as it depends on simulating the snooping process that might have been occurring. No formal tests can be
performed to answer this question, but figure 1 along with some historical facts about technical trading that
were given in BLL appear to defend the conclusions of Sullivan et al. (1999). BLL were careful to use rules
that had existed in the technical trading community for some time, and did not try to perform any extra
parameter tuning over their samples. Some of these rules have been in use since the early part of the century.
Given that they were not “tuned” to the previous 90 year sample, the results over the past 10 years are even
more interesting. This arbitrary feature of the data has changed dramatically, and in this context it looks
impossible that the past 10 years could be a draw from any 10 year period in the 90 year history. While it
is impossible to ever completely avoid the problems of data snooping the results here suggest that something
has changed dramatically