In this experiment agents repeatedly invest in three state-specific assets. Theory requires portfolios to match the probabilities of the states. The treatments endow none, one, three, or all (of eight) investors with probability information. Will professional consulting evolve as an entrepreneurial activity? Will decisions follow theory or the 1/n-heuristic of equal investments in all assets? Will agents with probability information be asked and paid for advice? Portfolios designed with probability information reect the benchmark but do not converge to it over time. Learning to avoid the 1/n-heuristic is extremely fast. Although advice is sought and readily paid for, recommendations are not always followed.