April 3, 2023
We take a look at how the latest innovations in Language AI can offer tangible solutions for the challenges of this year, as well as ideas for original investment approaches.
2023 started in the shadow of the Russian invasion of Ukraine, global inflation, and uncertain recession forecasts. It continued with turmoil in the banking system caused by the collapse of Silicon Valley Bank, Signature Bank and the last-minute rescue of Credit Suisse. These unpredictable developments have been illustrating in compelling ways the importance of managing risk and catching opportunities as they unfold.
Meanwhile, Language AI has taken the world by storm, with Large Language Models like GPT-4 reshaping how we work, ideate, learn and make decisions.
In this report, we take a look at how the latest innovations in Language AI can offer tangible solutions for the challenges of this year, as well as ideas for original investment approaches.
Here are 5 data-driven insights that can help you:
Recent research shows that inflation strategies that incorporate sentiment outperform models that depend merely on past observed inflation. RavenPack’s latest white paper proposes a multi-asset allocation strategy for hedging against inflation, which maximizes the portfolio's mean-variance criterion by investing in equities, commodities, and bonds based on inflation nowcasts. Back tested against the US equity market, a 10-year US government bond, and six conventional commodity indexes, the proposed strategy outperformed the S&P 500® in terms of return, volatility, Sharpe ratio, and maximum drawdown.
Graphic
Advanced language models are now able to “consume“ huge amounts of news and predict GDP in real time, solving the issue of official data lags. This is true for the largest economies like in the US, but also for developing economics like Brazil, India, and South Africa. One of our studies found that RavenPack news analytics helped improve timeliness and accuracy of traditional macroeconomic nowcasting models by up to 25%, depending on the country and the prediction period.
China is a special case, since economic figures are sometimes delayed by months. Using a cutting-edge approach that leverages sentiment indicators, RavenPack was able to nowcast China's macroeconomic indicators with enhanced accuracy and timeliness, reducing the uncertainty over predicted economic indicators by up to 7%.
Earnings calls can offer a window into the financial executives’ views about the past, present, and future state of the economy. A real-time Transcript Economic Sentiment Indicator for the U.S. uses NLP event detection and sentiment scoring capabilities to analyze earnings call transcripts. Among others, it was found that the indicator could identify significant inflection points in the U.S. economic activity, and detect meaningful changes in the business cycle. The model accurately detects up to 70% of upshifts of the U.S. GDP, and 60% of downshifts. Check out the latest indicator data here.
Staying ahead of controversies helps optimize bond selection. Media attention, real-time news and NLP technology can now help investors outsmart the negative impact of controversies on corporate bonds. Research shows that corporate bonds from issuers involved in severe controversies consistently underperform credit portfolios with low or no controversy bonds.
Bond prices can also be impacted by news related to credit ratings, analyst ratings, and price targets. This paper looks at the impact of credit-related announcements on corporate bond prices using sentiment-based trading strategies that analyze media content. The approach can bring annualized returns amount to 200 bps above the annual credit benchmark at a two-day holding period, and 90 bps at the one-week holding period.
Companies hiring for novel tech skills deliver higher investment returns, shows a new research using RavenPack Job Analytics to quantify tech adoption in hiring posts. For instance, a long-only portfolio leveraging the number of novel technology skills detected delivers annualized excess returns of 2.8%, relative to the sector.
Another paper looks at positions, hiring locations and job descriptions, after analyzing over 200 million job postings sourced LinkUp directly from employers websites. Research discovered that monthly hiring growth is indeed positively correlated with future stock performance. What's more, companies with higher hiring growth that also recruit in similar locations month on month, outperform the rest. A long/short sector-neutral portfolio delivers an Information Ratio of 1.1, and Annualized Returns of 2.9% with a weekly holding period.
As we dive into the second quarter of this increasingly unpredictable year, preparedness remains key. Our team of data scientists can help address your challenges or offer you inspiration on how to look at data from unprecedented angles. Book a call to explore how data can help you make more informed decisions.
RavenPack was recognized as the Alternative Data Vendor of the Year, by the prestigious Risk Markets Technology Awards 2022.
RavenPack Edge is an AI platform that collects, reads, and analyzes billions of documents to help finance professionals and businesses to make more informed decisions, stay ahead of competition and mitigate emerging risks.
Please use your business email. If you don't have one, please email us at info@ravenpack.com.