Googling and ARMS sentiment across economic phases and horizons: insights into Brazilian crises

Purpose This study investigates the impact of fundamental factors (returns, liquidity, and volatility) and investor sentiment, captured through indirect (ARMS index) and direct (Google Trends search volumes of positive and negative terms) measures across the Expansion and Contraction phases and three time horizons, on the risk of stock market crashes in Brazil. Design/methodology/approach The analysis applies the CMAX (current index level relative to the historical maximum) method for crisis identification and the local bull-bear indicator to classify economic phases over short-, medium-, and long-term horizons. Probit models were then estimated to assess how these factors affect the likelihood of crisis occurrence. Findings The results reveal that declining returns, reduced liquidity, and heightened volatility increase the probability of crisis. Investor sentiment, marked by optimism during expansions and pessimism during contractions, significantly predicts crisis risk, particularly over shorter horizons, although its effect diminishes as market fundamentals regain their influence. Notably, Google Trends exhibited strong predictive power, outperforming the ARMS index. Practical implications These findings provide timely insights for investors, analysts, and policymakers. Real-time sentiment monitoring via Google Trends supports early detection of market instability, aiding portfolio management, stress testing, and policy actions. The study also informs macroprudential regulation by incorporating behavioral indicators into systemic risk frameworks. Originality/value This study is among the first to examine direct and indirect investor sentiment across economic cycles in an emerging market, offering a robust framework to anticipate crises while highlighting the societal benefits of early detection and the importance of financial literacy in curbing emotional market behavior.