Quote from marketsurfer:
How many moves or series of moves in one direction increase the odds that the next move or series will be in the same direction?
Tudor Jones-- LOL. He was over a year wrong with his supposed 87 crash call with charts
From Victor Niederhoffer:
</b>Victor: Any trend that exists can be quantified and its departure from randomness can be measured with the usual statistical procedures, such as confidence intervals and likelihoods. Serial correlation coefficients, regression coefficients of current changes versus past changes, and magnitudes of the impact of past moving averages on the future, distributions of the length of runs, the correllelogram, the expected waiting times between peaks and valleys, survival statistics. All these techniques are very good at discovering any non-random elements.
To join a proper debate, such measures must be quantified for various markets and various times, and the degree of uncertainty and departure from randomness must be ascertained. I have never found a movement in prices that anyone could make money with by a trend following method that didn�t also show a major departure from randomness revealed by the standard statistical measures I mentioned. The tragedy is the mysticism and blind acceptance of trendism, that trend following exponents proclaim, without any evidence as to magnitude and uncertainty. No self-reported results that selected individuals or leaders might have made in the past shed light on the debate.
Dave: Your well known saying, �If it can be tested, it must be tested� comes into play here . Exactly what testing have you done to prove the above idea?
Victor: These tests can readily be performed My group of colleagues performs these tests maybe 2-3 thousand times a year over different markets and time frames. Those of a cognitive bent and those with their feet on the ground are always open to the existence of trends, but they test them with the best statistical methods existing. If you apply these tests to stock market moves, you will find that all such tests show negative serial correlation. In fact, they indicate a tendency for reversal.
Dave: What about the upward bias in stock prices? Why cant that be interpreted as a trend?
Victor: Well, all proper statistical tests take into account this upward drift. They would look for serial correlations over and above the basic drift of the market. One of the other market cons is the permanent bearishness of some of market pundits, and I am the last person to say that this upward drift, evidenced over the last 200 years, does not exist. This in no way refutes, but it does refine the statistical tests required for the stock market. However, I hasten to add that no such upward drift exists in any other market. </b>
Any questions?
surf surf
