By Abhinay Reddy, MBA Capital Markets, NMIMS
Algorithmic Trading makes use of computer algorithms to decide on the price, timing and quantity of an order and to minimize manual intervention. Direct Market Access -DMA- is a facility whereby brokers offer clients direct and full access to the exchange trading system through the broker’s infrastructure without any manual intervention.
Large volumes are traded through algorithmic trading in the U.S. and Europe and many firms have begun offerings in the Indian market in the last one year. This article examines some of the merits and challenges posed by algorithmic trading, its performance against manual trading and its impact on capital markets.
Direct Market Access -DMA- and Algorithmic trading was approved by the Securities and Exchange Commission of India -SEBI- in April 2008 but owing to the high volatility in the markets, the response was slow and it has picked up only in the last one year. India is in an Early Adapter Phase and moving towards a maturity phase where more market participants start using the new technology. It was first introduced in India by Lehman Brothers Inc. Other firms offering these services include Goldman Sachs, Credit Suisse JP Morgan, Religare Technologies, FuturesFirst etc. In equities market, about 5% is traded algorithmically and between 15%-25% in futures & options. Much larger volumes can be seen in U.S. Exchanges.
The Indian brokerage market is fragmented and characterized by high competition. Hence brokerage firms are offering DMA services to widen their offerings. Algorithmic trading is usually associated with institutional investors as they deal with larger volumes and speed and capacity become vital issues. Algorithmic trading models determine the optimum size of the next slice and its time of execution based on mathematical models and considering historical and real-time market data.
Selection of algorithm is the most important aspect but traders first need to perform pre-trade analysis to assess suitability of the algorithm for the order and then address micro and macro-level issues. Macro level issues include specification of desired benchmark price and implementation goals. Micro level decisions include specifying any desired deviation rules. In practice the most commonly used algorithms in the market place are: Arrival Price, Time Weighted Aver¬age Price -TWAP-, Volume Weighted Average Price -VWAP-, Market-On-Close -MOC- and Implementation Shortfall. Post-trade analytics are meant to improve execution quality and facilitate the making of investment decisions.
High Frequency Trading aims to take advantage of intra-day opportunities. The time scales are in seconds or even fractions of a second. It is a specialised form of quantitative trading focussed on short-term gains. Quantitative trading systems instigate trades whereas algorithmic trading systems merely execute them.
Algorithmic trading is also gaining momentum in other asset classes beyond equities and can be seen in Derivatives, Fixed Income and Foreign Exchange Markets. In FX markets, algorithmic trading strategies are designed to capture execution strategies in an automated and fragmented market.
Components Of Algorithmic Trading
• Real time and historical market data
• Algorithms to Perform correlation analysis , Identify trading opportunities , Determine optimal timing to launch, Measure trade execution against benchmarks
• Order management/order processing
• Connectivity to liquidity pools
• Integration with internal systems: Trading, Order Management, Risk Management, Compliance, Back Office
Algorithmic Trading Vs Manual Trading
Efficiency:
Algorithmic trading offers better capacity as computers can handle thousands of orders simultaneously. Additional computer servers can be added to increase capacity. Multi-tasking is handled better by algorithms and there is no drop in the quality of execution.
Algorithmic trading is the best option with respect to speed. Automated systems are perfect for monitoring and analysing thousands of variables in a fraction of a second. A trading algorithm can spot an opportunity and send an appropriate order to the exchange before a trader can even notice the quote flickering on his monitor.
Usability:
DMA allows investors to place and manage orders as if they were a broker/dealer and provides direct control. In comparison, both manual and algorithmic trading represent a slight loss of control, since the client can only issue general trade instructions or select an appropriate trading algorithm.
Brokers may not be able to divulge the exact inner workings of an algorithm but they should be able to explain the behaviour for specific orders to enhance transparency.
Another factor that affects usability is changing market conditions because of algorithmic trading and competition between venues. Trading volumes have increased dramatically and so orders need to be split to prevent market impact. Having multiple execution venues fragments liquidity and algorithmic trading is best suited as computerised systems can monitor each venue closely and decide where best to trade for thousands of orders.
As traditional securities experts are replaced by algorithm experts for operation and maintenance of the platform, there are changes in the employment patterns.
Performance:
For specific orders, manual traders outperform algorithms as they can infer more subtle signals. However, algorithms provide better consistency and overall are evenly matched with manual traders. DMA orders should be monitored closely by the trader to add value. Overall, better performance can be achieved by having the trader work the difficult orders and algorithmic trading to be used for more routine orders of a larger volume.
ADVANTAGES offered by Algorithmic trading and DMA
• Increased Liquidity: The market volume itself may not necessarily increase but the number of trades increases. Independent studies have shown that algorithmic trading has a positive impact and increase liquidity. Between June and November, 2009 the number of trades increased 4 times mainly due to the introduction of DMA and Algorithmic trading. When investors want to trade there are always buyers and sellers available. Access to liquidity is improved as algorithms provide Direct Multiple Access -DMA- to the exchanges/markets which offer the most favourable prices and liquidity.
• AT usually reduces spreads and lower spreads indicate more efficient markets.
• There is growing popularity on the buy side for algorithmic trading as they allow traders to handle a larger volume of aggregate order flow.
• Algorithmic trading focuses on minimization of implicit transaction costs in order execution and cut down transaction costs significantly.
• Increase in capacity and speed of computers offers a massive potential for news-based analysis.
• Algorithms reduce errors, offer greater control and more consistency.
• It should also open up newer avenues for trading, expanding the potential of multi-region and multi asset trading.
• AT increases the information of quotes and prices and the revenues to liquidity suppliers also increase with algorithmic trading.
• Increased competition from dark pools pushes all execution venues to compete for retail order flow with superior execution and industry wide price compression among trading destinations.
• Algorithmic trading tools can identify the cause and effect of a strategy and identify repeating market patterns to suggest newer combinations. Algorithmic traders can run thousands of permutations of an algorithm and retain the most effective versions.
• Firms are integrating electronic news into their algorithmic strategies and using live data feeds to reduce reaction times.
• New forms of traders like statistical arbitrageurs are expected to enter which will lead to an explosion in volumes. The liquid options market will benefit due to such statistical arbitrage orders.
AREAS OF CONCERN
Some of the Challenges and hurdles posed by algorithmic trading include
• Market data latency is an issue. The data fed into Algos may not be real time.
• Non-availability of tick-by-tick market data.
• Depth of market information is only 5 levels deep.
• Data which is needed to play back through the test system for the purpose of back testing is not available from the exchanges. The purpose of back-testing is to gauge the effectiveness of the strategy using historical data.
• Indian markets are often sentiment driven and it is very difficult to model for news based on election results, central bank policies, rainfall, inflation etc.
• Algorithms which are successful in global markets may not be easily used in India owing to factors like multiple trading exchanges and multiple listings of most instruments, sentiment of retail investors etc.
• In foreign exchange market, it is difficult to capture data from multiple locations scattered in different time zones. The latency issue is expected to cause concerns as FX market continues to attract high-volume, low-frequency firms.
• The cost of developing, implementing and upgrading algorithms is high and will be suitable only to institutional investors.
• Human traders are more adept at dealing with unexpected news or events and to infer more subtle signals. Too much reliance on software could be dangerous.
• The algorithms have to be submitted to SEBI and brokers are usually reluctant to share some classified information related to algorithms that they have invested in to develop.
Caveats and Demerits
• In the case of high-volatility, Algorithmic performance will suffer as pre-fed data and historical trends are not sufficient.
• Issue of information leakage: If brokers can gather information from client order flow and use it for their own proprietary trading. If the trading pattern of competitor’s algorithm is identified, then algorithms can be created to act on those patterns and exploit for unfair advantage.
• Need for more transparency: The users may be unaware of the intricacies of the algorithm and whether it is executing optimally. They might be reluctant to rely on algorithms.
• AT could impact intraday behaviour and with large orders being broken down it becomes difficult to ascertain the exact impact.
• High- frequency trading has reduced spread among large-cap stocks, but a New York Stock Exchange study suggests that volatility has increased and liquidity has decreased in small to mid-cap stocks
• Retail investors and brokers are worried if AT will fundamentally alter the Market or will replace traders.
Even though these constitute some serious concerns the regulators and exchanges are working together to keep the market more accountable and transparent. All these caveats will be taken care of as regulation in India is very robust. To enlist a few
• BSE has imposed some strong pre-trade risk controls and collateral checks on each order.
• NSE has imposed limits on the number of orders one member can submit per second.
• NSE doesn’t allow multi-exchange algorithmic strategies so cross-exchange arbitrage and smart-order routing cannot take place.
• A Securities Transaction Tax -STT- is levied on all securities sales.
• NSE imposes a penalty for too many unexecuted orders.
CONCLUSION:
1. Trading algorithms are very important for an emerging economy like India where there are huge numbers of small traders. Equity deals are 33 times more than that of Brazil, another developing economy. This underlying growth potential will certainly put Indian exchanges in global league if algorithmic trading is deployed to enhance the efficiency of market.
2. Algorithms are increasing in popularity as investors seek to trade more efficiently and avoid sharp spikes in volatility and yet minimize market impact. Back testing, simulation, and root-cause analysis are needed to learn from past performance and improve the effectiveness of trading strategies in the future.
3. A word of caution must be added as algorithms rely on historical data and rules and may not be optimum for all situations. Hence, role of human mind can never be underestimated as tight spread and high trade entail skill and expertise to execute a good trade. Algorithms can aid the process but they also need to be recalibrated and readjusted to reflect the dynamic market movements.
4. It is essential to integrate algorithms with human interaction, which would increase its dynamism along with impeccable accuracy and unmatched efficiency.
About the author
Abhinay Reddy is a 1st year MBA student at NMIMS, Mumbai. He holds a bachelor’s degree in Electronics and Communication Engineering from Anna University and can be reached at abhinay_2109@yahoo.com
