Nasdaq’s Machine Learning Hub Cuts Through Unstructured Data Noise Using iSentium Data

Turning Big Data into Small Data – Nasdaq, the world’s first electronic exchange, has ‘fintech’ running through its veins, and today introduces its latest innovation to support growing market demand for unstructured data sets.

The Nasdaq Analytics Hub will pull in data from Nasdaq’s own feeds as well as from a number of financial technology start-ups such as iSentium, which specializes in collecting and crystallizing massive unstructured social media content, into market intelligence; in other words, to turn Big Data into Small Data, and Small Data into trading and investment signals.

DATA

Four data sets are being introduced as the initial offering within the Hub. In addition to social conversations, it will take in central bank communications scored and profiled from the G-10, retail investor sentiment collected as investors make decisions to form strong short- and long-signals, and Nasdaq Dorsey Wright technical analysis that ranks securities, ETFs, and Mutual Funds based on relative strength.

The Analytics Hub is one of multiple initiatives coming out of Nasdaq’s Innovation Lab, a virtual workshop led by a select group of Nasdaq technologists and data scientists. The lab leverages cutting edge, scalable technology to provide Nasdaq’s clients with comprehensive multi-asset class market data solutions. Nasdaq Trading Insights was the first product to debut from the Innovation Lab last November.

New ways to find alpha-generating insights – Within the Analytics Hub, Nasdaq gathers normalized data from iSentium and other partners, before validating it and using machine intelligence to check that each data set possesses potential alpha-generating capabilities. Each data set is then thoroughly back-tested to ensure it stands up to rigorous testing in the field.

“We look at social sentiment data – that iSentium provides – as well as retail investor sentiment data and macro sentiment data. We also have our own proprietary technical analysis and machine learning-based events to draw upon. And, we are currently looking at a whole host of new data sets that we believe will be valuable to the marketplace,” says Terry Wade, senior vice president and head of business development and product, Global Information Services at Nasdaq.

He adds that further down the road, the Nasdaq Analytics Hub will have hundreds of data sets for clients to utilize:

“Our goal is to make it easier for the market (fund managers, active traders, quant funds etc.) to access valuable information in ways that would otherwise be difficult to do on their own.”

Social media sentiment from iSentium is one of the first four data sets available in the platform. Its team, which is led by Founder and CEO, Gautham Sastri, consists of linguists, quants and computer scientists whose primary mission is to find meaning or sentiment, in the terabytes of data that are produced on a daily basis.

Social data spans a wide spectrum. Facebook can be thought of as a cosy environment in which to share personal stories and events with one’s network, LinkedIn is the place for professionals to share business-to-business content, while Twitter is akin to someone shouting from the rooftops; a kaleidoscopic flow of 140-character data.

“When you arm millions of people with mobile devices and give them infinite bandwidth at low cost, they start messaging and tweeting, and it creates vast unstructured content. It represents the wisdom of the crowd, and is not machine-generated, so making the meaning obvious is valuable” says Sastri, who confirms that the Brexit decision started to become clear to his team some eight days before the final result was announced.

There is, says Wade, a desire to take people’s thoughts, emotions and actions and find signals and value in this vast unstructured content.

“The financial industry traditionally consumes oceans of high quality market data, so we thought, ‘How can we provide our clients with more of an edge in the market?’ How can we take it beyond what structured data offers and provide valuable information to people, to implement into their models and trading strategies?

“That’s what led to the development of the analytics hub,” explains Wade.

This underscores Nasdaq’s willingness to not only move in tandem with market evolutionary trends but to be at the vanguard of such trends, of which unstructured data consumption is just the latest example.

In Wade’s view, there will be increasing automation throughout the buy-side community, including increasing automation in research functions and trade selection. “To allow people to do that in an efficient way, we need to provide them with vast sets of accessible, cutting edge data,” he says.

Each of those data sets can then be confidently used and applied to clients’ trading strategies as a way to seek out new ideas and optimize entry and exit points.

Before determining whether to buy a specific stock, why not look at the market sentiment? If it looks favorable, it might be a good time to execute but if sentiment is trending down, maybe it would be best to hold off.

“One of the interesting things we’ve found is that while most people think of Twitter as a short-term signal, the bulk of work that both iSentium and Nasdaq have done, shows that these signals actually persist much longer than a minute or an hour,” reveals Wade.

Can you hear me…? While a stock exchange like Nasdaq generates very significant amounts of structured data per day, Twitter alone generates exponentially more: a terabyte of data every four seconds. People want to be heard in today’s world.

Still, that is a lot of noise to scan through and begin to make sense of.

“I don’t view stock market data as being competitive with what we do, which is to provide market intelligence. We are trying to add more color to the existing systems that deliver information, rather than replace them.

“The problem that we solve is the ability to process all this flow of social sentiment, which remember, doesn’t necessarily have to be rational,” says Sastri.

In effect, iSentium can best be thought of as a reporting business. It isn’t a quantitative business that provides people with a moneymaking solution, just as using a Nasdaq market data feed does not guarantee anyone will make money.

It all boils down to intelligence.

People use Nasdaq data feeds because they provide a reliable, accurate representation of the market at any given time.

“In our case, we look at a huge volume of social sentiment data, processing over 2000 messages per second with near zero latency to build patterns which show proven correlations to market performance. It’s an extraordinarily complex issue,” says Sastri.

Nasdaq receives iSentium’s processed data and then validates it. “We make sure the data is complete, we put it through our internal algorithms to back-test it using historical event data, and then we wrap it and send it to our clients. We do this for each and every data file we receive from our partners,” explains Wade.

The data–either raw or refined depending on the client’s preference—can be used across asset classes but it is mostly used for equities.

“Another of our partners, Prattle, mines central bank communications, and while this information is directly useful to many fixed income and global macro managers, we’ve taken that signal and shown clients how they might trade equities based on it as well,” confirms Wade.

The creation of the Nasdaq Analytics Hub to handle unstructured, non-traditional data sources shows just how fast the world is evolving. Sentiment data can reveal fear or optimism in the markets and, given it has been forensically analysed and processed, provide an important tool in refining one’s trading strategy.

Wade believes that people are going to want to consume ever more data to do their jobs better in a more automated way, “And we think we can extend our leadership position in the delivery of data to include this vast marketplace of unstructured, and unique, data..

“We’ve always been a strong technology company and we will continue to build on that heritage.

Link to original content can be found here:  http://business.nasdaq.com/marketinsite/2017/Nasdaqs-New-Machine-Learning-Hub-Cuts-Through-Unstructured-Data-Noise.html

About iSentium

Founded in 2008, iSentium uses patented Natural Language Processing (NLP) to extract sentiment from unstructured social content then instantly transforms it into highly actionable indicators in Finance, Brand Management and Politics. iSentium is proven in the market and is vetted by J.P. Morgan and NASDAQ.

iSentium’s management team brings a wealth of experience from both industry and academia with skills ranging from Big Data, finance, linguistics and signal processing. Our world-class team comprised of linguists, quants and computer scientists has collectively published over 200 papers and 18 books.

Knowledge is power. Knowledge is alpha. Knowledge is edge. iSentiumKnows.™

To learn more about what iSentium can do for you, call Jake Sedlock, Chief Revenue Officer, at (206) 683-6003. Or send him an email at: Jake.Sedlock@iSentium.com.

Author: isentium

Founded in 2008, iSentium extracts sentiment from unstructured social content and transforms it into highly actionable indicators in Finance, Brand Management and Politics. Knowledge is Power. Knowledge is alpha. Knowledge is edge. iSentium Knows.™

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s