Algorithmic Trading in Commodity Markets Trayport by liaoqinmei

VIEWS: 6 PAGES: 3

									                 Algorithmic Trading in
                 Commodity Markets
                                                                                                    By Guy Isherwood

                 ALGORITHMIC TRADING HAS                    AT is continuing its march through the asset classes with
                 expanded rapidly to become a            commodity markets increasingly in the frame. Whilst there is an
                 significant proportion of trade flows   established cadre of AT platforms being used in metals, freight
                 within the capital, foreign exchange    and soft commodity markets, the energy sector has been slower to
                 and money markets. The technology       adopt this technology. AT has been successfully applied in more
                 is now well established in equities,    liquid energy markets such as crude and refined oil products, as
                 bonds, FX and their derivatives –       well as natural gas in the US. However, broader application in
                 notwithstanding the “flash crash”       many other energy commodities is still in its infancy. This less
                 back in May.                            trodden path is seen as an opportunity by many, with commodity
                    On     some   equity    exchanges    algorithms becoming increasingly sophisticated. And with more
                 more than half of their trade           institutional money flowing into commodities [see page 44]
                 flow is generated by sophisticated      managers are looking for new and innovative ways of seeking
                 algorithms. Bond and FX markets         alpha. As such, AT is being applied to the most highly volatile and
                 typically quote between 30-45% of       active commodity subsets, especially crude oil.
                 trades as being computer generated.        An increasing proportion of algorthmically-derived trades is
                    Earlier this month the Bank for      now due to the acceptance of AT by smaller trading entities and
                 International Settlements reported      asset-backed buy-side firms.
                 that average daily turnover in the FX
                 market has jumped 20% in the past       Exchange Proliferation
                 three years to US$4,000 bn a day. The     The proliferation of new exchanges, multilateral trading
                 jump in spot trading was powered        facilities (MTFs) and alternative OTC matching venues has led
                 by “other financial institution”        to a range of trading activities, especially in the search for ‘dark
                 (i.e. non banks), and much of it        pools’ of liquidity. This requires market participants to do more
                 algorithmically driven.                 pre-trade processing (for example; aggregation, smart order
                                                         routing and order slicing or ‘iceberging’). Done manually, these
        AT is continuing its march through               tasks are labour-intensive and prone to error. With AT, users can
         the asset classes with commodity                work dozens of strategies concurrently generating hundreds of
                                                         orders across multiple venues. AT tools can also process these
         markets increasingly in the frame
                                                         tasks more quickly, accurately and effectively than the most
                    The range of use cases for           productive human trader or trading desk.
                 algorithmic trading (AT) is not           Energy traders now have an increasing choice of execution
                 limited to intricate algorithms         venues for many traded products, with the proliferation of new
                 for high frequency markets. AT          commodity exchanges, especially in Europe and Asia. New
                 can      include    straightforward     trading venues have emerged often providing (or promising)
                 limit-orders and stop-loss trades       faster executions, lower costs and a variety of different structures.
                 through to the use of more complex        However, innovation and competition has made liquidity
                 algorithms using event correlation      in some markets harder to find. As new trading venues siphon
                 for arbitrage trading and portfolio     market share from established exchanges a more fragmented
                 risk management, with the trading       market can result. “Where all trading was once concentrated at
                 algorithm essentially the logic that    the primary exchange [as in the case of equities] it is necessary
                 embeds a mathematical model             to examine the new MTFs to ensure receipt of the best price,”
                 that analyses market parameters         according to Kevin McPartland, senior analyst at TABB Group.
                 and automatically generates trade       For many market participants, this necessitates access to multiple
                 execution orders into electronic        trading venues – a major operational headache, without even
                 markets.                                considering trade execution and clearing costs.
                    AT can also be applied in a semi-      Many commodities such as oil and agricultural products are
                 automatic way where trade signals       traded within highly liquid markets. Liquidity is provided by the
                 are output to be authorised and/or      high proportion of traditional paper trades that typically represent
                 executed by a ‘human’ trader.           many times the value of the underlying physical products.

1   September 2010
                                                                                                   ALGORITHMIC TRADING




  In addition, as the number of trading venues and instrument         2-3 years ago, particularly from
types increase, so do the potential trading strategies. This          Trayport’s utility clients. Utility
becomes increasingly difficult for a ‘point-and-click’ trader to      groups seem particularly interested
monitor. But an AT solution can automate monitoring across            in Automated Trading given their
multiple venues. What traders need then is robust, low latency        aften limited IT resources.
connectivity to these venues – with many firms not wanting the          The feedback Trayport received
IT overhead of building and maintaining links to market APIs.         from users gave them the threads of
  AT solutions need to manage any number of trading strategies        commonality to build the common
and risk monitoring activities operating over multiple trading        requests. “It was a question of
venues. This requires an ability to connect and read input event      looking at the market and finding
signals in real-time from various APIs.                               out what everyone wanted to do




Example: European Energy Trading                                      (and all spending money in doing so)
  An example of the development of AT in the European energy          and understanding ‘this was not a
sector is that being developed and rolled-out by Trayport.            competitive advantage for anybody
The Automated Trading system connects to Trayport’s flagship          in doing it ... its just a cost replicated
GlobalVision Trading Gateway system, providing a ‘black-box’          across the system’ ... something
solution that allows traders to use their own IP.                     Trayport should be doing,” explains
  The system has been built specifically with the energy sector in    Davies.
mind. “From a trader’s point of view its the same screen, exactly       So where does this go next? “Our
the same interface, so it feels very much part of the system,” says   feedback tells us that customers,
Richard Everett, Technical Sales Specialist, Trayport.                including those with read-only
  Trayport have built five common requests into the system            screens, are causing a blurring
which allows users to build out so that as many synthetic orders      of the lines between traders and
can be added as required on a per trader basis.                       quantitative analysts (as seen in
  “Clients can build out their own synthetic orders, with their
own logic,” explains Everett, while Trayport stand by to assist                  Automated Trading provides
(particularly smaller firms) in developing their synthetics. Thus,               a template for customers to
Automated Trading provides a template for customers to develop
                                                                                     develop their own rules
their own rules.
  “We think the system is important now because the energy            other markets). “That’s where we
sector has a level of automated trading that is proportionately       want to expand this offering further,”
low,” [somewhere up to 10% at most] according to James Davies         confirms Davies. Its about linking
who heads Trayport’s Sales & Client Services. “We undertook to        groups within the organisation.
take our API to the next level to allow customer to put their own       “Trayport makes money by
rules into our system ... customers build the rule and the system     improving efficiencies and taking
builds the order management around it.”                               costs out of the market and that’s
  The appetite for automated commodities trading developed            what we are doing here again,”

                                                                                                             September 2010   2
    ALGORITHMIC TRADING




                 says Davies, who expects steady           consequence of investment decisions that were manually derived
                 growth in the use of AT in the sector.    and executed].
                 Eighteen months from now Trayport            From an IT perspective, automated trading is balanced by
                 expects around 50% take-up by its         real-time risk assessment and control, so within an AT platform,
                 130 or so trading client base. They       trading and risk should be seen as ‘two sides of the same coin’.
                 are also working on core systems          The key to risk mitigation is the provision of a risk framework that
                 updates which will allow enhanced         will allow risk managers to limit manage trading algorithms in
                 sophistication of Automated Trading       real-time.
                 as organisations evolve their AT             Ultimately, effective risk control can only be achieved if the
                 business to embrace spread trading        business gives a mandate to a risk management team with a clear
                 and other more complex algorithms         risk policy that is independently derived [see page 65]. This must
                 across multiple execution venues.         be underpinned by rigorous risk procedures which are managed
                    So, the question becomes “what         through continuous risk assessment. Properly implemented, an
                 happens when everyone has this            AT solution can do precisely this and will actually help to reduce
                 technology? Is it a race? The answer      market risk compared with manual trading and risk practices.
                 is ... its about latency.”
                                                           End of the Human Trader?
                 AT Architecture                              Certainly not. The significance of AT relative to traditional
                    An AT solution must be fully           methods of execution needs to be kept in context – algorithms
                 integrated    within    a    trading      are essentially automated trading tools that drive increased
                 infrastructure if it is to deliver        productivity in trading operations. Effectively implemented AT
                 full trading and risk capability          thus enhances profitability whilst mitigating downside market
                 seamlessly. To function effectively,      risk. However, it should be evident that the algorithms driving AT
                 a full straight-through-processing        are only as good as the trader/programmer who creates them.
                 environment will be required – tuned      Moreover, AT cannot replace human elements in many key
                 to model user-specific trading and        trading activities such as structuring origination contracts and
                 risk requirements.                        voice-brokered OTCs.
                                                              Human traders are also essential in the AT supervision role
           Effectively implemented AT thus                 – no more so than when markets are volatile and reactive to
                                                           external price shocks. Employing an AT solution also relieves
               enhances profitability whilst               traders of time-consuming recurrent trading activities, giving
           mitigating downside market risk                 them more time to focus on P&L enhancement through crafting
                                                           more complex, higher value deal structures. It also allows them
                   Risk management is an excellent fit     time to monitor algorithms and act on step-changes in market
                 for AT modelling. It can continuously     characteristics.
                 assess the exposure of open positions        Trade automation should lead to less risk, not more. Computers
                 throughout the trade lifecycle and        are not prone to manual trading errors such as confusing a buy
                 across different slices of a portfolio.   with a sell order or adding an extra zero to the trade quantity.
                 Another use is in auto-hedging. Many      They also react far faster than a ‘point-and-click’ trader to
                 structured energy deals are long-         capture the best bids and offers. AT solutions will also execute
                 term, creating residual exposures         stop-loss or auto-hedge trades to minimise exposure to adverse
                 to foreign exchange or interest           market movements.
                 rates. Logic can be used to buy, for         The volatility exhibited in commodity markets and the greater
                 example, FX and bond instruments          scrutiny being put upon them by the regulatory authorities
                 to hedge out these exposures.             means that AT can be applied to automate real-time responses.
                                                           And a well though out AT solution can also improve trade
                 Is AT Risky?                              performance by seeking out liquidity in thin markets. Users need
                   Yes. Poorly developed black-            seamless trade flow via an integrated framework that provides
                 box algorithms, allowed to run            access to dispersed liquidity by routing orders to alternative
                 unmonitored and uncontrolled can          exchanges or OTC venues. This is where the Trayport solution is
                 result in large trading losses. But       likely to have its main benefits. But remember; an AT solution is
                 the same can be said of any poorly        only as good as the trading strategies and risk procedures that
                 constructed and monitored portfolio,      are deployed. •
                 including those built on manually
                 executed trades. [Note: All of the                   Guy Isherwood is Editor of Commodities Now.
                 large-scale energy trading failures
                                                                             www.commodities-now.com
                 from ENRON to SEM Group were the

3   September 2010

								
To top