• Cognizant 20-20 Insights
Trade Surveillance with Big Data The rise of real-time, high-frequency trading has regulatory compliance teams working hard to keep pace with the industry’s widening pools of structured and unstructured data. By employing emerging tools and techniques, capital markets ﬁrms can improve trade surveillance and spot abuse and irregularities before they can do harm. Executive Summary Electronic trading has come a long way since the NASDAQ’s debut in 1971. Today’s fragmented electronic market venues (the result of nontraditional exchanges competing for trades with traditional exchanges) have created socalled “dark pools of liquidity.” Simultaneously, automated and algorithmic trading has become more sophisticated — now enabling individuals and institutions to engage in high-frequency trading (HFT).1 As a result, the number of trades has increased tenfold in the last decade, from 37 million trades in NYSE listed issues in February
This paper highlights some of the key issues faced by regulators and compliance teams. We will also describe how new “big data” solutions can help manage them.
Challenges and Changes on the Trading Landscape While traders march ahead with high-frequency trading and order books that allow them access to liquidity spread across geographies, surveillance and compliance teams cannot seem to catch up. A primary reason is their inability to access and harness huge volumes of data.
2004 to 358 million in February 2014. Traders at capital market ﬁrms have been at the forefront of these advancements — pushing the envelope along the way. How has this impacted trade surveillance and compliance teams? The rise of algorithmic trading, where split-second execution decisions are made by high-performance computers, plus the explosion of trading venues and the exponential growth of structured and unstructured data, are challenging regulatory and compliance teams to rethink their surveillance techniques. Those that depend on individual alerts can no longer meet most ﬁrms’ requirements. We believe that capital markets ﬁrms require a radically new and holistic surveillance approach. cognizant 20-20 insights | october 2014
Let’s look at some of the obstacles faced by the surveillance community today: 1. Capturing and recalling each event in the lifecycle of a trade. Trading ﬁrms need to maintain and recall the end-to-end lifecycles of individual trades. This helps internal compliance teams perform deep-dive analyses and dig into any issues that may arise. It also enables ﬁrms to respond accurately to regulatory investigations when needed. In some instances, this becomes mandatory, as detailed in the following examples:
ﬁrm must be able to provide step-by-step details of the complete lifecycle of any swap
market arbitrage. When the price of a stock changes on one stock exchange, an HFT algorithm picks up orders available on other exchanges before those exchanges have had a chance to react.
or associated transactions — including information on related deals.
> As stated in the article on Financial Indus-
try Regulatory Authority (FINRA) Rule 527 5270 0 (effective from September 3, 2013) no FINRA member broker-dealer shall execute an order to buy or sell a security or a “related “ related ﬁnancial instrument” when that member has material, non-public market information concerning an imminent block transaction in that security,, and that information has not been made rity public or has not become stale or obsolete.3 In the event of an inquiry inquiry,, ﬁrms must quickly recall and reconstruct the entire lifecycle of the block trade. This can become very complicated; a single block order sent via an algorithmic engine can be broken into thousands of smaller orders, and may get routed to multiple execution venues over the course of hours, days, or even weeks.
Today, there are no regulations in place that prevent an HFT algorithm from engaging in techniques such as “electronic front running” or “slow market arbitrage.” The reason? Prior to the emergence of HFT, no one considered such scenarios. Clearly, HFTs have exploited gaps in the regulatory frameworks — giving them just enough lead over the public investors to make substantial gains. In September 2013, the Commodity Futures Trading Commission (CFTC) announced plans to build a regulatory framework around highspeed and algorithmic futures trading. The CFTC Firms that subsequently released a actively use HFT 137-page document6 that have mastered requested public input on various proposed ways the complexity to control the associated of the market technology risks while structure, but enabling more trades to be made faster, and with less institutional human interaction. investors are far
According to ﬁnancial-services and software vendor Sungard, complete and rapid recalls of the details of a trade require: 4
Trade-related data from trade systems and ancillary systems systems..
and telephonic communications data from instant messaging applications, e-mails, phone call logs and transcripts.
from microblogging sites such as Twitter.
Detecting such market behind HFT. manipulation techniques requires real-time surveillance. But given the number of trades executed by human traders and their robotic partners at venues spread across continents, connecting the dots in real time can be daunting. A trader working at multiple venues, for example, could deploy a high-frequency algorithmic system
Firms must also be able to merge structured and unstructured data. In this way, they can recreate every detail of the trade and gain insight into market conditions and any other information that may have inﬂuenced the trader. 2. Curbing market manipulation in HFT. Market manipulation techniques such “quote stufﬁng, stufﬁng,” ” “spooﬁng” and “pump andas dump” are among the top-of-the list items that regulators on both sides of the Atlantic seek to detect and eliminate.
to process trades at exceptional velocities, making those trades impossible to track without equally fast technology.7 3. Assembling the bigger picture for efﬁcient and effective “cross-market and cross-asset surveillance.” At a large ﬁrm, traders have
Firms that actively use HFT have mastered the complexity of the market structure, but institutional investors are far behind HFT. In his new book, “Flash Boys: A Wall Street Revolt,” 5 author Michael Lewis points to some of HFT’s major issues:
access to multiple trading venues; they can view liquidity across all markets. They have a consolidated order book and can access its full depth across different asset classes and venues, including exchanges, ECNs and dark pools of liquidity.
> Electronic front running. This involves using extremely fast computer programs to detect market orders and jump to the front
One question to ask: Do compliance ofﬁcers have a similar view of the market? A trade surveillance application can scan traders’ executions, but it does not take into consideration
of that queue. Front-running results in the entity that placed the original order having to buy at a higher price point. cognizant 20-20 insights
the entire order book that was visible to the trader, or trades made by other players in the market across multiple venues worldwide. Trade sureveillance applications do not track a stock’s price movement against the volume traded for the stock throughout all venues from which the ﬁrm can trade. Nor do they compare a trader’s actions with the positions in related future contracts. According to one publication, compliance teams cannot compare the bid placed on an ECN with the best bid/ bid/ask. ask. They cannot detect spooﬁng if the trader is trading using several venues in different time zones and with different currencies. 8 In a nutshell, compliance ofﬁcers do not have a complete picture of the market. They cannot perform any cross-market analyses with the information accessible to them. They need to have the same view as the trader — a consolidated order book that spans multiple venues. Considering the large volumes of data that result from various venues worldwide and the accelerated speed of trading, it is simply not possible for a human being to detect all possible instances of market abuse. Consequently, future surveillance tools must be programmed to detect any suspicious activities, store very large volumes of data, and analyze that data in real time.
A Regulatory and Industry Roadmap for Overcoming the Hurdles The explosive growth of data over the last few years is taxing the IT infrastructure of many capital markets ﬁrms. Fortunately, there are emerging technologiess that can help these companies better technologie manage and leverage ever-bigger data pools. These tools can enable trading ﬁrms to end data triage and retain useful historical informatio information. n. By building big-data architecture architecture, , IT organizations organizati ons can keepaboth structured and unstructured data in the same repository, and process substantial bits and bytes within acceptabl acceptable e timeframes. This can help them uncover previously inaccessible “pearls” in today’s ever-expanding ocean of data. Big data analytics involves collecting, classifying and analyzing huge volumes of data to derive useful information, which becomes the platform for making logical business decisions (see ﬁgure below). Relational database techniques have proven to be inadequate for processing large quantities of data, and hence cannot be applied to big data sets.9 For today’s capital markets ﬁrms, big data sets can reach multiple petabytes (one petabyte is one quadrillion bits of data).
A Big Data Analytics Reference Architecture Front Office Consolidated Order Book
Different Asset Class & All Relevant Venues Client Orders
Users Compliance Dashboard
Historical Market Data
gh v ume str ct re nstru ure data ta oupled w th sc la le bill e . an l tic ca bi
Data Platformof Data Several Petabytes
(Real-Time Query & Updates)
Real-Time Market Data
Real-Time Analytic Engine
Executive Board Traders Data
Historical Action Data
Near-Term & Real-Time Actions
Phone Calls E-mails
Instant Msg. Twitter
cognizant 20-20 insights
Sales Reps & Traders
To keep processing times tolerable, many organizations facing big-data challenges are counting on new open-source technoloConsidering the gies such as NoSQL (not only large volumes of SQL) and data stores such as Apache Hadoop, Cassandra data that result from and Accumulo.
multiple venues worldwide and the
accelerated speed of trading, it is simply not possible for a human being to detect all possible instances of market abuse.
Planning. Develop a strategy and a solution
roadmap, bearing in mind the strengths and limitations of the business. Also, it is prudent to focus on the three 3 “Vs” of big data applicable to the organization:
> Velocity. The speed at which the data comes in and is stored.
The ﬁgure on the previous page depicts a representative
big-data architecture appropriate for modern-day trade surveillance.
> Volume. The data set size, which can read
A highly scalable in-memory data grid (e.g., SAP’s HANA) can be used to store data feeds and events of interest. Realtime surveillance can thus be enabled through exception exceptionally ally 10 fast open-source analytic tools such as complex event processing (CEP). CEP technologies like Apache Spark, Shark and Mesos put big data to good use by analyzing it in real time, along with other incidents. Meaningful events can also be
The types of structured and unstructured data. petabyte proportions.
Iterative development. Develop a solution to
prioritize issues for the initial phase. This will help the organization create a mechanism to resolve challenges faced along the way. Solutions can then be added to address all other identiﬁed issues as the implementation proceeds.
Data quality is essential. Last but not least,
the quality of the data being captured plays a vital role in big-data solutions. IT organizations must focus on allocating the required resources needed to ensure that the highest-qualit highest-quality y data is captured for proper and meaningful analysis.
recognized recogniz ed and ﬂagged in real time. There are some key guidelines that ﬁrms should bear in mind while formulating big-data strategies for surveillance and compliance. In our experience, superior results can be achieved by keeping note of the following while developing a big data solution:
New types of data sources should be included to gain a complete picture. Real-time news
and data from social media can be helpful in evaluating the circumstances under which a trade was executed.
Regulators and compliance ofﬁcers are aware of the advantages of putting big data to work. The issue is that a big-data solution cannot be implemented by one or two teams in an organization. Company-wide involvement and a commitment and direction from top management are required to initiate this effort. Senior executives must also share their vision with the entire ﬁrm. Open discussions about goals and objectives will help various teams collaborate more effectively.
According to industry experts quoted in an article published in TabbFORUM,11 “next-generation” architectural requirements for surveillance should encompass the following:
Fuse data across the timeline. Historical, new and real-time data must come together to provide a 360-degree view of market activity, sentiments and trading behavior.
Put activity in context. Once validated against historical data, a suspicious activity can turn out to be an aberration or a pattern. The NoSQL database aims to supersede and replace legacy RDBMS systems to deal with the rising data tide.
Real-time analytics is a must-have. Market
Identify issues and take small steps. Clearly
identify the issues that the organizatio organization n wishes to address with a big data solution. Pick only a few in the ﬁrst round of implementation.
cognizant 20-20 insights
data must be analyzed at the speed of automated trading. CEP technologies can provide the real-time analytics needed for modern surveillance systems.
Looking Ahead Firms need to equip their regulatory compliance teams with tools and technologies that can keep pace with robot traders. They also need to manage the torrents of data coming at them in various formats from various venues and media in order to discover and apply actionable intelligence. Using big data solutions, capital markets ﬁrms can:
Have a complete view of historical activities, including highly granular details down to very small time interva intervals. ls.
Enable quick recall, review and analysis of large volumes of historical data from algorithmic trading programs. This provides a better view of the trading patterns needed to uncover anomalies.
Deploy CEP technologies to uncover practices of market abuse that are hard to detect with existing technologies technologies..
Be seen as a proactive player in the eyes of regulators, the market and customers.
Large, global U.S.-based banks are already raising the bar by applying insights distilled from big data. Implementing big data and CEP together allows them to capture data from active feeds and interprett the signals in real time. In addition, they interpre are creating higher forms of business intelligence by bringing together streams from market news, information from social media such as Twitter, exchange data, market quotes, and insights distilled from their own trade executions in real time. By tapping into new tools and data streams, surveillance teams can effectively uncover more — and more complex — patterns of abuse, and move beyond the speciﬁc alerts that today only spot insider trading.
http://en.wikipedia.org/wiki/High-frequency_trading “High-frequency trading (HFT) is a type of algorithmic trading, speciﬁcally the use of sophisticated technologicall tools and computer algorithms to rapidly trade securities. HFT uses proprietary technologica trading strategies strategies carried out by computers to move in and out of positions in seconds or fractions of a second.”
Consolidated Volume in NYSE Listed Issues, 2014. http://www.nyxdata.com/nysedata/ http://www.nyxdata.com/ nysedata/asp/factbook/ asp/factbook/viewer_edition.asp?mode=tables&ke viewer_edition.asp?mode=tables&key=306&category=3 y=306&category=3
Charles S. Gittleman, Russell D. Sacks, Shriram Bhashyam, Michael J. Blankenship & Steven Blau. FINRA Rule 5270 FAQs: Front Running of Block Transactions, 2013. http://www.shearman.com/~/media/Files/NewsInsights/Publications/2013/01/FINRA%20 Rule%205270%20FAQs%20Front%20Running%20of%20Block%20Tran__/Files/View%20 full%20memo%20FINRA%20Rule%205270%20FAQs%20Front%20Runnin__/FileAttachment/ FINRARule5270FAQsFrontRunningofBlockTransactions__.pdf
Big Data – Challenges and Opportunities for the Energy Industry. Sungard, 2013. http://ﬁnancialsystems.sungard.com/~/media/fs/energy/resources/white-papers/Big-Data-ChallengesOpportunities-Energy-Industry.ashx
Flash Boys: A Wall Street Revolt, by Michael Lewis. Published by W. W. Norton & Company; 1st edition. March 31, 2014.
Steve Dew-Jones. Real-Time Surveillance: Mission Impossible? March, 2012. http://www.waterstechnology.com/waters/feature/2163111/real-surveillance-mission-impossible (The paper above can be read after subscribing to the Web site).
Michael O’Brien. Cross-Market surveillance is essential In an era of market fragmentation, 2011. http://www.nasdaqomx.com/digitalAssets/80/80361_75297_cross-marketsurveillancewhitepaper_ ﬁnal.pdf
cognizant 20-20 insights
Big Data – Challenges and Opportunities for the Energy Industry. Sungard, 2013. http://ﬁnancialsystems.sungard.com/~/media/fs/energy/resources/white-papers/Big-DataChallenges-Opportunities-Energy Challenges-Opportun ities-Energy-Industry -Industry.ashx .ashx
Data throughput rates that can range to millions of events or messages per second.
Mark Palmer. Real-Time Big Data and the 11 Principles of Modern Surveillance Systems, 2011. http://tabbforum.com/ opinions/real-time-big-data-and-the real-time-big-data-and-the-11-principles-o -11-principles-of-modern-surveillanc f-modern-surveillance-systems e-systems http://tabbforum.com/opinions/ (Access to the paper requires creation of a free account on the Web site of TabbFORUM).
• • • • • •
http://www.bigdata.com/blog/ http://www.ﬁnra.org/Industry/Regulation/Notices/2012/P197389 http://www.techrepublic.com/topic/big-data/ http://www.cftc.gov/LawRegulation/index.htm http://nosql-database.org/ http://hadoop.apache.org/
About the Author Pritesh Bhushan is Senior Manager — Consulting, in Cognizant’s Business Consulting, focused on the capital markets domain. Pritesh Pri tesh has worked on several projects for large investment banks, providing tech- nology consulting in the areas of compliance and trade surveillance. He has over 12 years of experience in technology consulting for capital markets ﬁrms. Pritesh has an MBA from Indian Institute of Management, Bangalore and a B.Tech from Indian Institute of Technology, Kanpur. He can be reached at [email protected]
. [email protected]
About Cognizant Cognizant (NASDAQ: CTSH) is a leading provider of information technology, consulting, and business process outsourcing services, dedicated to helping the world’s leading companies build stronger businesses. Headquartered in Teaneck, New Jersey (U.S.), Cognizant combines a passion for client satisfaction, technology innovation, deep industry and business process expertise, and a global, collaborative workforce that embodies the future of work. With over 75 development and delivery centers worldwide and approximately 187,400 employees as ofJune 30, 2014, Cognizant is a member of the NASDAQ-100, the S&P 500, the Forbes Global 2000, and the Fortune 500 and is ranked among the top performing and fastest growing companies in the world. Visit us online at www.cognizant.com www.cognizant.com or follow us on Twitter: Cognizant. World Headquarters 500 Frank W. Burr Blvd. Teaneck, NJ 07666 USA
European Headquarters 1 Kingdom Street Paddington Central
India Operations Headquarters #5/535,, Old Mahabalipuram Road #5/535 Okkiyam Pettai, Thoraipakkam
Phone: +1 201 801 0233 Fax: +1 201 801 0243 Toll Free: +1 888 937 3277 Email: [email protected] [email protected]
London W2 6BD Phone: +44 (0) 20 7297 7600 Fax: +44 (0) 20 7121 0102 Email: [email protected] [email protected]
Chennai, 600 096 India Phone: +91 (0) 44 4209 6000 Fax: +91 (0) 44 4209 6060 Email: [email protected]
transmitted in any form or by any Copyright 2014, Cognizant. All rights reserved. No part of this document may be reproduced, stored in a retrieval system, transmitted © means, electronic, mechanical, photocopying, photocopying, recording, or otherwise, without the express written permission from Cognizant. The information contained herein is subject to change without notice. All other trademarks mentioned herein are the property of their respective owners owners..