Political Communication and Cryptocurrency Volatility. Level Effects and Regime Transitions at High Frequency

We ask whether high-frequency political communication shapes the level of conditional volatility, its persistence, or the probability of transiting between volatility states. The question is addressed with hourly Bitcoin prices and the universe of Donald Trump’s social media posts from August 2017 to February 2026, scoring the directional tone of market-relevant posts and entering it as an exogenous regressor in single-regime and Markov-switching variance equations. Three results follow. First, the tone coefficient is recovered only in log-linear specifications, where more positive communication coincides with lower conditional volatility and adversarial communication with higher volatility; additive specifications place the estimate at the boundary of the admissible parameter space. Second, a two-state Markov-switching EGARCH identifies a calm and a turbulent state with expected durations of eight and four hours, so that regime alternation is an intraday phenomenon that daily aggregation cannot resolve. Third, the tone effect is proportionally identical across the two states, and a fourteen-month interruption in communication leaves regime alternation essentially unchanged. Political communication thus modulates volatility intensity within whichever state prevails, without governing transitions between states.