|
PureBytes Links
Trading Reference Links
|
Hi Yngvi
I've been trying to dig out your posts on Ruggiero, no luck so far. Please
re-post is you have them.
The problem with pre-processing data is not "missing data" as most of the
programs assume but the lack of "attributes" for the available data. Most
traders are able to install a complete set of O-H-L-C (OI & VOL) data for
their favorite tradable.
Where the Knowledge Discovery Process breaks down is in programming the
"attributes" into a second Excel workbook (after the data workbook) and
testing for statistical significance for each of the "attributes".
Such "attributes" of the tradable data include:
Stridsman's up (down) moves: a) period move, b) length of move, c) amplitude
... for Daily, Weekly and Monthly data. Likelihood for periods in a row.
Percentage of moves of a certain length. Percentage of moves of a certain
amplitude. Correlation analysis.
Larry Williams has a whole chapter on data for "attributes", i.e., key
reversals, gaps, inside and outside days, etc
Walter Downs' patterns, Schwager's chart patterns, etc.
How an indicator reacts with the tradable, etc.
So there are lots of pre-processing approaches.
... will write later
Walter
|