Reserve currency

What is it? Lets do what Mish does and go with the Investopeida definition:

"A foreign currency held by central banks and other major financial institutions as a means to pay off international debt obligations, or to influence their domestic exchange rate."

There are two parts, liquid and stable makes a reserve currency. The stable currencies are liquid over the largest domain, normally that is the US. By having the largest, liquid economy, the Dollar changes its relative value the slowest and remains liquid. So central banks do not have to constantly trade currencies over the global domain, they get a good representation value.

The world is changing, however, in one respect. Robots can do the trading for us. So liquidity and a measure of uncertainty are the two things we need to hold currencies over time. We need to measure entropy so the robots can propery trade currencies and get a measure of the global economy, no longer relying on a single large economy.

It is relativly simple to measure entropy if the robot has access to transaction data and can run the equivalent of a Huffman encoder. In the new world we have robots distributed to each of the central bankers, and these robots constantly measure the amount of mutual entropy among the currencies, and make trades to maintain minimal redundancy. As simple as that.

What makes this work is relatively transparant trade of goods, intra and inter national trade. China does not have that, the US does, Japan does, Europe does, Brazil does. There are no more reserve currencies, they are gone. There is simply the measure of transparancy in  the various economies. Gold really doesn't work, it is too heavy to trade, and so there is always a paper trail between the gold and the trades.

You cannot beat the robots as long as you can gather good trading sets.

If you need a measure of mutual entropy, follow my blog and I will dig it out. Basically my approach is to constantly measure the combined real yield curve of all currencies by mutual quantization, a multi-variate Huffman encoder. Then the central bank need only trade to make the holdings a close match to the measured yield curve. Probably the best way to do this is the same method I do with the SP100 trades. I put them all into a group and encode them as one. Individual currency trades, then matched to the cobine curve generate trading vectors, directional trades. In fact, there will exist the ideal yield curve for all currencies, treated as if they we a single channel. That ideal curve will be the matching fibnacci set, just rrade each currency against the ideal.

Now, what can I use for the trading values. Probably the price of oil, copper, gold, wheat, and other internationally taded commodities would be a good start. Any graduate students out there? Let's see, I need my SQLite to hold trading sets, my R Project and SQL interface, and my Huffman encoder. Check, now I just need more data.