Using Sentiment Analysis to Navigate the Italian Government Bond Market
The BTP/Bund yield spread made headlines in 2018 when it spiked to over 300 basis points, after the election of a Eurosceptic coalition in Italy; and in this study Gábor Komáromi, Head of Fixed Income at RavenPack, wanted to find out if information aggregated from online news sources could have helped investors steer a course through this volatile market.
In order to do this, he used the RavenPack platform’s natural language processing engine to aggregate and quantify sentiment extracted from online news sources about Italian and German bond market themes and drivers.
From this data, he then constructed a time-series of sentiment scores that could be tested for predictive power in forecasting yields.
The time-series was used to construct two similar bond trading strategies: a bond strategy in which buy signals were given when news sentiment crossed above a positive threshold and sell signals when it crossed below a negative threshold.
Results of Our Sentiment-Based Bond Strategy
The results from back-testing the two strategies were encouraging, with both showing to have significantly outperformed a simple buy-and-hold strategy, as can be seen from the table below.
The outperformance of the RavenPack bond strategy suggests sentiment data can provide the basis for forecasting BTP/Bund spreads.
Sovereign Bonds Events and Sentiment Data
Easily implement the results of this white paper using the RavenPack Analytics Platform, which includes event and sentiment data on all major sovereign bonds, available via dashboards or our web APIs. Email firstname.lastname@example.org to set up a trial.Request White Paper