I may be missing a point here, but it seems we assign some TA positive properties that other TA apparently lacks. For example, splines = good, moving averages = bad. Discarding the obvious mathematical advantages splines have what makes them any more a metric of anything worth measuring than a moving average?
I guess lol Well, I'll be posting a full blown T.A. in a few mns... quite amazing considering this can be seen as a perfectly RANDOM time/price series...
Well well, here it is... Hard to tell this is out of a stochastic data series generation. The double-bottom accurate to the "tick" (the 0.5 granularity of my experiment) is staggering. I'm hating this...
I wonder if this isn't a distinction without a difference. If a trader has a profitable methodology built around price patterns -- and many certainly claim to, in this thread and elsewhere -- does it matter to his P&L or future prospects that he's not making direct use of what you describe as the underlying stable distribution? Isn't the fact that these patterns do exhibit stable behavior good enough? To return to your car on the road analogy, one could drive a car quite well without knowing anything at all about the intricacies of what goes on inside the cylinders. Your thoughts?
How often do these patterns fail though? There have been a string of threads here for months where people claim "the top is in" based on technicals such as head and shoulder patterns. They have all failed. I wonder if there's any edge in placing bets against the direction of a pattern banking on a short squeeze? Then again I don't know how much of volume is made up by retail traders but I assume it is not a huge percentage. Probably still a coin flip.
good questions. When one takes their framework and looks for instruments in which their framework makes money - they are acting less efficiently but similarly to someone who is running lots of models across their database. Our easiest profits were manually arbing the major NYSE s&p stocks back into alignment with their typical value relative to the s&p. Or sometimes when we were in big up or down we would just pick and intrument we expect to out perform the s&p and buy the next pullback or short the next rally. We were eventually replaced by ecns eminis and computers. But our moving averages were very useful for comparison with the s&p. For the very reason I knew what were doing I am blown away by people who think that analysing a database is any different that putting up bollinger bands and moving averages. Properly used its the same thing. But, when some people happen to expect the points to mean something... they frequently do.
Here is a insight about flocking (my type). I thought about this more. And now I see why I was attracted to this flocking argument. We put up 20 ema on 1 minute and 5 minute chart. We compared our stock to the s&p. We would look for points we expected the s7p to bounce. if it did we then bought or shorted our stock. We beat the rest of the flock. here is a link to something my ex business partner wrote at trading markets showing what we had been doing since 1997. http://www.tradingmarkets.com/.site/stocks/education/lssnoftday/08272002-27276.cfm The citi chart in the middle of the article is what really was up on our screens all the time. And we pulled buckets very consistent money out of the market. By the way - his Floyd Numbers - just the floor trader pivots applied to multiple time frames. (another trader and I showed them to Dave)
" Can linear regression analysis really predict the future? " No, it can only come up with some "best fit" for the past.