOBEX. One example is noise or
information trading. This refers to the increased noise in
an extremely active market. Traders sometimes choose
which pit and its respective contract to risk their money in
by the noise associated with it. On a screen based system
such as GLOBEX there is no way to hear price movement.
There is no replication of the hundreds of shouting voices,
waving hands, or stamping feet that one commonly sees on
the trading floor. Thus, it can be argued that this lack of
auditory and visual stimuli on GLOBEX may preclude
some involvement by traders seeking profit.
Interface Design Issues
The primary concern in creating a screen-based system is
the impact on market liquidity [9]. Many factors on the
trading floor contribute to an efficient market, and these
may not be able to be fully replicated on a computer [10],
but full replication should not be the immediate goal.
Rather, a system developer should be concerned with the
implementation of an interface which incorporates some of
the defining characteristics of a floor-based market. These
include visualization of the market and its participants, a
trading process which allows active participation, and the
human interaction between each of the participants. The
last characteristic clearly is not easy to create. How does
one duplicate the traders' pitch and tone of voice and level
of urgency in order execution? The answer partially
depends upon the state of technology.
The Pit
In creating an automated trading environment, it is
necessary to establish a central object which will represent
the place where "virtual" traders will reside and bid and
offer contracts. A pit structure, such as those on the CME
or COMEX, may be chosen as an appropriate central
object. As the pit is static in nature, it may be constructed
using a graphics program or bitmap image. It is
worthwhile to consider that on the different floor
exchanges, the actual physical characteristics of the pit
differ. This suggests that they serve little purpose other
than to provide a central area where efficient and open
exchange is possible. However, in creating this image, it is
necessary to indicate leveling. An example of the
aforementioned psychological information, the grading of
the pit exhibits a hierarchy among traders residing there.
Figure 3 illustrates the idea.
Traders
Traders view each other as information segments. Not
only do they help relay public and private information to
the marketplace, they signal their intentions by their
bid/offer price and quantity. Furthermore, there are
psychological ramifications to their every action. Thus, by
having a list of the best bids and offers, as GLOBEX does,
one does not see the entire market and the intentions of its
participants. By modifying GLOBEX to show the entire
limit order book, this would be solved, but at the
significant cost of feasibility since these limit books are
generally very large. Another way is to use an icon
representation of each trader. Since some markets are very
large with respect to the number of participants, there is an
issue of available screen "real estate." A solution to this
would be to dynamically display those traders who actually
have standing bids and offers. Unfortunately, this avoids
the psychological component of viewing the entire market,
but it is feasible given the current state of screen
technology.
Trading Process
Optimizing the process by which trades are executed has
important implications for interface design. It is necessary
to note that this process is primarily governed by the order-
execution algorithm rather than the interface. However,
the ease of trading on a screen-based system is directly
dependent on the user's perception of the information and
the facility of her cognitive processes in interpreting it.
An advantage of using the icons in the screen design is
their functionality. Not only do they serve as visual
representations of human traders, they also can be used as
order execution vehicles. By using the mouse to click on
the desired trader's bid/offer, the user can quickly choose
her best price and quantity. This is assuming traders are
actively involved in the market as opposed to simply
entering limit orders. After viewing the marketplace and
it's best prices, a trader can quickly execute the trade by
clicking the respective trader's icon. Thus,
communications bandwidth is maintained as is the speed of
order entry and execution.
Other Design Issues
Other market information such as spot and forward data,
should be included as well. These quantitative facts may
be displayed in separate windows. Thus, this element of
the configuration would be very similar to GLOBEX.
Since they are solely quantitative in nature, there is no
need to be overly intricate in designing its interface. One
technique would be to make use of scrolling which
GLOBEX currently uses or create a substitution routine to
simply replace the current data.
