I certainly am not the person to backup the EMH (my complaints about it are more about the lack of provability/usefulness) but you are completely misstating some very central points about the EMH and more specifically about Richard Dennis.
1 - no version, even the most strict, of the EMH claims that returns above market over time are not possible. The claims are about the prices of assets and their relation to information. So even a very strict interpretation of the EMH wouldn't say you can't make above market returns over time, only that you can't do it because you know something no one else does. Random movements or other non-knowable events are not part of the EMH.
2 - all EMH versions talk about risk adjusted returns. So if you place an incredibly risky bet, that pays off, that does not violate any form of the EMH. The Dennis system was particularly risky, especially in comparison to other market participants like him.
3 - Dennis' system did not perform well over a long period. The system does not perform well past 1986 and was only first codified in either 1983 or 84. Dennis performed better over that time, but his system did not. This is evidence that he was lucky, not beating an efficient market. Even if we take the most charitable view of Dennis' system, not that it was luck, rather it was some new found previously unknown information, that backs up less strict forms of the efficient market hypothesis. The market responded to the actors in it finding an inefficiency by making that efficient to the point of the opportunity disappearing.
1 - no version, even the most strict, of the EMH claims that returns above market over time are not possible. The claims are about the prices of assets and their relation to information. So even a very strict interpretation of the EMH wouldn't say you can't make above market returns over time, only that you can't do it because you know something no one else does. Random movements or other non-knowable events are not part of the EMH.
2 - all EMH versions talk about risk adjusted returns. So if you place an incredibly risky bet, that pays off, that does not violate any form of the EMH. The Dennis system was particularly risky, especially in comparison to other market participants like him.
3 - Dennis' system did not perform well over a long period. The system does not perform well past 1986 and was only first codified in either 1983 or 84. Dennis performed better over that time, but his system did not. This is evidence that he was lucky, not beating an efficient market. Even if we take the most charitable view of Dennis' system, not that it was luck, rather it was some new found previously unknown information, that backs up less strict forms of the efficient market hypothesis. The market responded to the actors in it finding an inefficiency by making that efficient to the point of the opportunity disappearing.