Remarks on reply to Johansen’s comment :-)

Any reader of the paper by Laloux et al. [1] would have been left utterly confused after consulting the time series for the price of Japanese Government Bonds (JGB) around May 1995 looking for a log-periodic power law acceleration in the data. The authors of [1] now confess their error and admit that the prediction date given in [1] was incorrect and that the correct prediction was for August 1995. However, they now report [2] that the analysis leading to the prediction was made in May 1995. Obviously, this is also incorrect since the data used in the analysis, see fig. 1 of [3], does not start before June 1995. That the authors of the reply (LMAB) have severe problems regarding dating certain events becomes even clearer reading the second section of the reply. Reference [3] of the reply (corresponding to reference [5]) is essentially a 19 page answer to the criticism put forward in [4]. Nevertheless, LMAB writes that I and Didier Sornette (JS) ignores the work of J.A. Feigenbaum! Let us now turn away from the prediction experiment in question as well as the issue of whom cites whom and address the physical issues raised by LMAB. That crashes occurs for a number of different reasons and unfolds in various ways is quite obvious. For example, the bond crash of October 1998 mentioned by LMAB is generally explained as a result of the heavy investments made by German Banks in Russia. When the Rubel crashed, so did the Bundes bond. Other crashes triggered by external events are the ones on Wall Street in 1973 (OPEC oil embargo) and 1974 (Nixon's resignation). However, no consensus regarding the origin of the worldwide crash of 1987 exists. This clearly illustrates that crashes occurs for a variety of reasons of which speculative bubbles are just one. What we (JS) has argued in a number of papers [5, 6] is that speculative bubbles on the financial markets are more often than not quantifiable by a log-periodic power law acceleration ending in a crash or a large correction. That the crash/correction is not certain was already made clear in 1998 [7] where a probabilistic framework was developed. I wish to stress that such a probabilistic framework is essential in order for our hypothesis to make sense. In the same paper, an explanation to why that the time of the crash/correction predicted by the governing …