Sebi mulls overhaul of algo trading rules

Capital markets regulator Sebi is looking at a number of measures to strengthen algo trading framework, including by mandating the exchanges to offer shared co-location facilities and providing some services for free.

Sebi has also proposed a review of trading requirement for algo software for strengthening the algorithmic trading framework by mandating stock exchanges to provide a simulated market environment for testing of software used for such high-frequency trades.

The proposed move is expected to be discussed later this month by Sebi's board, along with a number of important matters including amendments to rules on angel funds, mutual funds, buyback of shares, takeovers, registrars and bankers to issues, suspension and revocation of trading and distribution of cash benefits by listed companies.

Sources said other important matters to be discussed by Sebi include new regulations for fiduciaries in securities markets and a new framework of compliance to Sebi rules by listed entities under insolvency resolution.

Regarding algo trade, which has been under public glare because of an ongoing probe concerning leading stock exchange NSE, an official said various measures are being proposed to address the concerns relating to market quality, market integrity and fairness on account of usage of algo trading and co-location.

Algo trading includes automated rule-based trading where decision making is delegated to a computer model, while High Frequency Trading (HFT) is a type of algo which is latency sensitive and is characterised by high daily portfolio turnover and high 0rder-to-trade ratio (OTR).

Co-location is a facility provided by exchanges to trading members and data vendors whereby their trading or data systems are allowed to be located within or at close proximity to the premises of the bourses.

This facility enables the co-located entities to access the trade/order related data before other non-co-located entities. It also enables co-located members to minimise the time for sending orders to the trading system of the exchange.

Sebi has taken various steps in the past to frame regulatory guidelines for algo trading, which has been continuing to attract the attention of investors and regulators across the world. To address concerns, including those regarding unequal access, Sebi in August 2016 issued a discussion paper to introduce some further measures and also got a simulation exercise done.

Now, the regulator is proposing fresh measures to address the concerns related to algo trade and co-location to make the market more fair, equitable and transparent, a senior official said.

Sebi has also got representations that many small and medium sized members of exchanges find it difficult to avail the co-location facility services due to high cost.

To address this issue, Sebi has proposed that exchanges may provide these facilities as a shared service in the co-location facility, as opposed to each member setting up a server and individually incurring charges. The proposed move has potential to reduce the co-location access cost substantially by more than 90 per cent.

Besides, lower cost would help more number of trading members to avail the facility, thus reducing the latency for accessing the trading system.

Sources said another proposal calls for providing tick-by-tick data feed free of charge to all trading members.

It is also proposed that algo orders placed within 0.75 per cent on either side of the last traded price may be exempted from the framework for imposing penalty for high OTR.

Further, the OTR framework may be extended to orders placed in the equity cash segment and orders placed under liquidity enhancement scheme.

Also, the exchanges may be asked to allot a unique identifier for each algo. Presently, one specific code is attached to all algo orders to distinguish them from non-algo trades.

Sebi may also ask the exchanges to publish additional details regarding the latency observed within the exchange trading infrastructure and also publish a reference latency.