Big Data and
Summary of the Latest Research and Trends
Bigger data and more intelligent algorithms are being processed and analyzed faster in an API-enabled, open source environment. J.P. Morgan is committed to understanding how this technology-driven landscape could differentiate your stock, sector, portfolio, and asset class strategies.
Here, J.P. Morgan summarizes key research in machine learning, big data and artificial intelligence, highlighting exciting trends that impact the financial community.
Featured animation powered by machine learning
Scroll to view the reports
Investment Possibilities with Artificial Intelligence
AI adoption at an inflection point
In practice, Artificial Intelligence is a group of technologies that help facilitate the discovery and analysis of information for the purpose of making predictions and recommendations, support decision making, facilitate interactions, and automate certain responses. Since AI applications are continually transforming business models, the scope of traditional technology applications will scale up towards a multi-channel world with recommendation systems, virtual assistants, chatbots, and AI-managed marketing platforms.
Essentially a model that helps identify patterns and associations from large amounts of data, Artificial Intelligence enhances quality control and improves operational effectiveness through digitized information assets. This allows businesses to focus time and resources on identifying new opportunities and customers, as well as different channels to market.
Apart from large technology firms, the Banking and Securities sector is one of the leading industry verticals in Artificial Intelligence adoption. The business has relatively high digital maturity, access to troves of data, a desire to glean patterns from historical events as a guide for future decisions, and certainly has tasks that are amenable to automation.
Thus far, AI has made its way into Financial Services with automated trading and investment discovery, trading strategies, robo-advisors, voice-based commerce, customer behavior analysis, and chatbots for customer services, identity verification and fraud detection.
WHAT IS THE CURRENT STAGE OF ARTIFICIAL INTELLIGENCE SOLUTIONS ADOPTION WITHIN YOUR ORGANIZATION?
Hover over bars for values
SAMPLE AI USE CASES ACROSS DIFFERENT INDUSTRY VERTICALS
|INDUSTRY VERTICAL||INDUSTRY AS A % OF TOTAL GLOBAL IT SPEND||SAMPLE AI USE CASES|
|Banking & Securities||19%||Automated trading and investment discovery, trading strategies, robo-advisors, voice-based commerce, customer behavior analysis, chatbots for customer services, identity verification and fraud detection.|
|Government||17%||Smart surveillance, threat detection, Smart Cities and Utilities, AI-enhanced and personalized education and training, chatbots for info distribution and citizen engagement.|
|Manufacturing & Nat. Res.||17%||Predictive maintenance, machine learning-driven insights for yield improvement and optimization.|
|Comms, Media & Services||16%||Customer analytics, forecasting, customer demand trends, video analytics and computer vision interactivity (e.g. in video games and other immersive media).|
|Retail||7%||Customer analytics, forecasting, anticipating demand trends, reducing revenue churn, supply chain management, warehouse automation, chatbots for customer services and conversational commerce.|
|Insurance||7%||Claims management and fraud detection, analyzing customer behavior and reducing revenue churn, automated underwriting, pricing, conversational platforms for customer services, complying with regulations and trading strategies.|
|Utilities||5%||Enhanced supply-demand management based on AI-driven analytics, predictive maintenance, dynamic pricing based on consumption analytics provided by smart meters, for example, chatbots for customer service.|
|Healthcare Providers||4%||Diagnostics, image analytics for early disease detection, drug discovery, patient monitoring (pre-emptive warning systems) and personalized medicine and treatment.|
|Transportation||4%||Self-driving vehicles, advanced driver assistance systems and personalized content delivery/productivity enhancement tools used by providers of transportation services.|
|Education||2%||Customized/adaptive learning programs and skill upgrade programs based on real-time insights gleaned from job market trends.|
|Wholesale Trade||2%||Warehouse automation and inventory management based on insights gleaned from demand analytics and autonomous delivery.|
Scroll to view more >
Connecting Dates with Trading Decisions Using Machine Learning in Interest Rate Markets
The Impact of Machine Learning on Investor Decisions in Fixed Income Markets
Machine learning in trading is entering a new era. While previous algorithms were hard-coded with rules, J.P. Morgan is exploring the next generation of programming, which allows machine learning to independently discover high-performance trading strategies from raw data.
In a recent initiative focused on interest rate markets, a team fed in some 1,250 raw input features from a wide variety of sources, such as daily close levels of U.S. Treasuries, dates of Federal Reserve meetings and international interest rates. The model was built on data from 2000 to 2016, with the intention of then determining whether it could be applied to timing and sizing contemporary trades in 2017.
After testing a variety of machine learning methods, a technique that created an interlinked collection of decision trees emerged as the most effective. Known as “random forest,” the method’s results for U.S Treasuries are shown below.
The third and sixth bars indicate the return on simply buying bonds through conventional methods, which act as a “control.” In terms of machine learning, the first and fourth bars indicate the returns from short selling and the second and fifth bars from both buying and selling.
Performance of Weekly and Monthly RF Predictors
Natural Language Processing in Equity Investing
Machine Learning in Big Data for the Classification of News Sentiment for Equities
Applying machine learning to words, rather than to numbers, is an exciting and rapidly developing field of study. Natural Language Processing creates the potential for a machine to digest hundreds of thousands of written reports and classify the language as sentiment to create a broad investment picture.
In a case study, J.P. Morgan Research built an algorithm based on some 250,000 analyst reports that provided the source material for learning the implication of financial terms such as “overweight,” “neutral” and “underweight.” The team then tested the model on 100,000 news articles that focused on global equity markets with a view to informing future equity investment decisions.
As the table below shows, the signal produced strong returns and outperformed several benchmark indices.
Results when using our classifier to trade the news sentiment (long-short) in different indices
|INDEX||BACK TEST PERIOD||AVG # L/S COUNT||AVG INDEX CVG||HIT RATE (WEEKLY)||L/S RETURN P.W.||L/S RETURN P.A.|
|MSCI UK||12m||66.0||61%||55.6%||8 bps||4.2%|
Scroll to view more >
Clustering Data to Enhance Returns
Using Unsupervised Machine Learning to Enhance Returns and Reduce Risk
Grouping similar items, or clustering, to uncover natural relationships within a set of data is one of the core machine learning techniques that can be utilized to expose additional insight into the global equity markets.
In this report, J. P. Morgan examines various clustering algorithms applied within and across countries, sectors and stocks. In certain equity markets, Dynamic Cluster Neutralization has proven to be a better way to enhance returns and reduce risk than traditional country or sector neutralization.
HIERARCHICAL CLUSTERING DENDROGRAM ON CORRELATIONS
Scroll to view more >
Cluster analysis of MSCI GDM country monthly USD returns between 2000 and 2018 shows two broad groups consisting of North America plus Asia-Pacific nations (countries 1-8) and Europe (countries 9-23). Within these clusters, sensible groupings are also found lower down in the dendrogram, such as USA & Canada, Australia & New Zealand, and Germany (DE) & France which happen to exhibit the highest correlations and therefore shortest branch length in the dendrogram.
Source: J.P. Morgan QDS (March 31, 2018)Report from August 29, 2017
Machine Learning in Value Investing
Incorporating Profitability Measure and Sentiment Signals to Identify Winners and Losers
While there are a number of valuation metrics to account for when calculating the “fair value” of stocks, machine learning has proven to offer a new perspective when assessing value strategies.
In this study, J.P. Morgan implemented machine learning algorithms to assemble a valuation-based strategy to predict the “fair value” of stocks. This is formed around a large number of equity characteristics and the connection between profitability and the quantification of a “mispricing” signal. To further enhance the valuation strategy, RavenPack’s news sentiment data was introduced as a useful overlay to the mispricing signal, along with measuring the impact of investor sentiment. The advanced valuation strategy showed that a combination of ML models can help improve predictions, as opposed to using one model.
AN OVERVIEW OF OUR STOCK SELECTION STRATEGY
Scroll to view more >
Alternate Stock Lending with Unconventional Data
Data on shorting/utilization/borrow cost
In market transactions, stock lending is the act of loaning a security for a fee to an investor who has a short-term need, such as in the case of short selling.
The practice creates unconventional data on volume and borrowing costs. This formed the basis of a J.P. Morgan research project that sought to find strategies providing strong returns from lending patterns.
In particular, the quant team used global data to determine which stocks had high lending rates, assuming strong demand was being driven by short selling. These shares were then also sold short.
At the same time, the model went long on shares that were not in wide circulation in the loan market, based on the hypothesis that the market viewed these as fairly-valued.
The overall approach showed “following the short interest herd,” on average, pays off, particularly in Europe where the information ratio measuring return-to-risk payoff was high. In quant terms, anything above 0.5 is generally considered a success.
15 TRADING STRATEGIES FOR EUROPEAN STOCKS BASED ON ALTERNATE STOCK LENDING DATA WITH THEIR CORRESPONDING INFORMATION RATIO (IR)
|Short Interest (SI)||1.29|
|Short Interest Trend||1.20|
|Composite (SI + SI Trend)||1.18|
|Short Interest Ratio (SIR)||1.29|
|Short Interest Ratio Trend||0.81|
|Composite (SIR + SIR Trend)||1.12|
|Short Int. / Sh. Out. (SISO)||1.52|
|Short Int. / Sh. Out. Trend||1.16|
|Composite (SISO + SISO Trend)||1.57|
|Utilization Ratio (UR)||1.13|
|Utilization Ratio Trend||1.11|
|Composite (UR + UR Trend)||1.50|
|Borrow Cost (BC)||0.26|
|Borrow Cost Trend||0.48|
|Composite (BC + BC Trend)||0.15|
The Global Research Quantitative Council was launched in 2017 to evolve Global Research products and capabilities to enhance the client experience.
The Global Research Quantitative Council will also promote the integration of quantitative techniques in all research products and encourage sharing of best practices, and guide the Global Research strategy on initiatives such as:
Monetization of internal modeling capabilities
Acquisition of external datasets and tools
Model and research tool review and issue resolution
Use of proprietary / internal datasets
This communication is provided for information purposes only. Please read J.P. Morgan research reports related to its contents for more information, including important disclosures. JPMorgan Chase & Co. or its affiliates and/or subsidiaries (collectively, J.P. Morgan) normally make a market and trade as principal in securities, other financial products and other asset classes that may be discussed in this communication.
This communication has been prepared based upon information, including market prices, data and other information, from sources believed to be reliable, but J.P. Morgan does not warrant its completeness or accuracy except with respect to any disclosures relative to J.P. Morgan and/or its affiliates and an analyst's involvement with any company (or security, other financial product or other asset class) that may be the subject of this communication. Any opinions and estimates constitute our judgment as of the date of this material and are subject to change without notice. Past performance is not indicative of future results. This communication is not intended as an offer or solicitation for the purchase or sale of any financial instrument. J.P. Morgan Research does not provide individually tailored investment advice. Any opinions and recommendations herein do not take into account individual client circumstances, objectives, or needs and are not intended as recommendations of particular securities, financial instruments or strategies to particular clients. You must make your own independent decisions regarding any securities, financial instruments or strategies mentioned or related to the information herein. Periodic updates may be provided on companies, issuers or industries based on specific developments or announcements, market conditions or any other publicly available information. However, J.P. Morgan may be restricted from updating information contained in this communication for regulatory or other reasons. Clients should contact analysts and execute transactions through a J.P. Morgan subsidiary or affiliate in their home jurisdiction unless governing law permits otherwise.
This communication may not be redistributed or retransmitted, in whole or in part, or in any form or manner, without the express written consent of J.P. Morgan. Any unauthorized use or disclosure is prohibited. Receipt and review of this information constitutes your agreement not to redistribute or retransmit the contents and information contained in this communication without first obtaining express permission from an authorized officer of J.P. Morgan.