The effect of price volatility on judgmental forecasts: The correlated response model

Traders often employ judgmental methods when making financial forecasts. To characterize judgmental forecasts from graphically-presented time series, I propose the correlated response model, according to which the properties of judgmental forecasts are correlated with those of the forecasted series. In two experiments, participants were presented with graphs depicting synthetic price series. In Experiment 1, participants were asked to make point forecasts for different time horizons. Participants could control the graphs’ time scales. In Experiment 2, participants made multi-period forecasts, and could apply moving average filters to the graphs. The dispersion of point forecasts between participants (the standard deviation of participants’ point forecasts) and the variability of individual participant’s multi-period forecasts (local steepness and oscillation) were extracted. Both forecast measures were found to be significantly correlated with variability measures of the original, scaled, and smoothed data graphs. Thus, the results supported the correlated response model and provided insights into the forecasting process.