Following the frequency and severity of recent market downturns, investors have turned their focus to managing the risks of unexpected losses. These events have also attracted the attention of academics, who are exploring new frameworks for pricing risk.
A growing corpus of research shows that news analytics data has proven useful in forecasting market volatility and that the inclusion of news-related information delivers more precise and timely hedging strategies compared to the competing models without this information. The independent nature and high frequency of news are the two fundamental reasons why the inclusion of news-related information in risk modelling can produce more accurate and timely results.
In this paper, RavenPack summarizes the advantages of using its Analytics product to detect and measure financial risks through news by looking at the results of 14 relevant research papers.
The paper outlines how news metrics can be integrated into volatility models by incorporating them into Heterogeneous Autoregressive (HAR) or GARCH models. Furthermore, it highlights that news has a quantifiable effect on the bond market through a positive relationship between news and bond liquidity. It also outlines various research papers that combine news sentiment, volume and relevance scores to create new indicators, which can be used, for example, to capture changes in volatility and to predict sovereign bond yields.
While the paper covers a wide range of applications, news can also be used to predict and measure crash risk, option implied volatility and portfolio risk. The academic literature presented in this paper highlights the applications of news data beyond the traditional approaches in investment management.
RavenPack is also developing new applications for macroeconomic news to increase the accuracy of forecasting and nowcasting models for Non-Farm Payroll (NFP) and GDP indicators. Furthermore, company relationship co-mention networks derived from news offer a differentiated approach to uncovering linkages that are otherwise not disclosed in traditional sources. This ongoing internal research on such networks aims to showcase a range of risk management applications by understanding potential signal dissemination effects around economically linked companies.