On 5 August 2016, the Securities and Exchange Board of India (SEBI) released a discussion paper on the ‘Strengthening of the Regulatory framework for Algorithmic Trading and Co-location’. This is yet another step in the global movement towards increased control on algorithmic trading. In the paper, SEBI proposes several suggestions for comment by all stakeholders on potential new measures.
Algorithmic trading is the use of computers to make and execute the trading decisions based on imputed strategies. A subset of this is High Frequency Trading (HFT) which employs technologies, such as colocation, which involves placing servers as close as possible to the exchanges, and high speed networks in order to trade at very high speeds and take advantage of very small windows of opportunity. In fact, HFT can take advantage of opportunities that may only be open for a fraction of a second. It is no wonder, then, that there is an outcry for regulators to level the playing field between HFT firms and other market users. With such an edge, it is clear why many are pushing for regulations which will “allay the fear and concern of unfair and inequitable access to the trading system of exchanges.” According to the SEBI discussion paper, algorithmic trading comprises about 40% of the trades on the exchanges. Although HFT has led to the tightening of spreads, improvement of liquidity and increased speed of assimilation of information into market prices, there have been concerns, not only that HFT causes unfair market access for non-HFT traders, but that it could also increase the prevalence of ‘flash crashes’.
For these reasons, SEBI has proposed the following potential mechanisms for controlling the unfair advantages of HFT. Firstly, minimum resting time for orders has been suggested. According to the document, ‘resting time’ is considered to be “the time between an order is received by the exchange and the said order is allowed to be amended or cancelled thereafter”. This is intended to solve the issue of the ‘fleeting’ liquidity which is caused by HFT-traders’ ability to modify their orders in response to new information, and something that currently occurs in large volumes. If the mechanism were to go ahead, there would be a defined amount of time required to elapse before an order could be cancelled or modified. SEBI does note that no regulator currently employs a ‘resting time’ mechanism, although it has been proposed and rejected in other countries in the past, which could suggest one reason to reflect carefully on the decision to implement such a practice.
Secondly, SEBI has proposed frequent batch auctions, in terms of order matching. Currently in place on the exchanges is a system of ‘continuous matching’ in which buy and sell orders are constantly matched together. In efforts to curb the edge that HFT traders have in terms of latency, SEBI proposes a ‘batch auction’ system in which orders would accumulate for a specified period of time, before being matched. Latency is the time required for data to travel between its source and its destination and can pose either significant advantages or disadvantages for trading – a market participant that uses colocation can trade multiple times in the time that it can take for a non-colocation participant to make a single trade. At moments of particularly volatile market activity, the non-colocation participant may not even receive current price quotes. This proposal is aimed at addressing such a disparity.
The third proposal is random delays, or speed bumps, which would hopefully deter the use of latency-sensitive strategies, which will inevitably give the algorithmic trader an advantage. Globally, there has been movement in this area to attempt to slightly hobble algorithmic traders, without hindering or discouraging non-algorithmic traders, and the hope is that these ‘speed bumps’ will achieve just that.
Furthermore, a revision of the order queues has also been proposed; the ‘randomisation of orders received during a period’, such as a period of 1-2 seconds. This is yet another way of nullifying the latency advantage available to co-located traders. Because participants using colocation can produce more trades higher in the queue than non-colocation participants, if the queue were randomised, then market access would become less preferential.
SEBI also want to implement a maximum order message-to-trade ratio requirement. This would mean that, for a specified number of orders, a market participant would be required to execute at least one of those trades. SEBI notes that the UK reviewed this option and exposed some potential worries, such as an increase in bid-ask spreads and an exacerbation of liquidity withdrawal during volatile moments. Despite these limitations, there are also several advantages, such as the increased chance that a viewed quote will be available to trade and a reduction in ‘hyper-active order book participation’, which is currently being discussed as a problematic aspect of algorithmic trading.
The discussion paper also explores the possibility of having two separate order queues. By having separate queues for colocated and non-colocated orders, SEBI hopes to allow for a fair chance of execution and avoid non-colocated orders being ‘crowded-out’ by colocated orders. The mechanism would function by alternately taking an order from each queue, offering a fair chance to non-algorithmic traders. The subject is contentious and has already been discussed with SEBI’s Technical Advisory Committee (TAC) and deliberated in an earlier consultation paper. The discussion paper notes that colocated market participants would still be able to very quickly react to market data as they retain their proximity to the trading platforms and their high-speed systems.
Finally, a review of Tick-by-Tick (TBT) data feeds has been suggested. TBT provides real-time information about orders and trades. It is a tool paid for by – particularly – HFT-traders who use it in addition to colocation to theoretically recreate the order-book in order to forecast the impact of their execution. It is noted by SEBI that most smaller market participants do not utilise it, as it requires an additional fee and is very data-heavy. Many people have argued that it allows for an unfair advantage. In order to address this disparity, SEBI proposes for ‘structured data’ to be available to every market participant as either a real-time feed or at prescribed time intervals. The feed would contain the top 20/top 30/top 50 bids/asks and market depth, amongst other data. This should create more parity in the information available to market participants.
The aim of the entire proposal is to increase market fairness by ensuring that, regardless of financial or technological strength, all participants engage on a level playing field. There are those, however, that see the measures as going too far. One of the most critical concerns is the damage that such measures could pose for liquidity, with the potential further detrimental effect of deterring foreign investors who may view the measures as too restrictive and damaging for liquidity. It will also be important to discuss and predict how effective the measures would be, and this will be part of the next steps as SEBI receives public comments regarding the proposal.
In a global climate that is calling for restrictions on HFT to limit the disparity between algorithmic and non-algorithmic trading without damaging liquidity, SEBI’s proposal is an important one and, going forwards, many eyes will be fixed on the discussions and actions in India to see just how effective they are at satisfying the calls for market fairness.