Rapid advances in trading technology are revolutionizing the landscape

of the futures industry. A new type of quantitative trader is migrating into

the futures markets, applying new strategies and new execution tools
around the world. At the same time, an increasing number of futures mar-
kets are now accessible for purely electronic trading, shaking up tradi-
tional methods for trade execution.

Behind the scenes, a massive increase in market data is putting tremen-

dous pressure on the industry’s ability to process transactions. Faster
connections to exchanges result in a feedback loop between outbound
market data and inbound order messages, with each flow driving the
other higher and higher. This is exposing weak links in the processing
The Domino chain that can cause bottlenecks and slow down the entire process.
Effect:
These changes are making established players re-evaluate their trading in
The Changing
order to avoid slippage. Even though the quants represent a relatively
World of
small part of the total daily trading volume, the tools they are using are
Futures Trading
affecting the entire landscape. Clients are having to adjust to changes

By Greg Wood such as reduced bid-ask spreads and decreasing trade size, and in many
cases they are asking their brokers to provide the same advanced execu-
tion technologies deployed by the quants. The evolving landscape also
requires significant advances in the efficiency of post-trade processing to
cope with the massive increase of information flows among brokers, cus-
tomers, and exchanges.

44 Futures Industry
Recent Migrants have to process more fills for their orders brokers to adopt these standards for elec-
to the World of Futures than they would have seen a few years ago. tronic communication and are driving the
Another consequence has been the phe- industry as a whole in this direction. But
The futures industry is no stranger to nomenon of “phantom quotes”. The deploy- these standards are just part of the answer for
quantitative approaches to trading. ment of high-speed black-box trading firms providing greater efficiency across the entire
Commodity trading advisors have applied into a wider range of markets has noticeably trade life-cycle.
mathematical models for decades. What accelerated the frequency of order book Existing practices for futures settlement
makes the new generation of quants different updates, to the point where quotes come and based around Street-standard operations sys-
is the impact that their trading techniques go from the screen faster than a human being tems groan under the current weight of the
are having on the markets themselves. can respond. This is driving a demand for load that they are carrying. Post-trade alloca-
There are many types of quant trading brokers to provide a new generation of tion of thousands of small trades force all par-
firms and quant strategies, but one common “smart tools” for executing trades. ties in the give-up and clearing processes to
theme is that they are bringing automated look for electronic efficiencies and better
strategies refined in other asset classes into Setting the Bar for Efficiency straight-through processing, as opposed to
the futures markets. This trend has been the short-term solution of having more peo-
developing for several years, but lately the The most extreme users of quantitative
trading are amongst the most demanding of ple working longer hours.
trend has accelerated and expanded into new Possibly the biggest challenge with regard
areas of the futures markets as they have efficiency. They require consistent, reliable
trading solutions for all instruments, even to to processing is accepting that for every
become more accessible for this type of trad- information superhighway built for the bene-
ing. This includes not only futures markets the point of mandating specific latencies for
transactions. And they demand the lowest fit of the front-office, one equally as large if
for commodities, such as energies, metals, not larger must exist for operations to use.
grains and even livestock, but also certain possible “touch” for settlement across one or
more clearing brokers. Every effort to improve efficiency in getting
overseas futures markets that until recently to the execution venue has an impact on the
attracted mostly domestic participants. Exchanges are reacting to this demand by
offering their own solutions in this quest for post-trade process.
One consequence of this migration has
been a steady decline in average trade size. speed. Most major exchanges now have a co-
location solution for member firms to place Feedback Loops,
As computers replace human order entry in
the transaction process, large orders can be their—and their clients’—black boxes as Slow Consumers
quickly and efficiently divided up and sub- close as possible to the trading host. The and Domino Effects
mitted as many small orders, thereby reduc- major exchanges are also finding ways to
With the adoption of black-box trading,
ing costs associated with market impact. deliver more data, more quickly. The CME is
the primary issues affecting clients, brokers,
At the same time, a small but growing a technological leader in this space with its
and exchanges are often now related to the
number of firms are engaged in high velocity adoption of newer versions of FIX and data
speed of sending/receiving data and getting
arbitrage, seeking out tiny variations in price compression tools such as the FAST protocol
orders to the market as quickly as possible.
that might only exist for a few moments. for its market data. Similarly, Liffe now offers
Such latency demands are now measured in
“Ultra low latency” and “co-location” are premium services such as optional 100MB
ever decreasing metrics of milliseconds down
now common buzzwords in the futures indus- high-bandwidth connectivity that will accel-
to nanoseconds.
try, reflecting the drive for exceptionally fast erate trading and eliminate the need for mar-
These demands have far wider impact
connectivity to exchanges. This phenome- ket-data compression.
than might be imagined:
non may be driven by only a handful of firms, Technology standards such as FIX 4.4 and
• Feedback loops are phenomena that are
but it affects everyone in the market, includ- FIXML, which originated in the equity
often overlooked. The faster speeds you
ing traditional futures participants who now world, set the bar for entry into the business.
can obtain using ultra-low latency result
The most demanding firms are forcing their

November/December 2007 45
 Feedback loops, slow consumers, and
domino effects in futures execution and clearing

This diagram gives a schematic representation of the information flows within the trade processing chain. All information flows within the
central feedback loop are real-time and an increase in any one flow triggers a corresponding increase in related flows within the loop.
On a typical day, the number of order messages transmitted from customers to brokers and onto the exchange can reach several hundred
messages per second. The ratio of messages returned starts at 1:2 (one order: one acknowledgement, one execution) and increases as clients
amend orders and/or many executions occur. The number of price updates on a particular contract reflects this activity and scales as more con-
tracts are subscribed to.
A slow consumer anywhere in this process can lead to bottlenecks in the flow of information and a domino effect on the downstream recip-
ients of that information.

46 Futures Industry
 in increased updates in market data, gen- mation and managing execution. Theirs is and when to hold back, it is now an algo-
erating more buy/sell indicators, generat- often a pure quest for speed. Get closer to the rithm electronically making that decision.
ing more orders, generating more market exchange, receive market data faster, dissem- This allows clients to “fire and forget”, send-
data, etc. inate, digest, and react faster. Look for pric- ing their orders into the broker’s execution
• The combination of increasing volume ing errors and exploit them, draw out management system and putting the onus
with decreasing trade size means that the liquidity that might be potentially hidden, back on the broker to deliver the optimal
number of messages being generated for and use that to their advantage. They need fill. The use of smart tools also allows the
both trading and operations is growing solutions for nanosecond latency through broker to handle complicated instructions
exponentially. proximity hosting or co-location at the from clients such as contingent orders or
• Expanding real-time market data band- exchange, allowing them to focus on partic- precise execution instructions throughout
width specifications require innovative ular trading venues and/or to distribute their the trading day.
solutions to compress data with minimum execution methods across the globe.
delays in processing the data. The majority of players, however, cannot Where Do We Go From Here?
• The cost of storage for tick data for later ref- justify the massive investment necessary to The futures landscape will continue to
erence rises as message volumes grow, coun- build this infrastructure. Yet they are finding evolve. The quants are here to stay, and
teracting the decreasing cost of disk space. that current methods for order execution are everyone who trades futures will need to
adjust their systems to cope accordingly.
The events of August were a call to
Keeping up with the latest exchange API action in this regard, as a sudden outburst of
model-driven trading triggered an explosion
can be costly and time-consuming. Execution in volatility and record amounts of trading in
virtually all global futures markets. The
can easily slip as prices on the market come demands of everyday trading were already
and go in milliseconds. And snipers and posing efficiency challenges. This market
turmoil reinforced the importance of invest-
guerrillas are waiting for every move, ready ing in the excess capacity necessary to han-
dle unexpected bursts in market activity
to pick off perceived pricing inefficiencies. without experiencing delays—particularly
with regard to handling the explosive quan-
tity of market data that was witnessed on
some days. But investing in an information
• The effect of a slow consumer of data can no longer as effective in the new landscape. superhighway to get orders ever faster to the
cause a bottleneck that then backs up the This actually highlights an irony in the evo- marketplace means little if clearing and set-
entire process. A slow consumer is a tech- lution of execution technology. tlement processes cannot keep up with the
nical term for a process that cannot react In the open-outcry era, clients gave their speed and volume of transactions. If a single
fast enough to the information that is orders to brokers for execution. One of the link in the chain—or more likely, multiple
being fed to it. Examples of slow consumers notable changes that occurred with the links—cannot keep up with the flow of infor-
exist everywhere. Participants might not extension of electronic trading facilitated by mation, then potentially everything is
be able to take a clear view of the masses of broker front-ends, multi-broker vendor sys- affected.
data produced by volatility explosions. tems, and the FIX Protocol is that control At best, the traffic is slowed down—
Exchanges might not be able to handle the over execution moved back to the client. including the clients leading the flow who
volumes of messages generated by an unex- Now, however, managing execution has now cannot be certain of their positions at
pected interest rate decision. Settlement become a struggle. Keeping up with the lat- the start of the next trading day if their state-
processes might take several hours to est exchange API can be costly and time- ments are late. At worst, one failure in the
process trades on busy days, leading to consuming. Execution can easily slip as chain brings down several others behind it,
delays in generating statements. prices on the market come and go in mil- creating a domino effect.
• Finally there is the domino effect as one liseconds. And snipers and guerrillas are There is no single answer to the problem
component teeters on the edge of stabil- waiting for every move, ready to pick off per- but rather a variety of solutions for meeting
ity, falls over, and brings down several ceived pricing inefficiencies. different customers’ needs at different points
other pieces that are closely stacked Brokers have recognized this and have in the transaction chain.
around it. Such events have wide-reach- developed tools to remove the execution
ing effects and demonstrate the need for component of the order from its original cre-
greater efficiency throughout every aspect ation. Execution management systems using
of the trade life-cycle. algorithms developed by the brokers’ own Greg Wood is vice president and head of client trading
technology professional services at Credit Suisse
quants are now available for the buy-side
Securities (USA) in New York.
Weapons of Mass Discretion community. The irony is that clients are
Some players are willing and able to build again giving their orders back to brokers to
their own infrastructure for consuming infor- work at their discretion, but instead of a
human making the decision when to trade

48 Futures Industry