In my previous article I have explained why no market-timing indicator is good enough alone, then introduced the systemic risk indicator MTS4, similar to an avalanche danger scale for investors. I have presented an ETF strategy based on MTS4.
MTS4 is calculated as follows:
- If the unemployment rate is above its value 3 months earlier, then a=1; else a=0.
- If S&P 500's EPS estimate is below its value 3 months earlier, then b=1; else b=0.
- If S&P 500's 50-day sma is below the 200-day sma, then c=1; else c=0.
- If the 52-week sma of S&P 500 companies' average short interest is above the 104-week sma, then d=1; else d=0.
MTS4 is simply the sum a+b+c+d.
The problem with the avalanche danger scale is that zero risk doesn't exist. There are sometimes fatal accidents on low-risk days. I push the analogy with MTS4: a low systemic risk means that the environment is not favorable to black swans, not that they are impossible. The risk is low, but the damage may be the same if something bad happens. To cope with this, I propose to normalize the stock/bond ratio at 60/40 for the lowest values of MTS4. Here is an example starting from the previous article's model and adding 2 fixed bond ETFs to the portfolio. I have chosen ETFs in long term treasury bonds (NASDAQ:TLT) and mixed intermediate term bonds (NYSEARCA:BIV). Other choices are possible.
The portfolio is composed of 5 ETFs depending on MTS4 value.
For MTS=2, holdings are MDY,QQQ,IEF,TLT,BIV.
For MTS=3, holdings are MDY,IEI,IEF,TLT,BIV.
For MTS=4, holdings are SHY,IEI,IEF,TLT,BIV.
The underlying ideas are:
- Increasing the bond ratio when MTS4 goes up.
- Investing in various stock indices and keeping the less volatile as long as possible.
- Using shorter-term bonds at higher values of MTS4 to limit the volatility when the risk is high.
In the next simulation since 1/1/2001, positions are rebalanced once a week. Dividends are included. Transaction costs are not.
The return is lower than in the first version, but the drawdown and volatility are drastically cut. The Sharpe ratio jumps from 0.92 to 1.20, showing a better risk-adjusted performance.
MTS4 doesn't take into account the oldest market anomaly: seasonal patterns. Rather than counting it as a component of systemic risk, I propose to bet a 6th ETF position only on seasonals. This position would be in SSO on the 4 best months of the year (November, December, March, April), and UST the rest of the time. Beta-slippage is usually acceptable for these 2x leveraged ETFs. The database has synthetic prices before their inception dates. If you don't like them, this is almost (not exactly) like taking a double sized seasonal position alternatively in SPY or IEF with a portfolio leveraging factor of 1.167.
Backtests give clues to compare strategies in the past. They are not a guarantee for the future and don't aim at calculating expected returns.
Here is a backtest on the same period with the 60/40 adjustment and the seasonal rule:
The Sharpe ratio is significantly higher. With a 0.1% transaction cost, which is twice InteractiveBroker's rate (IEX:IBKR), the simulated annualized return is still above 12.6%.
MTS4 calculation rules are given above. I also send it for free once a week. Ask me here if you want.
Data and charts provided by P123.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.