The bulk of of trading on stock markets is done algorithmically, by computerised clerks working at the behest of human traders.
Most trading on modern exchanges, worldwide, is done by computerised clerks working at the behest of human traders.
Like every new technology, this has induced gainers and losers.
The statistical analysis of data has generally shown that algorithmic trading has benefited market quality. Algorithmic trading calls for careful thinking by regulators in terms of the impact on their functions such as enforcement against market abuse.
Algorithmic and high-frequency trading have brought about large changes in financial markets. Rigorous analysis is required to identify areas of potential market failure, as well as possible corrective interventions.
At the IGIDR Finance Research Group, we used trades and orders data from the National Stock Exchange to answer five commonly asked questions about algorithmic trading:
How much of the trading activity in India has gone algorithmic?
The share of algorithmic trading started rising in the equity markets in January 2010, with the introduction of co-location at exchanges.
At the time, algorithmic trading was around 20 per cent of traded volumes.
It became 40 per cent in 2011 and 60 per cent by 2013, and has remained between 60 per cent and 75 per cent since then.
How does algorithmic trading affect market quality?
There is ample academic evidence of the impact of algorithmic trading on measures of market quality such as liquidity and sudden large crashes. The core purpose of financial markets is liquidity. Higher algorithmic trading brings higher market liquidity. Liquid stocks are less easy to manipulate, which helps make the market safer for small and vulnerable investors.
The nature of market liquidity is that large stocks tend to be liquid, while small and young stocks tend to be illiquid.
Our research finds that algorithmic trading improves liquidity, not just for large stocks, but also for small stocks. Small stocks with higher algorithmic trades have better liquidity compared to small stocks with lower algorithmic trades.
Intuitively, what may be at work is that a financial firm cannot justify putting a human trader to work on the shares of a small company, but once there is more automation, trading activity extends to smaller stocks to a greater extent.
Looking ahead, the deepening knowledge about algorithmic trading in Indian financial firms and exchanges is likely to lead to a further percolation of liquidity into smaller securities.
Does algorithmic trading take away more liquidity than it provides?
In our analysis of the data from 2013 to 2016, we see that algorithmic traders are on both sides, supplying and demanding liquidity in equity spot and equity futures. With equity options, we see algorithmic trading as net-buying liquidity from human traders.
Some people fear that algorithmic traders flee the market when there is stress.
Our analysis shows that these patterns hold in periods of stress also. For example, a closer examination of trades during the Emkay crash of 2010 shows that algorithmic traders continued to supply liquidity in the market, even after the crash.
Do algorithmic traders mislead the market?
A concern is that high-frequency trading can generate misleading signals about current prices and liquidity. One example is fleeting orders. These are orders that appear inside the touch (the best price to buy and sell in the market at any time), but which can disappear in a very short period. A high percentage of fleeting orders in the limit order book could persistently mislead other traders.
We find that less than eight per cent of orders placed at the touch in are cancelled within a second of its arrival. This is a small percentage of fleeting orders.
Most orders, which are cancelled within a second, are usually at prices far away from the touch. Algorithmic traders do work rapidly and change their mind often, but this is not happening to a great extent at the quoted prices.
What should policymakers do?
Algorithmic trading has changed the range of features that exchanges can offer. Some exchanges are creating facilities more supportive to algorithmic trading and vice versa. As an example, an exchange in Canada (TSX Alpha) introduced randomised delays in orders. They felt this would attract business from people who fear trading against the algorithms. This has resulted in higher "quote fades" and lower aggregate market liquidity. Such experimentation is legitimate and valuable.
Regulations, on the other hand, need to be carefully nuanced because they carry the power of the state and impact all exchanges. There is a need for careful scientific work before using the power of the state. The US Securities and Exchange Commission and UK Financial Services Authority have created regulatory sandboxes to test corrective mechanisms before deploying them as regulations.
In Australia, the SEC has done thorough scientific studies, and has fed this knowledge into improving surveillance procedures at the SEC.
In August 2016, the Securities and Exchange Board of India proposed new regulations for algorithmic trading with seven interventions that have been individually proposed by exchanges or regulators elsewhere in the world.
Sebi’s proposal did not demonstrate the presence of a market failure or evidence about the costs and benefits of the seven kinds of interventions. If interventions are proposed, it would be wise to bring them about in a small, controlled manner, to measure the impact, before pushing them out on a market-wide scale.
The most important change required is in market surveillance. In the algorithmic world, an episode of market abuse can start and finish in 10 seconds. Sebi and the exchanges need a new level of scientific firepower in detecting and enforcing against these.
The bulk of algorithmic activity in India is on the equity market. These benefits will percolate to commodity futures and the bond-currency-derivatives nexus, as the knowledge spreads from one part of the Indian financial system to all other parts. It is likely that greater algorithmic trading will assist India's goal of achieving a more liquid and efficient financial markets system.
Nidhi Aggarwal is a research consultant with the Finance Research Group, Indira Gandhi Institute of Development Research (IGIDR). Susan Thomas is faculty at IGIDR.
