| April 08, 2014
In this study, working with Arialytics’ predictive financial modeling platform, we construct long-only and long-short equity portfolios based on sector level 1-month expected returns.
In this study, we utilize Arialytics’ predictive financial modeling platform optimized for highlevel computing and RavenPack’s news analytics derived from real-time content published by Dow Jones Newswires, The Wall Street Journal, and Barron’s - to construct long-only and longshort portfolios based on sector level 1-month expected returns. The constructed portfolios have been found to consistently outperform the S&P 500 on both an absolute and risk-adjusted basis.
Expected Return Indicators
The complexity of financial markets makes it challenging to model the impact of news on individual stock prices, sectors, or markets – especially over the long-term. Although the intraday reaction to a news event is often relatively straightforward, the impact over days and weeks become harder to model. More so when investors are bombarded with lots of new information - some of which can be extremely valuable and some merely noise. The sheer volume of news and financial data available to investors present serious analytics challenges that cannot be solved with traditional tools.
Due to the size of these ‘big data’ sets, estimating predictive models using conventional statistical means is not feasible. We therefore perform our analysis using the Arialytics predictive financial modeling platform called “Aria”, a distributed computing platform optimized for finding statistically robust solutions from large amounts of financial (structured) and non-financial (news) data.
Catering to longer-term investors, we examine U.S. equity portfolio formation strategies utilizing one-month ahead expected return estimates derived from RavenPack News Analytics and priced market factors. The underlying data files for this equity portfolio return analysis consist of daily one-month forward raw (log) return estimates for the following major U.S. industries: basic materials, consumer services, energy, financials, healthcare, industrials, consumer goods, technology, telecommunications, and utilities. These major industries are tracked by Dow Jones industry-level price indexes: DJUSBM, DJUSCY, DJUSEN, DJUSFN, DJUSHC, DJUSIN, DJUSNC, DJUSTC, DJUSTL, and DJUSUT, respectively. The expected return estimates we generate are published daily 90 minutes prior to market open. Based on our daily predictions, we construct a set of long-only and long-short strategies (including 130/30, 150/50, and 200/100); All of which are shown to outperform the long-only US market benchmark over nearly 4 years of backtesting - covering the period February 2010 through September 2013.
In the following sections, we provide an overview of how we estimate our expected returns. We describe the portfolio construction process including how the benchmarks and our longonly and long-short trading strategies are created. We evaluate our strategy results including applying different portfolio turnover restrictions and finally present our conclusions.
The underlying data for each target security’s portfolio weights, or expected return estimation, consists of a single RavenPack U.S. market file, multiple RavenPack sector files (one per sector), and their corresponding Arialytics market and sector files.
Please use your business email. If you don't have one, please email us at email@example.com.
We will process your personal data with the purpose of managing your personal account on
RavenPack and offering our services. You can exercise your rights of access, rectification,
erasure, restriction of processing, data portability and objection by emailing us at firstname.lastname@example.org. For more information, you can
Your request has been recorded and a team member will be in touch soon.
High inflation has returned in developed markets after decades of lying low. In our latest paper, we show how to build an inflation-based asset allocation strategy using sentiment data and we illustrate that sentiment-based strategies outperform models that depend merely on past observed inflation values.
This year's RavenPack Research Symposium brought two intense days of knowledge sharing in London and New York, from 25 top experts in natural language processing, quantitative investing and machine learning. Together, we explored how firms can leverage new language models to generate alpha, better manage risk and respond to calls for more sustainable investment practices.
Human capital is at the heart of value creation. Our latest research demonstrates how unprecedented workforce insights, sourced from over 200 million job postings, can generate more alpha.