Market regulator Securities and Exchange Board of India (SEBI) recently came out with an order penalising NSE, India's leading exchange, some of its past officials, and several organisations and individuals for misuse of the NSE's colocation services.
What is the NSE colocation controversy all about, who were the players and what was the modus operandi? Here's an explainer.
Q: What is the NSE colocation controversy all about?
A: It was alleged that certain trading members had faster access to the price feed distributed by the National Stock Exchange to its members. A whistle blower had written to market regulator SEBI in January 2015, explaining the flaws in NSE’s colocation system, and how some brokers had exploited those flaws in collusion with some employees of the NSE.
Q: Before we go into the details, tell us what exactly is a colocation facility
A: In August 2009, NSE introduced the colocation facility, which allowed brokers to place their servers in the data centre of NSE. Of course, there was a charge for this service, and only the big brokers can afford it. The benefit of colocation is that it allows faster access to the price feed (buy/sell quotes) distributed by the stock exchange. That is because the broker’s server sits right next to the stock exchange’s server and so the data has to only travel a very short distance from the exchange server to the broker’s server. This was a big advantage to those trading members engaging in high frequency trading strategies.
Q: What is the time lag in the buy sell quotes reaching a co-located server vis-à-vis a non-co-located server?
A: The difference is in milliseconds, but it can make all the difference between a profitable trade and not being able to catch the trade at all. Because the co-located server gets the quote first, the trade would be executed by the time the broker on a non-co-located server got to see it.
Q: OK, before we get into high-frequency trading, tell us: is there anything illegal about the colocation facility that NSE is providing
A: No, it is not illegal. World over, stock exchanges provide colocation facility to brokers who are willing to pay for it. The NSE has done nothing illegal by offering colocation services. However, the introduction of this facility was not transparent. SEBI usually puts out a discussion paper to seek the views of market participants on major policies like these. In this particular case, SEBI did not take the feedback of market participants before allowing NSE to offer colocation services to its broker members.
The other point is that of fairness. World over, there has been a debate on whether exchanges should discriminate between those who can pay for faster access and those who cannot.
Q: The whole controversy is referred to as the NSE algo trading scam. Is that a right description?
A: Technically, it cannot be called algo trading scam. Algorithmic trading or algo trading is a method of executing a large order using pre-programmed instructions. These instructions could be based on many variables, such as time, price, and volume. In general, it could be a fast strategy or it could be a slow strategy where small slices of the order are sent out to the market over time.
High frequency trade is a type of algo trading characterized by high speeds and high turnover rates. The algo is rapidly shooting buy and sell orders at a speed that cannot be matched by a human. The key factors that drive HFT are highly sophisticated algorithms, fast access to order books and colocation.
Q: What happened in the NSE colocation case?
A: Three months after introducing the colocation facility, NSE started offering tick-by-tick market data. Again this was available for a fee. Each tick constituted a packet of information like a buy/sell order, order cancellation, order modification, or a trade that has happened. The data was exhaustive and only those brokers with sophisticated IT systems could take advantage of it.
Initially, this data feed was distributed through a tech architecture called Transmission Control Protocol/IP. In TCP/IP, the information would be delivered one-by-one, unlike a broadcast, where everyone got the information at the same time. Also, the data was distributed sequentially in the sequence in which trading members logged in to the NSE server.
So the first one to connect to the server with the least load would receive data faster than others.
One broker in particular, OPG Securities is alleged to have gamed the system because it was said to be aware of the workings of the NSE’s IT department.
Q: How was OPG able to game the system?
A: The whistleblower alleged that OPG Securities, with help from some officials in NSE’s IT department was able to able to figure out which server was working better, what time the servers would be started, and would access the least crowded servers.
Brokers availing the colocation facility are given details of the servers and the ports to which they could connect to access the price feeds. Investigations revealed that many brokers used multiple IP addresses to connect to the servers. IP is a unique address that identifies a device on the Internet or a local network.
OPG Securities was alleged to have mapped multiple IPs to a single server such that it would get the first two or even three connections to that server and crowd out other members.
There were allegedly no policies and procedures for allocation/mapping of the IPs of members to the exchange servers, nor was there a Standard Operating Procedure to deal with requests for change in IP mapping to a particular server.
Such requests were left to the discretion of the NSE's Project Support and Management Team
OPG Securities would allegedly regularly connect to the secondary server, commonly known as the backup server, which had the least load. The NSE took no action to prevent this despite having knowledge of this.
Q: Were the allegations of the whistleblower probed?
A: Yes, by the Technical Advisory Committee of SEBI and by Deloitte India, which was appointed by the NSE board.
Q: What were their key findings?
A: The TAC concluded that NSE’s TBT architecture was prone to market abuse thereby compromising market fairness and integrity, in that it provided quicker order dissemination to those who managed to login early. OPG tried to exploit this architecture by not only logging in 1st on select servers but it even tried to crowd out others by occupying 2nd, and 3rd positions on those servers.
The reported also said that it was likely that OPG and some other brokers were given preferential access to back up servers of the NSE TBT system.
The Deloitte report too said that NSE’s TBT architecture the system was prone to manipulation, and that sequential distribution of price information reached brokers logged on to less crowded servers, faster than it reached somebody on a crowded server.
Q: Could the NSE have negated the advantages some brokers had by connecting to least crowded servers?
A: Yes. The advantage of connecting to least crowded servers would have been nullified if the NSE had a load balancer system in place.
There was a prescribed limit of 30 connections for each port of the server from which the NSE disseminated price data to trading members. However, this limit was not followed and the number of connections on one port often exceeded 30. This put members who were on more crowded ports at a disadvantage. NSE’s colocation support and project support management teams were aware of the load on each server.
A load balancer would have distributed network/traffic load across a number of servers based on specific algorithms like least connections, least response time and ensured equitable load distribution across trading members.
Q: Why did the NSE not implement the load balancer?
A: The NSE’s stand was that deployment of a load balancer would have introduced an increase in time lag because the additional hardware device would add an extra step through which the data would need to flow.
Q: Could the NSE have nullified the advantage to brokers logging in early?
A: Yes. It could have removed that advantage by introducing a randomizer. A randomizer, as the name suggests, would randomly pick a connection to begin dissemination of data rather than starting with first connection each time. But the NSE did not implement the system on the servers disseminating Tick by Tick data. This gave an advantage to brokers who logged in first.
Q: Did the NSE plug the loophole that gave an unfair advantage to the broker who logged in first?
A: In April 2014, the exchange introduced the Multicast Protocol transmission, which meant that all brokers would have access to the tick-by-tick price feed simultaneously irrespective of when they logged into the exchange servers.
Q: There was also the issue of some brokers with colocation facility benefitting by logging on to the secondary server of the NSE. How were they able to do that?
A: When a broker’s application for colocation facility was approved, the activation e-mail gave details of the secondary server as well, in addition to those of the primary server. Also, the secondary server was active at all times. There was no system whereby the secondary server would start only when the primary server failed. Since there was no documented policy on connecting to the secondary server, and since trading members had the details, they could log in to the secondary server anytime. Also, NSE did not have a monitoring mechanism to check if the brokers connecting to the secondary sever had a valid reason for doing so.
Q: What was the advantage for brokers connecting to a secondary server to access price feeds?
A: Since the secondary server was to be accessed only when the primary server was not working, few brokers connected to the secondary server. As a result, the load on that server was low, and the few members connected to the secondary server would receive the price feed faster than members connected to the primary server.
Q: If it was so obvious, why did more brokers not access the secondary server?
A: Initially many of them did. Following warnings by the NSE’s IT team, most brokers reverted to the primary server. But some like OPG Securities continued to connect to the secondary server despite warnings. The SEBI report observed that the NSE did not penalize the brokers who were repeatedly flouting NSE’s directive on connecting to the secondary server.
Q: What is SEBI’s conclusion?
A: The regulator has ruled that the stock exchange has failed to ensure a level playing field for the trading members who had subscribed to NSE’s colocation services and were getting the tick-by-tick data.
However, SEBI feels the exchange cannot be charged with fraud as there is no proof of collusion of employees with brokers, or proof of some brokers having been discriminated against, or some NSE officials or brokers having gained financially because of the lapses. Also, SEBI feels failure to have a randomizer or load balancer cannot be seen as a breach of the principle of fairness and equity.
So NSE cannot be held guilty under the provisions of fraudulent and unfair trade practices.
At the same time, the exchange did not exercise the requisite due diligence while putting in place the TBT architecture. Also, the exchange’s policy on retention of electronic records were weak. For the above failings, the exchange has been ordered to disgorge a part of the profits made from the tick by tick data dissemination between 2010-11 and 2013-14. That works out to Rs 625 crore along with an interest of 12 percent from April 2014.
Also, NSE has been barred from accessing the securities market for six months
Q: Were any of the other NSE officials penalized?
A: SEBI ruled that Ravi Narain and Chitra Ramakrishna, who were MD and CEO of NSE in succession during this period, failed in their responsibility to ensure equal, fair and transparent access. Narain has been asked to disgorge 25 percent of the salary drawn from FY 2010-11 to 2012-13 and Ramakrishna has been asked to disgorge 25 percent of her salary drawn for FY2013-14.