After my last post I decided to change my plans for the day and run a backtest on the model discussed in my last post. I ran the test from June, 2000 to present and each time my US timer gave a long signal the model bought the two ETF’s having the best prior six-month performance and went to cash on a sell signal. This model produced an 8.5% compound annual growth rate with a 21.7% maximum drawdown versus a compound annual growth rate of 2.1% and a maximum drawdown of 59.9% for the Russell 2000. However, a model based on simply buying IWM (iShares Russell 2000) each time my US timer went long and going to cash on a sell signal produced an 11.6% compound annual growth rate with a maximum drawdown of only 14.3%.
There may be a way to improve the two sector ETF model: instead of holding the two ETF’s for the duration of the long signal we could run the selection process every month and switch ETF’s as necessary. In order to match the IWM model, this monthly switching would have to add 3.1% per year. That’s a study for a rainy day! For now, I’ll leave my IWM/RWM model on Collective2 as it is – hold IWM when my US timer is long and hold RWM when my US timer is short.
FJP
