Thinking near and far: Modeling the formation of traders' beliefs in asset markets using experimental data

We present a dynamic model that expresses how traders use historical price patterns and recent trends and the weights they assign to each information sets when predicting future market behavior. We show that although traders initially have more faith in the validity of information about historical events, in light of recent price changes, they gradually shift their reliance to more current markettrends as they realize that the prices they are witnessing do not follow the historical patterns. As soon as the price trend changes again, however, traders' confidence in the information contained in historical events is restored, and they consequently rely less on the more current price changes. We estimate the model using experimental data and explain the role of the model parameters.