As Matt said, the two basic assumptions of the efficient theory has some problems.
I would like to make my statement clearer.
1. People invest on their decisions, not information.
Different people may make different decisions even based on the same information. Since we do not know all people’s tastes or transformations of tastes, it is very difficult to tell or even predict where the demand curve actually is.
Take the impact of the financial crisis on the stock market as an example. If we explain the phenomenon by the changes of tastes of most inventors, we actually explain nothing because it a tautological theory. On the other hand, if we assume the tastes of all investors remain the same, we have to figure out the changes of the opportunity costs of holding,selling, or purchasing the stocks, so that we can tell what changes will be in the stock market. This is a refutable theory but it is also not easy to achieve. That’s why most economists are unable to predict the crisis. (I would also like to point out the anyone who claims that he predict a crisis successfully, he is based on luck more than ability.)
Further, if we can know all the constraints, the “Pareto condition” will always be achieved, which means the market is always efficient(being cleared). This explanation fits the efficient theory, but it is also a tautology, which has few explanatory power. We are still unable to tell when would the whole market generally goes down or up.
2. In an economic theory that has explanatory power, it is unnecessary to assume that people are rational. We only need to assume that people are oriented by their own interests(or selfishness).
Since the latter assumption can let us draw a demand curve with negative slope(which means a person tends to buy more if a good is cheaper than before), which can create refutable theory to explain or predict people’s behaviors. However there is no exact definition of “ration,” we can not build up any refutable theory based on the rational assumption.
Conclusively, it means the efficient theory is to be a tautology at all.