Posted on 9th Jun 2020 by the LSS team
The LSS US Treasury model has provided excellent (if somewhat boring) signals
One of the best performing - if perhaps least exciting - models this year has been our US Treasury model which maintained its recommendation to be long US Treasuries since December 2019. In fact, except for a quick trend break in December 2019 that lasted for all of 17 days, the signal from our US Treasury model has pointed in the same direction since November 2018, when the US 10 year yield peaked out at approximately 3.25%. In recent days though, yields have moved back up somewhat but the model hasn't budged. What gives?
The LSS US Treasury model has returned +19.7% YTD, in line with the market.
When we construct our models we maintain a certain degree of flexibility in terms of techniques, variables used and timeframe. This is in recognition that what works for some markets may not work for others, and while Bitcoin may go on a 100% return rampage or drop 70% in a matter of weeks (much of which may be missed out on if you calibrate your model on mid or long term trends), the same would be far less likely for government bonds.
Developed market bond yields have been on a steady downward grind over the last decades, with only occasional periods in which that trend is broken, largely in line with US economic growth. While changes can still come relatively fast and furious as could be seen during the COVID-19 melt-up and subsequent bounce back, in general, we find that models get sufficient time to adjust to the new reality by picking up on medium-term momentum shifts.
We don't just look at bond prices, but also at equity volatility and stock prices
When we look at the US Treasury market, we are not simply looking to pick up on statistical changes in the way bond prices behave. We also take into consideration equity volatility levels, the economic situation and how the stock market is faring, amongst others. While in recent weeks we have seen the S&P500 recover all of its losses from February and March, volatility remains elevated and the world economy obviously remains in dire straits. In a sense, our quantitative model is saying it doesn't fully buy the recent rally in risk assets and nothing has materially changed for government bonds.
We believe that this is one of the strengths of our government bond approach. It enables us to often capture most of the move and is not easily distracted halfway through. When yields came down from 3.25% to 3% the model switched and never looked back. Only when in late 2019 yields came up from 1.5% to 1.9%, it recommended staying on the sidelines halfway through to buy back in at 1.9%.
- Courtesy of LongShortSignal's Quant Team.