The mannequin produced a number of observable patterns in each market habits 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 increased than long-term volatility. This contrasts with conventional principle and means that, in rising markets, buyers 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 buyers at longer maturities.
As well as, 84% of day by day yield curve actions fall into 4 of the eleven normal configurations recognized within the literature, with parallel upward and parallel downward shifts among the many most frequent (additionally confirming this quick time period volatility taste). This focus highlights the significance of accurately classifying a small set of dominant curve dynamics.
Extracting Sign from Language
To organize the textual content knowledge, widespread phrases corresponding to “committee,” “situation,” “billions,” and “costs” had been eliminated as cease phrases, as they don’t contribute to classification. Phrase frequencies had been then mapped for every yield curve motion class, permitting comparability of language patterns throughout totally different curve configurations.
Seasonality in Curve Actions
When inspecting the language related to particular actions, a seasonal sample emerged. For instance, bear flattening actions had been steadily related to references to August, September, and October, whereas bull flattening actions had been extra usually linked to January, February, and March. A chi-squared check offered statistical proof of seasonality throughout a number of yield curve actions.
