TA is still toast
I have nothing against TA per se. However, like weather forecasting, it requires excellent models and serious computing power to be done effectively. Most, (arguably all) TA you will find in the public domain is too simplistic to be really useful .
The weather forecasting analogy is very useful. In San Diego, I can quite easily make the following prediction
"Late night and early morning low clouds, burning off in the afternoon. Hazy sunshine on the coast, brighter inland. Temperatures
low 70's on the coast, approaching 80 in the inland valleys."
With this forecast I would be correct 70% of the time. year round (if you live in SD, you know this is true). Trouble is, I'm only interested in when it rains!
To predict rain accurately, I then need the resources of the Meteorological Office: their thousands of Weather Buoys; their aircraft; their mathematical models and their supercomputers.
Alternatively, I can buy a farmers almanac, hang up a pine cone and some seaweed, and divine the weather.
The pine cone and seaweed may have some utility, since they actually do react to moisture in the air, but they can be fooled, say by fog - a common occurance.
I'm suggesting that quasi-technical terms like stochastic, momentum, bollinger band, MA etc etc are little better than seaweed and pinecones.
Real TA is done on supercomputers, sampling thousands of data points
in real-time and feeding models that have taken years to develop.
Thats why TA (the type posted here) is toast. |
