The mannequin produced a number of observable patterns in each market conduct and language construction. These findings illustrate how text-based alerts align with subsequent yield curve actions.
Market Construction and Curve Dynamics
First, short-term volatility within the Brazilian fastened revenue market is greater than long-term volatility. This contrasts with conventional concept and means that, in rising markets, traders react extra strongly to short-term information and coverage alerts. Lengthy-term devices seem to commerce with comparatively decrease volatility, reflecting the dominance of institutional traders at longer maturities.
As well as, 84% of each day yield curve actions fall into 4 of the eleven customary configurations recognized within the literature, with parallel upward and parallel downward shifts among the many most frequent (additionally confirming this brief time period volatility taste). This focus highlights the significance of appropriately classifying a small set of dominant curve dynamics.
Extracting Sign from Language
To arrange the textual content knowledge, frequent phrases resembling “committee,” “state of affairs,” “billions,” and “costs” have been eliminated as cease phrases, as they don’t contribute to classification. Phrase frequencies have been then mapped for every yield curve motion class, permitting comparability of language patterns throughout completely different curve configurations.
Seasonality in Curve Actions
When inspecting the language related to particular actions, a seasonal sample emerged. For instance, bear flattening actions have been incessantly related to references to August, September, and October, whereas bull flattening actions have been extra typically linked to January, February, and March. A chi-squared check offered statistical proof of seasonality throughout a number of yield curve actions.
