I rarely make changes to the gauges’ construction. The fundamental thesis is that they include all the necessary variables or inputs which drive future returns. I must then be vigilant and honest when we get into a situation where a new temporary variable is a factor that I have not included in the gauge construct.
We encountered such a situation after the last election when the prospect of significant trade tariffs by the new Trump administration did great damage to the bond market. I noted it in a blog and turned off associated algo’s that use the index. These are the first updates I have made in two years.
1. The Long-Term Stock Index/Gauge The propensity of the S&P to trend in a longer time frame is far greater than in a shorter (daily) time frame. The old version of the long-term index was reverting to a very low level during such trends so positions were only being held intact after buy signals by virtue of trailing stops.
I have allowed for a slightly heavier weight to the “trendiness” factor. You can see that this doesn’t represent a huge change. During trendy periods the gauge will be less inclined (ceteris paribus) to crash down to very bearish levels.The distribution is altered but the long-term average value of the index is not changed. We just won’t see such low gauge levels that might drive us to sell the market too early, during trends.
2. Daily Bond Index Adjustments The bond gauge has three main inputs:
- Bond Fear
- Stock market conditions (as a proxy for economic activity)
- Inflation – several variables have been used to measure this indicator such as crude oil, the US dollar, and even the utility (stock) index.
The time has come to recognize that the bond market simply doesn’t care about the impact of oil prices on inflation. Either the bond market doesn’t believe that rises in oil will be sustained or it thinks they will not be passed through to the economy at large.
It would appear at first that rises in oil are consistent with falling bond prices but in fact, they are coincidentally associated with declines in Vix. Normally when Vix rises oil declines – we see that benefit in the green line. The problem is that falling oil prices provide no edge with regard to bond performance. This exposes the weakness of the relationship. If we go back in time there was a higher beta and it could be improved by considering oil’s moves relative to associated changes in Vix. The time has come to deemphasize oil as a predictive variable. At the same time, I must stay on watch to see if the oil factor becomes relevant again in the future.
There is a logical association between US dollar rises and a reduced fear of (imported) inflation. We could even improve this variable by adding extra weight to the inflation-sensitive currencies like the Australian and Canadian dollars. Let’s look at the performance of bonds under different dollar scenarios:
The profits from shorting bonds when the dollar rises are really the result of a falling market. The gains from buying bonds when the dollar falls are de minimus. Performance using the inflation currencies as filters is not materially improved either even if you look at them on a relative basis. They too must, therefore, be deemphasized as inputs in the daily gauge. On a longer-term period, both oil and the dollar are relevant so no similar changes have been made to the weekly bond index.
Material changes in the construction of the gauges are rare and I prefer to avoid them. If I make any more, I will certainly update you.