Scary stuff.
If someone has a five color graph coloring, then we can approximate a five color currency pit. Drop in dollars, withdraw yen. Add on Euro and out yuan. and pound.
We have five dual queues with paired exit and entry for each currency. Match each pair separately,then whiten the interleave of the five. Their 'tint'is the spectral density, how much volatility does it need to emit the typical sequence against the other four, treated as an aggregate uncertainty. Then implement a swap protocol to keep the trade space dense across tints. We should get close to a five coloring.
This is simply the five way queue matching. Go read that five coloring paper.
In our case, for each of the five you get the bit error matching, so you end up swapping order across bit error sequences to remove observed correlations, keep conserved trade space maximally packed (sphere packing again) . But that means doing some swaps in the structured distribution, reversing the order of some swaps. It is like a backward revision in prices.
There should exist a connecting graph of all currency pairs. Properly colored the graph is as close to white as possible Links, between nodes, generate events between adjacent nodes, maintaining its proper tint for each node. Read the paper, but my interpretation fits the pit method above. The five color whitening is as if the node was white but emits adjustments along its links to keep its own tint. Sandbox does good coloring.
Five color binder, the optimal FX tool, none better. Also quite geeky stuff.
No comments:
Post a Comment