harrytrader wrote: I assume Random Market Hypothesis is the same as Random Walk Hypothesis (that is what it is referred to in the literature). In any case remember a random walk is a markov chain, but not all markov chains are random walks. This is why Technical Analysis maps to (multidimensional ) Markov random processes that are not simple random walk processes. __________________
I don't know if you consulted the answer I gave you to another thread since you asked if there was a difference in litterature between RMH and EMH http://www.elitetrader.com/vb/showthread.php?s=&threadid=25864&perpage=6&highlight=rmh&pagenumber=2 http://www.e-m-h.org/random_walk.html "with the benefit of hindsight and the theoretical insights of LeRoy (1973) and Lucas (1973), it is now clear that efficient markets and the random walk hypothesis are two distinct ideas" Saying that Technical Analysis maps to (multidimensional ) Markov random processes that are not simple random walk processes once again doesn't distinguish from today's EMH which uses also these markov chains and don't restrict them to being random walk since at least Samuelson. Random Walk is dated from Bachelier at the beginning of 20th century whereas "modern" EMH which doesn't lie on random walk is dated from Samuelson in the 1970 : he didn't receive the Nobel Prize for random walk of course but for his martingale theory applied to financial market. Fama, one of the so-called Chicago School like Samuelson and who popularize the 3 different forms of EMH, dictates the "official" theory of finance notably to regulators, clearly reaffirms EMH predominance in his article where he says: "efficient market hypothesis dominated financial theory for more than 30 years" and continues to do so since there is no rigorous proof from other schools above all from TA (remember that CFTC even considered that TA is probably charlatanism) - as I pointed already: http://www.econometric-wave.com/home/home.html.html
harrytrader wrote: My question was the difference between what you were calling RMH and RWH. So what TA is not Markov (or perhaps modelled as a Hidden Markov Model)? I could model some 3-step candlestick processes (let the states be [0,C,H,L] tuples) as third-order Markov chains for example.