Sounds interesting but what do you mean by "plain market data"? Something needs to get assigned to pitch. Do you mean treat the volume of each buy or sell as a click, then treat those clicks like samples of a wave?
yesEvery tick would be a sample. The volume could be used to do some stereo mangling (phase difference for example). Pitch shifting because the frequency range would be beyond the hearing range of 10 octaves.
Your approach isn't that bad either, read up on granular synthesis to perfect it
Your ideas about sound mangling are hideous. Hideous i tell you!
You cannot change phase by modulating amplitude (well, actually, to be completely nerdy, you do sortof distort the phase temporarily while the volume change is in the process of actually changing, but its nothing like you mean).
You don't want to pitch shift, you want to reframe the data into hearable frequencies, which in this case would be a simple rescaling of a frequency representation of the data.
Anyway, the above 'auralisation' is pretty useless.
No new information is presented in the audio that could not be observed in a single gaze at the actual data.
Also, it is horribly coarse.
It may be interesting to hear high res derivatives of the original data, but making the sound meaningfull is just going to be very difficult, no matter how you look at it.
