Outlined below are the filters for the VT Model and their weightings.  In total there are 9 filters. They will trigger at various times. In effect, I am running 9 trend following strategies on VT.  I use 9 filters because if one or more do not work well, the overall returns will only be partially affected. Also, the results of the 9 combined will be smoother because they trigger at various times, some intentionally more often than others.

1.  Economic Filters (30%)

Filter

Weight

Negative annual change in real retail sales combined with negative Trend in VT

10%

Negative annual change in industrial production combined with negative Trend in VT

10%

US unemployment rate above 12-month moving average combined with negative Trend in VT  

10%

For an economic filter to trigger, there must be negative economic data, often associated with a US recession, and the “Trend” for VT must be negative (i.e. the market is trending down).   I measure Trend by the 5-DMA crossing the 210-DMA (10-month). I use total price returns.   If the market is trending higher despite negative economic information, the filter is not triggered, and the filter remains long VT.  I learned about these at the blog Philosophical Economics, which I recommend (link). 

The economic filters will trigger very infrequently, so 30% of the VT Model will be long VT most of the time.  Thus, this 30% will likely be subjected to less “whipsaw” than the rest of the filters. 

2.  Price Trend/Momentum Filters (30%)

Filter

Weight

5-DMA of VT below 210-DMA of VT

15%

Negative 12-month ROC in VT (smoothed using 5 DMA)

15%

These are pure price trend/momentum filters.  They are designed to capture any situation where the price of VT deteriorates to a certain degree.  Usually, large drawdowns are associated with recessionary periods, but this may not always be the case for whatever reason.  I expect these filters to generate more false signals than the economic filters and in choppy markets they will perform poorly, usually underperforming VT.

The trend and momentum periods are common periods used by trend/momentum followers.  There is nothing special about these period lengths. Other period lengths may or may not work better in the future.

3.  Sentiment Filters (20%)

Filter

Weight

50-DMA Equity Put/Call Ratio > 0.68

10%

Annual % Change in NYSE Margin Debt < 0 and negative Trend in VT  

10%

The Put/Call filter has historically triggered during periods of prolonged market volatility, which often coincide with poor risk adjusted returns.  Many people use high short-term put/call readings as contrary indicators, which often makes sense. However, this filter is more of a trend following filter.  

Change in margin debt outstanding often reflects a change in investor sentiment.  In the past, negative annual margin debt growth has coincided with poor market returns and higher volatility.  Like the economic filters, I use a double trigger. Margin debt growth year over year must be negative, and the Trend must be down.  

4.  Credit Spreads (10%)  

Filter

Weight

Negative 3-month rate of change (smoothed using 5 DMA) in the ratio between the ETF JNK (high yield bonds) and the ETF IEI (3-7 year US treasuries)

10%

Often corporate credit deterioration precedes weakness in the equity markets.  This indicator is designed to capture such weakness before it translates into weaker equity prices.  As such, it is normally an early filter.

5. Market Breadth (10%) 

Filter

Weight

Less than 50% of NYSE stocks trading above their 200-day moving average (smoothed using a 5 DMA)

10%

This filter captures weakness in market breadth which often precedes, or coincides with, price weakness.  It will typically trigger during minor corrections as well as bear markets. As a result, like the price trend/momentum filters, it will be whipsawed and do poorly during choppy markets.