Correction. The comparison was 0.8242/0.8179 (N/50*N), an extraordinary convergence and will within sample error for N.
Even more astounding, to me, the pattern frequency was also very similar, which is counter intuitive ... I expected a much lower frequency occurrence for the 'bullish pattern' in a bear market. --- In [email protected], "brian_z111" <brian_z...@...> wrote: > > Just a heads up for newcomers. > > 2008 was an extraordinary year that will be in our databases for years to > come. > > IMO it is worth preserving a database for the year ... it is a lot easier to > do it now than to try and isolate the 'bear year' data later. > > It can be very instructive to run tests in a contrary year. > > For example: > > - I ran a bull pattern on 2008 bear EOD data > - Yahoo adjusted ... daily ...S&P500 == a - 50% bear market > - it is a high frequency pattern so N (sample number) was very high > - it is not a tradeable pattern as such ... only indicative of tendencies > - after the event occurred the liklihood of X occurring was 0.8411 > (probability) > - tested on JimSwindles Yahoo US database ... NYSE stocks 1997 - 2007 == a > +100% bull market with a significant down period in the middle of it > - sample size was approx 50 * the previous N and the probability of the > reactive event occuring was 0.8179 > > To me that is an optimistic result i.e. the bullish effect is persistent in a > bear market and consistent between a bull and a bear market, with a > significant N. > > It will be relatively difficult in the future to obtain an intraday database, > for the bear 2008 bear cycle.... worth keeping if you have one already? >
