The efficient market hypothesis is the idea that markets quickly take new information into account. Broadly speaking, it explains why a random person cannot achieve guaranteed profit by picking up the business section of a newspaper and buying stock in companies that seem to be doing well. According to the efficient market hypothesis, all of this news will have already been taken into account by the stock price; future events will affect the stock unpredictably.
The economic principle underlying the efficient market hypothesis is arbitrage. Arbitrage is the practice of making a guaranteed profit by exploiting some flaw in the market. A basic example of arbitrage would be buying something at a low price when you know you can immediately sell it for more money. Applying the principle of arbitrage to information yields the efficient market hypothesis. The idea is that, if the information is out there, it will have already been acted upon.
The strongest version of the efficient market hypothesis predicts that the market will follow a 'random walk.' That is, it predicts that, at any given time, any given stock, and the market as a whole, is just as likely to rise as to fall. The long term tendency of the market and all stocks within it will thus be nothing but an accumulation of random decisions. Long-term trends should be impossible to identify. More precisely: as soon as trends become identifiable, they disappear, because investors will buy and sell stocks according to any apparent trend. In doing so, they negate it. If the stock can be reasonably expected to rise for the rest of the year, investment decisions will incorporate this future value—with appropriate discounting—into the present price.
By necessity, the efficient market hypothesis can only ever constitute a rough approximation. For the hypothesis to work correctly, the market has to be filled with a number of intelligent and rational agents who act on their assessments of trends and value. Paradoxically, if these agents were to all assume the efficient market hypothesis, the system would fall apart. The most active participants in the market must believe, to some degree, that they are capable of making profitable decisions on the basis of new information or evaluations.
Because of this paradox, and because of a large amount of data opposing it, the efficient market hypothesis is extremely controversial. It remains a core element of neoclassical economics, and is still widely taught. Many economists probably consider the hypothesis to be a good description of a market functioning ideally. However, real world markets all deviate from perfect efficiency, some more than others. For example, the oil futures market, in which many well-informed and well-financed investors participate, probably conforms better to the hypothesis than does the used car market.