My mind keeps dropping back to the physics paper that describes gravity as an emergent property of a quantization system. The basic idea is that in an world of expanding entropy it is possible to capture some entropy into the work of quantization, making particles. The entropy in becomes quantization out plus entropy out, the entropy out being a bit less than the entropy in. The work of quantization has a coding gain that restricts the possible trajectories and mass of the particles out, and effect we call gravity.
I bring it up because our generalized yield curve is basically a black body radiator, so how can radiation losses become work? The answer is that the coding gain is the work, but each channel component in our yield curve will radiate entropy out in amounts that expand with the work done, so radiation losses are a consistent measure of work.
The bankers, remember, set yields according to past coding errors, past radiation losses. But the resetting of the yield to cover greater expected losses implies more gain from quantizing larger trades. See the connection? We invest according to optimum radiation, but we really take home the coding gains, which grow consistent with past loss corrections.
It becomes relevant because a paper referenced by Rajiv (the LeBaron paper) sets me on course to model the stock market trading system as quantization system within the optimum Shannon channel. The LeBaron paper was based on minimizing variance, and I will rework it minimizing redundancy (the complement of entropy)
I want to distinguish between coding gain and radiation losses because my model will use yields as variance measures, in the normal sense, but the gains from trade will be affected by the coding gain. Dunno how much work I will do, and I have no objection to some graduate student doing the work for us. So I will start doing that theory piece meal until I get a hit from some university somewhere.
My plan is to construct the generalized yield curve as a set of trader groups operating along the X axis. They gain by requantizing stock values to fit their trading bins, thus pushing the same information into less bandwidth. The optimization is to maximize coding gain and minimize radiation loss. The result should be the defacto model of the stock market, a fairly bold prediction.
No comments:
Post a Comment