Experimental Investigation of Percieved Risk in Random Walk Processes

The hypothesis that, on average, people accurately estimate probabilities in random walk processes is experimentally investigated.Individuals are confronted with a process that starts with $X, and in every stage either goes up or down by $1, with probabilities p and 1 - p respectively.For different values of p, individuals were asked to estimate what is the chance that after 10 stages the system will be at a point higher than or equal to $X.Systematic mistakes in estimations were observed.In particular, estimations were centered around the stage-by-stage probability (p) rather then around the actual probability. Implication of this result to random walk processes in finance is considered.