there are even more "*MHs" to this palette, like AMH (adaptive market hypothesis (Lo, 2004, 2005)), and FMH (fractal market hypothesis, (Peters, 1991)), which instead of answering EMH question, offer another way of asking it. Generally speaking, whether the market is informationally efficient or not depends heavily upon the way of measuring it, so there is no - and won't be - the one and only answer.

It would be interesting to split this hypothesis in short/long term.

On short term EMH is wrong. There is simply not enough information for a rational behavior of investors. On a long time perspective, required data for a rational behavior is available, so market is more likely to correctly determine the prices.

From time to time, market behavior defies rationality, thus creating bubbles. But bubbles do not prove market is inefficient, they prove the opposite.

One of the most important concepts in modern portfolio theory is the Efficient Market Hypothesis (EMH): the market is assumed to be efficient, since there are a lot of rational, profit-seeking and risk-averting investors competing in forecasting the future value of stocks. Any new change related to a stock is rapidly known in the entire investment community and soon reflected in the price of the stock. However, it can be assumed without any doubt, that the market is always efficient? The answer to this question is not a simple one. With respect to market behavior there are, at the extremes, two views. At one extreme is the well-known EMH which says that the prices are always fair and quickly reflective of information. In such a world neither professional investors nor the proverbial little investors will be able to systematically pick winners or losers. At the other extreme, we have the so called the “Market Failure Hypothesis (MFH)”. According to this view, prices react to information slowly enough to allow some investors, presumably professionals, to systematically outperform markets and most other investors.

Under the EMH, as soon as there is some information available indicating that a stock is underpriced and profit can be made out of it, investors will buy that stock and as a consequence the price will rise to a fair level. After prices reach fair levels, they will increase or decrease only in reaction to future new information. For this reason, if an investor wishes to take advantage of this situation, he/she should have access to new information available in the market before any other investor. However, future new information is unpredictable; otherwise it would be part of today’s information. Thus, stock prices are also unpredictable and follow a random behavior. It is important to note that most academicians support the random walk theory but practitioners, for example those that work in Wall Street, believe that past stock prices show predictable trends and this way it is possible to forecast the market using past performance information.

Perhaps using to forecast the market a short past period of stock prices information as a basis would not provide useful information to investors. However, if investors use to forecast the market a series of past information on stock prices but in a longer period then perhaps, under certain circumstances and using the appropriate mathematical tools, the outcome of the action could provide useful information for investors to predict the market. To achieve this goal it is important to use the information of past stock price in a sufficiently long period, an adequate mathematic model for forecast the market, identify very well the circumstance under which the market worked and the possible changes of these circumstances in order to avoid that a possible crisis in the sector could change the normal evolution of the market, among other relevant factors.

The EMH has very important implications for investment analysis and corporate management, but the hypothesis is impossible to be tested directly. To do this it would be indispensable to know in advanced the following elements; a) the market’s anticipated net operational cash flows; b) anticipated required rates of return for all future periods; and c) all information relevant to security prices and the way this information is reflected in prices. Thus, for the use of EMH in investment analysis and management corporation is necessary to create tests of the EMH using existing information and statistical techniques, and these tests are usually made under the assumptions that: a) there are no transaction costs; b) no taxation; and c) free access to all available information for all traders. Based on what has been said before can be concluded that there are evidences supporting and rejecting at the same time the EMH.

I think that it is hard to accept that a market is always efficient, but also that a market is always inefficient. It's better to discuss about markets with a higher or a lower level of efficiency.

I agree that Lo (2004, 2005) is a turning point in the EMH theory.

I also appreciate the survey of Lim and Brooks (2011) : Lim KP, Brooks R (2011) The evolution of Stock Market Efficiency over time: a survey of empirical literature. J Econ Surv 25(1):69 -108.

BTW, portfolio theory could be applied to inefficient markets, as well as to efficient ones. Efficiency is not a necessary condition for MPT.

Markets are not efficient.

If they were, all past information would be incorporated in prices of traded assets.

Price changes would depend only on future (unexpected) events, and would follow random laws.

This contradicts the serial correlations (or autocorrelation) of price changes that can be observed in many markets (stocks, commodities, foreign exchange, ..).

https://www.researchgate.net/publication/285590240_Sensitivity_Analysis_of_Macroeconomic_Variables_and_Stock_Returns_A_Cross-Country_Comparison_between_Bangladesh_and_India

Book Sensitivity Analysis of Macroeconomic Variables and Stock Re...

Jorge: some excellent answers so far. I have one more to offer:

you may want to go to my web site, paradoxinvestments.com, where i clearly show (as nobel laureate robert shiller has stated) that the EMH is a hoax,  pure and simple. click on the OVERVIEW tab on my site's home page for an explanation. also see article 9, "market efficiency in doubt," and a second article #9, "evidence of  the mispricing of common stocks --  the DJIA reexamined."

there are other articles on the site that apply--especially article #10, "cycli-graphs of company earnings and stock prices illustrating earnings dominate influence over prices."

 thanks for your question -- this topic interests me a lot.  - gene hawkins

How do you test for semi strong form market efficiency using the Adaptive Market Hypothesis? - ResearchGate. Available from: https://www.researchgate.net/post/How_do_you_test_for_semi_strong_form_market_efficiency_using_the_Adaptive_Market_Hypothesis [accessed Apr 7, 2016].

Only two words are enough to defy the EMH: "Information Asymmetry".

It applies in all walks of life. Financial markets are driven by IA. Deeper you read into it more you understand why the wealth distribution is so skewed. 

Jorge: I have one more point regarding the efficient market hypothesis. that incomparable genius john maynard keynes also thought it was absurd! .  thus, on my web site, paradoxinvestments.com, click the INVESTMENTS tab on the home page and under the first heading, "science vs. reality on the new york stock exchange," click on article #1 " entitled "profile of john maynard keynes...etc" which contains his classic four page essay on how investors (mis) behave because they cannot "correctly price" or value a long-term PERPETUAL claim. he does it with one brilliant  metaphor after another!