The Gateway to Algorithmic and Automated Trading

Plumbing or Trading: The financial technology maze

Published in Automated Trader Magazine Issue 25 Q2 2012

As traders find themselves exploring new markets, time-frames, data types and geographies in search of alpha, an important question arises. Do they want to manage the integration and maintenance of all these variables, or just trade? Mark Pesonen, CEO, Enterprise Products & Solutions at Bloomberg explains why the latter is the right answer, and also how it can be accomplished.

Markets are constantly changing. Not just in terms of price fluctuations, but also structurally. The dynamics of trading have fundamentally shifted in recent years due to the convergence of three areas:

  • Markets - The move to electronic trading platforms has meant that more trading is now conducted between machines rather than humans. This is due to commercial and performance pressures and in some cases regulation.
  • Technology - Technology used to be developed by the government or private sector before entering the consumer sector. That flow is starting to reverse. The convergence of telephony and data, mobility and the development of standards for inter-connectivity/communication is one example. The consolidation of software platforms has made interoperability easier while the cost and availability of network bandwidth has improved. These changes have made previously non-viable trading models feasible.
  • The Economy - The financial services industry is facing decreasing margins and challenging markets. It needs to find efficiencies wherever possible. In the past, better margins meant that firms were willing to absorb significant infrastructure costs. The objective today is to allocate scarce dollars towards gaining a competitive edge and wherever possible outsource generic infrastructure and operations.

Breaking down silos

Collectively these three factors represent a massive challenge to the way technology and services are delivered and maintained. Organisations have historically functioned on the basis of siloed individual businesses with technology and services kept separate. Independent data repositories were often created horizontally to support several businesses.

This model doesn't work anymore. Global markets strategies fit poorly within regional frameworks, and cross asset strategies collide with functional business units. Segregated information becomes problematic as trading models and risk platforms grow in complexity, requiring an ever-greater amount and variety of data. In a world of decreasing margins, the cost of re-engineering and ongoing maintenance becomes prohibitively expensive to defend.

This challenge is the driving force behind Bloomberg Enterprise Products & Solutions. As an organisation serving over 300,000 users globally across multiple instruments, regions and time horizons, Bloomberg has learned how to effectively and transparently navigate these issues each day. Now, EP&S takes this one step further, directly supporting organisations to redefine and strengthen their core technology value proposition.

The information continuum

The Bloomberg data universe does not discriminate between news, pricing, or other feeds. A continuum of data flows through the system at blindingly low-latency, providing clients the tools and opportunity to piece together the parts of the puzzle into their own market narrative.

Segmenting, merging, analysing and acting on these unique data streams is made simple through Bloomberg's state-of-the-art operations and technology, along with the world's largest private network; supporting varying demands transparently to the end user. Data formats and storage are no longer of concern-Bloomberg manages the hardware, software, services, networks and operations.

Mark Pesonen

Mark Pesonen

Expanding demographic

It is no longer sufficient to continue doing what has worked in the past. The past no longer works, trading strategies become stale, and the expansion of algorithmic trading provides downward pressure on this trend. Smaller firms, historically maintaining their own data and infrastructure, realise that the modest requirements of the past no longer cut it in this brave new world.

Traders leaving bulge bracket firms to start their own funds find that they have not the resources, expertise, or interest to install and maintain the plumbing. They prefer to focus on trading, and outsource the rest.

As alpha opportunities become increasingly ephemeral, the need to be actively present across a wide number of markets is critical. The number of organisations capable of continually delivering this rapid deployment with internal resources is small and declining.

Gaining an edge through event-driven trading feeds

This holistic approach to data and services also has the advantage of begetting new tools for alpha capture. Bloomberg's news platform has historically been limited to display applications. Every news function on the terminal-search, alerting, TOP, news charting, etc-is crafted to facilitate the reading of news by humans. Traders open, read, interpret, and react to the news (see Figure 1). It's a core part of the Bloomberg professional service, and a critical part of most financial professionals' workflows.

Bloomberg News consistently moves markets, providing unique edge and insight into the world of finance, and a benchmark for the industry. As such, Bloomberg's news distribution has been constrained to the terminal, providing incredible and lasting value to the world's most powerful financial professionals. The proprietary content was not made available for consumption outside the terminal, and was purposefully limited and delayed to ensure professional trading clients were able to extract the most value from breaking releases.

Now, the entire universe of Bloomberg's content, along with its next-generation low-latency distribution mechanism is available as a real-time, machine-readable, meta-data-rich feed, allowing clients to leverage the power of their strategic value proposition with the strength of Bloomberg's editorial supremacy, marrying operational efficiency with cutting-edge performance and alpha opportunities.

A paradigm shift

Bloomberg's Event-Driven Trading Feed has been serving real-time applications consuming machine-readable news data since 2008. From black box trading, to risk management platforms, to other back-office functions, the feed provides a comprehensive solution for firms looking to incorporate news and events data into their trading applications faster than ever before.

Figure 1: {MMN<GO>} – Market-moving events for global equities. MIPS US, example of M&A exclusive.
Figure 1: {MMN} - Market-moving events for global equities. MIPS US, example of M&A exclusive.

Bloomberg's Event-Driven Trading Feed consists of the following modules:

Low-latency Machine-Readable News Content: access to all of Bloomberg News' unique and exclusive market-moving stories. Fast, comprehensive coverage of global markets, press releases, regulatory filings, premium third-party wires, and scraped web/social media content from over 80,000 providers globally.

Rich Metadata: All content is tagged with tickers, people, and topics for the easy translation of business logic into trading signals in over 30 different languages. Thousands of topic codes arranged taxonomically allow for structural dependencies to be made explicit, and for the creation of advanced filters and triggers. Bloomberg's strong editorial process, coupled with a client-driven approach, consistently leads to the creation of added value through a fast, high-precision story classification system.

Historical Archives: Years of historical data is provided at no additional cost for the back-testing of models. Bloomberg recognises that a large variety of strategies involve time horizons not in the millisecond range. Clients looking for predictive signals can rely on Bloomberg historical news and analytics archives to effectively back test their models and find alpha.

News Analytics: a comprehensive suite of powerful news analytics across Bloomberg's entire content spectrum, leveraging collective insights from Bloomberg's network of over 300,000 professional terminals in five key areas:

  • Sentiment: real-time, predictive measures of the expected effect the information conveyed in a story will have on the market for a particular security. Bloomberg's sentiment models provide normalised scores and confidence levels from a set of highly tuned proprietary machine learning algorithms. Scores are provided both on a story level by ticker, as well as on an aggregated basis by security over a trailing eight hour window. The scoring methods are competitive with expert human judgements, and are provided in a matter of 20-50 milliseconds from the story's time of arrival.
  • Readership: real-time measures of the demand for information on securities and individual news stories. Bloomberg's readership score leverages the aggregated activity of the Bloomberg Professional Service users to provide a pertinent treasure-trove of information, revealing what the most important news is at a particular point in time.
  • Publication Flow: real-time measures of supply and velocity of data produced about a security. The publication flow indicator provides timely alerts to highlight the events and topics people are writing about right now, and how publication flow changes over time.
  • Volatility & Impact: real time measure of the likelihood that a story will affect asset prices. Bloomberg's unique market volatility and impact product leverages proprietary data-sets to perform large-scale regression analysis and tick-by-tick prediction of the effect a particular news item will have. Bloomberg's custom machine learning algorithms applied across tens of thousands of securities and historical data samples work together to produce normalised impact scores. Coupled with years of historical data, the Volatility and Impact analysis product (dubbed "Market-moving News", see Figure 2) provides invaluable insight into potential market reactions before anyone has a chance to act on the news item (10's of milliseconds).
  • Novelty: real-time measure of the likelihood that a particular story is introducing new information to the market. Providing large-scale similarity analysis of news stories within a rolling time window, novelty scores give traders the ability to tune in to what's important within their news universe.
Figure 2: Historical market-moving event data at your finger tips.
Figure 2: Historical market-moving event data at your finger tips.

Global Economic Statistics & Central Bank Announcements: lowest latency access to global economic indicators from government lockups, central bank, treasury, and finance ministry websites, and other contributors.

Structured Data & Fundamentals: low-latency access to structured corporate earnings, guidance, industry-specific figures, analyst rating changes, and corporate actions.

Practical Applications

Bloomberg's Event-Driven Trading feed offers the potential to easily construct advanced data-driven strategies incorporating real-time market signals with relevant topic, pertinent entities, and rich linguistic metadata to find alpha-whatever your market, instrument, and time horizon. Nonetheless, even simple strategies using Bloomberg's unique content and attributes can have remarkable results.

Defensive Market Making

Equity, equity derivatives, and FX market makers can leverage news for defensive strategies. The need to widen spreads or stop trading when new information on a symbol emerges is important in order to avoid mispricing. A simple to implement yet powerful strategy is to leverage Bloomberg's speed desk operations to track potentially market moving events. Bloomberg News is tracking company press releases, regulatory filings, live events, and other news organisations and distilling the breaking news in the form of flash headlines. These represent timely information happening right now. A simple strategy is to take action whenever a headline is published that is tagged to a security being traded (see Figure 3).

Figure 3
Figure 3
Figure 4: {MIPS US Equity QR <GO>} - Notice the 3 second delay between the timestamp of the headline and the volume spike? Human-introduced delay.
Figure 4: {MIPS US Equity QR} - Notice the 3 second delay between the timestamp of the headline and the volume spike? Human-introduced delay.

Ahead of the Crowd

Human traders have physical limitations in their ability to react to unexpected news events. It takes the human brain about a second to read and interpret the news, another second to key in an order, plus additional time for execution. This adds approximately 2.5 seconds of latency (see Figure 4) assuming that all other variables are correct-no bandwidth or CPU issues, no complex filtering. You can use QRto view events and tick data that support this analysis:


BN Apr 12 2012 14:02:53.271 EST

Instead, a simple strategy leveraging Bloomberg News' strict style guide and our real-time sentiment and impact indicators can place you far ahead of the curve. Using the query "SAID TO" IN HEADLINES AND TOPIC:HEADS AND TOPIC:MNA, we can isolate breaking exclusive Bloomberg M&A headlines. For headlines where a single traded security is listed, it becomes a simple buy indicator. In more complicated examples, sentiment can be leveraged to strengthen the trading signals (see Figure 5).

Figure 5: Sample story-level sentiment message.
Figure 5: Sample story-level sentiment message.


A new trading environment requires a new approach. A trader cannot keep using a trading model that has gone stale and is losing money-and the same applies to the business infrastructure model. Anything that does not deliver proprietary edge should be outsourced to a suitable provider. The Bloomberg Event Driven Trading Feed provides a unique edge, and the peace of mind that your infrastructure is secure and free of maintenance costs. Bloomberg provides a complete solution to the financial data challenge, built on top of the world's most trusted sources for financial information, and augmented with state-of-the-art analytical tools; leaving you to pursue your proprietary value proposition to its full potential.

Bloomberg Enterprise Products & Solutions is a complete package of enterprise-class managed services, applications and data that enables firms to leverage the same data and technology that supports the Bloomberg Professional service for its internal applications and processes. We are uniquely equipped to meet the data acquisition, management and distribution needs of global firms across a broad range of asset classes. With Bloomberg, you leverage far more than data; you leverage our entire operational infrastructure - proven solutions designed to withstand today's volatile markets.