Portfolio Update

In last week's episode, I had just finished a portfolio model, and included its output for the data as of the Friday before. Since then, I've reworked it a couple times, to make it easier to update with new price data, to be able to set the number of issues in the output, to make suggestions for using short funds during sell-offs, and to improve performance by modifying the way allocation amounts are computed.

I think I'm on version three by now.

For the initial model recommendations, I put together a test portfolio using Google Finance, assuming that the entry price would be the average of last Monday's high and low. Over the first week, it gained a little more than 2%. I also built a benchmark portfolio, allocating equal shares to DBC (commodities in aggregate), VEU (all world equities), and BND (the universe of bonds). That baseline also returned a little more than 2%, but less than the test portfolio.

This week, new latest version of the model produced the following allocation:

Commodities (45%)
  13% JJT (Tin)
  13% BAL (Cotton)
   5% SLV (Silver)
   5% FUE (Biofuels)
   3% JJN (Nickel)
   3% JJC (Copper)
   3% DBB (Base Metals)

Equities (55%)
  12% EPU (Peru)
  10% TUR (Turkey)
   7% THD (Thailand)
   6% INP (India)
   5% KOL (Coal Producers)
   5% EWO (Austria)
   5% EWH (Hong Kong)
   5% BJK (Gaming and Casinos)

Again, I updated the test portfolio using the average of Monday's high and low as the purchase or sale price. As of the close on Thursday, the test porfolio had gained 4.62% overall, while the benchmark portfolio had risen only 3.04%. I'm rather pleased with that.

I'm also very leery of the current market environment. It feels overextended, and I'm expecting a pullback fairly soon. This rally in everything is the result of the dollar dropping rather dramatically, and if correlation were causation, then it would be the result of investors discounting the effects of QE2, expected to be soon issuing from the Fed. There's a Fed meeting on November 2-3rd, and the election on the 3rd. I expect that this nice little run will come to a sudden stop in the last week of October, as traders take profits in front of some obvious uncertainty.

I back-tested the model through the 2008-2009 market crash, and while it performed much better than the benchmark portfolio, it still had a peak to trough drop of 10-20% in the most favorable part of the parameter space. There'll probably be at least one more episode like that in the next 2-3 years, so a little pullback to see how it performs in real time would not be such a bad thing.

On Saturday, I'll have the new allocation based on today's data. I'm curious to see haow the positions held change, and what the changing positions say about aggregate investor behavior.