p-Values of the Silverman test for bimodality (column 2 and 3).

The null hypothesis is unimodality with a significance level of 0.02. The test is on the mispricing δ, once with the empirical filtered fundamental value and once with the smoothed. ✗: acceptance of the unimodal null hypothesis for both empirical time series, ✓: twofold rejection (bimodality), : inconclusiveness. Center Right (Numerical): modality test results of simulated time series of the non-linear model, using parameters from Table V in S1 Appendices. Right: Jensen-Shannon (J-S) distances between the empirical and numerical mispricing distributions. A J-S distance of 0.2 corresponds to a Kolmogorov-Smirnov distance of 8% (i.e. the maximum distance between two cumulative distributions functions). For comparison, the J-S distance between LCATTLE and AU is 0.19.