S1 File - Decrypting Financial Markets through E-Joint Attention Efforts: On-Line Adaptive Networks of Investors in Periods of Market Uncertainty

Table A. The network metrics during low and high volatility phases. Values indicate the means of variables. We used a t-test to evaluate the statistical significance of the difference between the values. It is worth noting that * indicates that the difference was significant at the 5% level. Table B. The network metrics during initial and high volatility phases. The values indicate variables means. We used a t-test to evaluate the statistical significance of the differences between values. It is worth noting that * indicates that the difference was significant at 5%. Table C. The network metrics during initial and low volatility phases. The values indicate the variable means. We used a t-test to evaluate the statistical significance of the difference between the values. It is worth noting that * indicates that the difference was significant at 5%.Table D. Summary of the features of our trust schemes. Table E. The aggregation of our trust schemes. Table F. Test of the trust scheme. Table G. Test of the global scheme. Table H. Autoregressive coefficients (different columns in each panel) and intercepts for each equation (different lines in each panel), in the different regimes (different panels) of the MS-VAR model.* indicates that the parameter was significant at the 5% level. Table I. Model selection following log-likelihood, AIC and BIC. Table L. Mean value of the network metrics during the initial period, and the low, moderate and high volatility phases. Fig A. Typical performance of the SVM classifier used to classify messages into trading related ones and non-trading related ones. Fig B. Unicredit stock log-return, rt, series (blue line, left axis) and the filtered volatility regime st|t (red line, right axis) for K = 4.
(DOCX)