RavenPack Q1 Research Roundup

In this article, you will read about:

The power of sentiment-enhanced language models

Media coverage as a stock performance indicator

AI-Powered risk management for multi-asset strategies

Earnings intelligence: A powerful signal for investors

During the first quarter, our data science team published new research on the performance of finely-tuned large language models for sentiment classification, the link between media coverage on innovation and stock performance, enhancing risk-based multi-asset allocations in various inflationary regimes, and a deeper dive into earnings intelligence. Check it out:

The Power of Sentiment-Enhanced Language Models

Our research demonstrates that optimizing language models for sentiment analysis can significantly improve their performance. Even FinBERT, a model pre-trained on financial data, showed a 48% increase in Information Ratio (IR) when fine-tuned with sentiment-enriched annotations.

Portfolio	IR		AR		Portfolio Size
Model	Raw	Fine-tuned	Raw	Fine-tuned	Raw	Fine-tuned
US Mid/Large-Cap	0.80	1.27	4.35%	6.74%	149.5	169.9
US Small-Caps	2.63	3.42	25.55%	29.63%	128.2	160.6
EU Mid/Large-Cap	1.42	2.36	8.82%	14.76%	62.6	65.8
EU Small-Caps	0.99	2.03	9.79%	20.15%	15.8	18.4
APAC Mid/Large-Cap	2.86	3.07	19.05%	23.29%	47.4	49.0
APAC Small-Caps	2.47	2.94	23.91%	28.12%	26.4	30.5

Key portfolio performance statistics for various universes, comparing raw FinBERT and fine-tuned regression FinBERT models from January 2010 to September 2023.

Media Coverage as a Stock Performance Indicator

By analyzing historical news articles and patent filings, we discovered a correlation between media coverage of innovation and a company's future stock performance. This finding can be a valuable tool for identifying potential market leaders.

FIGURE: Portfolio performance associated to the Global Innovation long-short strategy rebalanced at a monthly frequency from 2011 to 2023 as a function of investor’s preference for sector bias. RavenPack, September 2023

AI-Powered Risk Management for Multi-Asset Strategies

Our research explored how AI could enhance risk-based multi-asset strategies to adapt to various inflationary environments in real-time. Backtesting demonstrated that this strategy consistently outperformed the S&P 500 MARC by 5%, with a Sharpe Ratio of 1.29.

Chart — **FIGURE:** Multi-Asset Inflation-Based Strategy with a 5% volatility target employs the RavenPack inflation indicator to identify inflation states and construct the unscaled portfolio. The strategy considers the cost of leverage, calculated by subtracting the Federal Funds rate. The out-of-sample backtest covers the period from November 2013 to August 2023 and when compared to the S&P MARC 5%.

Earnings Intelligence: A Powerful Signal for Investors

Combining earnings-related signals from various alternative datasets can deliver Information Ratios of up to 2.6. RavenPack's Earnings Intelligence simplifies complex earnings data (news, transcripts, and announcement dates) and provides these valuable signals as off-the-shelf factors for investors.