The momentum-volatility rankings system provided on this website was designed to protect against downside market moves. This orientation was adopted because of the strong belief that we live in dangerous financial and economic times. The possibility (likelihood?) of a financial collapse is real and a consideration that should be reflected in all trading/investment endeavors.
A recent conversation with a subscriber touched on this subject and the downside bias built into the system. The subscriber asked me to quantify that. I could not and realized that I had never formally verified this belief. The purpose of this article is to do that.
The momentum-volatility algorithm was developed because of strong skepticism regarding current financial markets. Economic and financial conditions were and continue to be ugly.
The irresponsible behavior of the Federal Reserve and the US government and their International counterparts have driven markets beyond normal levels. Financial repression and liquidity injections are likely to continue to provide financial asset inflation.
Obviously, nothing can continue forever, but for the foreseeable future government intervention is unlikely to cease. Ultimately external events (market driven) will put an end to the game. Before then, it is likely that QE will be increased before it is decreased.
Those of us with some seniority have no business in today’s equity markets. But financial repression has made it impossible to have income via traditional fixed income markets. In an attempt to survive this period, the momentum-volatility ranking system was developed.
Momentum is a well-known technical tool used to ensure positioning in the direction that the market is moving. The rankings system does not engage in shorting, however it does signal market exits when conditions warrant. That and a volatility add-on appealed to me as a workable alternative to sitting in bonds and watching one’s portfolio rust away.
The need to do something didn’t change the risk of doing it. I likened trading in these markets as akin to trading to Armageddon. That reality had to be incorporated into the design of the rankings system.
Two key considerations were the following:
- Downside corrections are faster and more violent than upside.
- Markets are overvalued as a result of government interventions and making such corrections more frequent and likely.
These two factors influenced the design and made a short-term orientation necessary.
A monthly system was determined to be optimal. Quarterly rankings did not react quickly enough to protect on the downside. Nor did they produce the returns that a monthly system did. Semi-monthly did not produce meaningful different returns and roughly doubled the number of trades required.
A month can seem like a long time in a severe market plunge. Some intra-month protection was deemed necessary.
There was no way to incorporate early exits into the software so stop-loss limits provided a means of intra-month protection. Unfortunately, it would be impossible to quantify their effects. The system only allows for the purchase of ETFs at the beginning of a month and then their disposition at the end of the month. Suggested stop-loss limits are recommended for all ETFs. I would not be in these markets without some form of downside protection.
Stop-loss limits kick in when markets turn down in general or for particular ETFs in particular. I suspect, but have no way of documenting, that stop-losses reduce the reported losses generated by the software each month. Only in the instance where an ETF hits its stop-loss limit during the month and then turns around and finishes higher by month end would that not be the case. The possibility of that occurring depends upon market behavior and the placement of stops.
The data below are based on buying and holding for a full month. No effect of stop-loss protection is reflected.
In designing the algorithm, decisions to minimize losses were incorporated where possible. Choices in the timing of ranking re-balancing, the lengths of variable calculations and the internal weighting of inputs all were biased to toward that end. Downside losses were to be minimized. Maximizing upside gains was of secondary importance.
Believing this to be true and achieving it are two different things. Proving it is another thing. What follows are my findings regarding upside and downside performance.
Downside Vs. Upside Performance
Overall Cumulative Performance
The cumulative returns of the one-month buy-hold and then sell strategy for the period from January 1, 2007 through December 6, 2013 follow:
- US(6): 173.7%
- International(6): 168.9%
- Inflation-hedge(3): 128.0%
- Overall(10): 182.6%
Each of these represent collections of rankings available each month. The parenthetical numbers are the maximum number of ETFs included by portfolio. The returns were achieved via backtesting.
If enough ETFs did not meet the minimum threshold, cash replaced them in those months. Depending upon the month and the portfolio, cash ranged from 0% on up to 100%.
A traditional buy and hold strategy for the period provides a reasonable benchmark. For US, the return of SPY, an ETF for the S&P 500 is appropriate. For International EFA was used. Both of their returns were substantially lower than any of the strategies above. SPY returned 47.2% for the period and EFA returned 8.8%.
Returns in Up Markets and Down Markets
To determine the performance in up or down markets, the following methodology was used. If the appropriate benchmark (SPY or EFA) was up in a particular month, that was considered an up month. Down months ere and under what market conditions returns were generated was determined by reviewing the data on a monthly basis. The US and International portfolios were used. For the relevant portfolio, the incremental performance was listed. So, as an example, if the benchmark was up 2% that month and the portfolio was up 1%, a -1% was listed in the up-month column. If the portfolio had been up 4%, then a +2% would be placed in the column.
Similar calculations were made for down months. If the benchmark were down 1% in a month and the portfolio were breakeven, a +1% would be entered into the down column.
To envision this, for each portfolio there were 84 months that were classified as either up or down depending upon what the benchmark (market) did that month. In each column was an incremental number was placed that reflected the portfolios bettering or missing the market return.
Adding the two columns up provided the summation of performance in up markets and in down markets. This difference was the effect of the portfolio strategy as opposed to a buy and hold of the market.
Here are the results:
|Up Market||Down Market||Total|
These numbers represent the cumulative portfolio effects over and above what markets did. These are not returns, although it might be proper to think of them as incremental returns produced by the portfolio(s).
The bottom line is that in up-markets the portfolio did not outperform the market. In down markets it did. Overall, it outperformed markets.
One of the important reasons why the portfolios performed so well in down-markets was the ranking threshold. When not enough ETFs passed the minimum requirement, cash was substituted for some or all of the ETFs that month. As markets get become more volatile and returns diminish, fewer ETFs exceed the hurdle.
Cash always outperforms in down markets.
Nowhere is that more apparent than the 2008 – 2009 stock market collapse. In just three different months (not consecutive) during this period, the US portfolio gained 30 percentage points relative to SPY. Actually, the portfolio gained only one percent in these three months, but it avoided losses in the SPY of 29%.
Cash is good in down markets and the algorithm is quick to go there when rankings deteriorate. That is a good thing for someone more concerned about protecting what he has rather than growing it. Although, the overall cumulative performance above shows dramatic growth relative to a buy and hold strategy.