This is an update to the Trend-Momentum strategy that trades the All Weather ETF List. This article starts with a review of the strategy and the importance of the All Weather ETF list. We then examine some signals to understand the methodology and update performance. I expanded on some key points so there is something new, even for those who have read the prior articles.
First introduced in February 2022, this strategy benefits from bull markets by focusing on stock-based ETFs when the Composite Breadth Model turns bullish. It shuns stock-based ETFs during bear markets by shifting its focus to non-stock ETFs (alternatives). The broadness of the All Weather ETF list and the ability to shift focus allow for true diversification across different asset classes.
I will add a video for this article later in August (ran out of time in July).
Trend Composite Strategy Series
- Introduction to the Trend Composite and its Indicators (1-Feb)
- Trend Composite Signal and StochClose Ranking Table (2-Feb)
- All Signals Tests and Indicator Comparison (8-Feb)
- Trend Portfolio: TrendComp Crosses, StochClose Rank, Mkt Regime (15-Feb)
- Momentum Portfolio: Trend Status, StochClose Rank, Bear Mkt ETFs (22-Feb)
- The All Weather Portfolio, Bull-Bear Market ETFs, Market Regime (17-Mar)
- Adding an ATR Trailing Stop and Testing Time Exits (24-Mar)
- Live Example as Composite Breadth Model turned Bullish (28-Mar)
- Adding a Profit Target (31-Mar)
Note that the Trend Composite is part of the TIP Indicator Edge Plugin for StockCharts ACP. I use Amibroker because this is where I can build the indicators, test the indicators and fully customize the chart views.
Setting the Stage
The backtest period extends from January 2007 until June 2022, which covers 15 and a half years. As the chart below shows, this period includes the Global Financial Crisis (GFC), a few small bear markets, the covid crash and the current bear market. It is important to cover different environments when testing a strategy and this period has it all.
Other testing criteria include:
- All Weather ETF List for universe (50 ETFs)
- 14 Equal-weight positions
- Starting Portfolio: $100,000
- Signals based on closing prices
- Buy/Sell based on opening prices the next day
- Commission per Trade: $5
- Slippage per Trade: 2.5 basis points (.025%)
- NO Dividends
There are no dividends included so actual returns might be slightly higher. To err on the safe side, I added a small commission for each trade as well as some slippage. These ETFs are very liquid so I suspect that slippage would be minimal. The signals are based on closing prices with the buy/sell occurring on the next open (one day delay).
The performance tables show the Compound Annual Return (CAR), Maximum Drawdown (MDD), Win Rate and some other metrics. The Gain/Loss Ratio is the Average Gain divided by the Average Loss, which is pretty straight forward. Trend-followers would like to see this ratio above 3. Profit Factor equals the number of winners multiplied by the Average Gain divided by the number of losers multiplied by the Average Loss. A Profit Factor of 2 means total profits are outpacing total losses by a two to one margin.
Setting the Benchmarks
The first table shows results for buy-and-hold SPY, buy/sell the 5/200 day SMA cross for SPY and buy-and-hold for the 50 ETFs in the All Weather List. The Compound Annual Return (CAR) was the same for Buy-and-hold SPY and the 5/200 cross (green shading), but the Maximum Drawdown (MDD) was much higher for buy-and-hold (56.5% vs 20.2%). Even basic timing can help lower drawdowns and achieve similar returns. Buying and holding all 50 ETFs produced a lower return and a high Maximum Drawdown (red shading).
The PerfChart below shows the six biggest losers since 2007 and the six biggest winners. Not everything goes up over the long-term. The Oil & Gas Equipment & Services ETF (XES), Clean Energy ETF (PBW), Gold Miners ETF (GDX), Yen ETF (FXY), DB Agriculture ETF (DBA) and DB Base Metals ETF (DBB) are down over the last 15 and a half years. Five tech-based ETFs are up more than 480% (XLK, SOXX, FDN, IGV, QQQ) and the Medical Devices ETF (IHI) is up over 500%. I think we know which ETFs are driving the returns.
Testing the Trend-Momo Strategy with Daily Signals
The table below shows the current performance metrics for the Trend-Momo Bull-Bear Strategy using the All Weather ETF list. The Compound Annual Return (CAR) is 9.45% with a Maximum Drawdown (MDD) of 15.9%. There were 272 trades with a win rate of 53%. The average gain (20.4%) was more than four times the average loss (4.6%) and the Profit Factor was above 4. Winning positions were held an average of 277 trading days, while losing positions were held an average of 73 days.
Overall, the results for this strategy are good. As with most trend-momentum strategies, a few good trends juice the profits and drawdowns are contained by keeping the losses small. 16 of the 272 trades (5.8%) gained over 50%, 18 (6.6%) gained between 30 and 50 percent, and 17 (6.2%) gained between 20 and 30 percent. Roughly 19% of the trades (51) produced gains greater than 20% and these trades accounted for the lion’s share of the Compound Annual Return (CAR). As the Profit Distribution shows, the vast majority of losing trades (80) lost less than 5%. Let the profits run and keep the losses small.
Charting Performance Metrics
The first image shows the equity curve since 2007 (start = 100,000). The system was flat into early 2009 because it was out of stock-based ETFs during the 2008-2009 bear market. The equity curve rose when the bear market ended and the Composite Breadth Model turned bullish in early May 2009. The equity curve pretty much zigzags higher along with the S&P 500 buy-and-hold (blue line).
Containing drawdowns makes a difference to the bottom line and the stomach. Drawdowns using this strategy were less severe than drawdowns for buy-and-hold and basic timing. This is because the strategy does not trade stock-based ETFs during bear markets and preserves capital. As such, equity drawdowns are capped and total capital is higher when the bear market ends. We will also be less likely to suffer ulcers. The red arrows show steep drawdowns for buy-and-hold in December 2018, March 2020 and June 2022. Drawdowns for the strategy were contained (not eliminated).
The next image shows the equity drawdowns over time. This is the percentage decline from a high in the equity line. There were some steep drawdowns in 2010, 2011 and 2012 ( greater than 12%). Drawdowns since 2013 have extended to the 10-12 percent range and the strategy is currently experiencing a 12% drawdown from the equity high, which was in April. For reference, SPY was down around 20% from its early January high to its June close. The equity line for the strategy hit a new high in April and experienced a big drawdown in June. Drawdowns are part of the process and pretty much unavoidable.
The next table shows the yearly and monthly returns for the strategy. On average (bottom line), all months are positive. May, June, September and October were the least positive months (weakest). The red shading highlights some sizable down months with May, June and September experiencing a string of bad months. It is important to see these so we understand the strategy and set realistic expectations. There will be some pain to get the gain. The green shading highlights some of the bigger up months.
Testing First and Last Day of Week Entries
The backtest above is based on daily signals, which can be any day of the week. Not everyone can or wants to watch the market every day, especially if we can get similar results from weekly signals. I have not had much success with signals based on weekly data, but I have had some success using daily data and trading at the end of the week, Fridays for example. The next tests will look at trading on the first and last day of the week.
Most weeks have five days so the first trading day for entries and exits would be Monday and the last trading day would be Friday. Weeks without trading on a Monday would have Tuesday as the first trading day, while weeks without a Friday would have Thursday as the last trading day. For explanation purposes, I will simply refer to Monday as the first trading day of the week and Friday as the last.
Trading on the first trading day of the week means signals are based on the last trading day of the prior week. There is always a one day delay between the signal and the entry/exit. For example, we take signals based on Friday’s close and the entry/exit is the following Monday. Trading on the last day of the week (Friday) means signals are based on the previous day (Thursday) and the entry/exit is Friday.
The results below show that trading on the first or last day of the week does not help performance. In fact, the Compound Annual Returns (CARs) are lower and the Maximum Drawdowns (MDDs) are higher (red shading). It is interesting to note that the last day of the week (bottom line) performs better than the first day of the week (middle line). In other words, buying/selling on Friday performs better than buying/selling on Monday.
Highest StochClose for First/Last Day Entries
Buying on Friday means using signals from Thursday’s close, which means we are ignoring signals and StochClose ranking values from the previous five trading days. This does not usually affect in-state Trend Composite signals, but it can affect the StochClose ranking. In other words, the ETFs with the highest StochClose values on Thursday may not be the same as the ETFs with the highest StochClose values on Tuesday. Keep in mind that the strategy buys up trending ETFs with the highest StochClose values and StochClose is the momentum score
As a remedy, the idea is to identify the ETFs with the highest StochClose value during the prior five trading days (StochClose H5). The entry-exit day is Friday and signals are still generated on Thursday. Instead of taking just Thursday’s value, we take the highest StochClose value for the week (including the prior Friday). This means we choose the week’s strongest ETFs, not just the strongest ETFs on the signal day (Thursday).
The table below shows performance when selecting upward trending ETFs with the highest StochClose value over the prior five trading days (StochClose H5). Trading on the Monday means the highest StochClose for the prior week (Monday to Friday). Performance when trading the first day of the week is uninspiring, but trading on the last day of the week appears worth considering.
As the green shading on the table above shows, trading on Friday produced a Compound Annual Return (CAR) of 9.93%, which is higher than CAR for the daily signals (9.45%). The Maximum Drawdown (MDD) is 16.62%, which is higher than the MDD for the daily signals (15.9%). The Average Gain/Loss Ratios are about same, while the Profit Factor is significantly higher (4.81 vs 4.16). Overall, basing signals on Thursday’s close and trading on Fridays is worth considering.
Conclusions
The Trend-Momo Bull-Bear ETF Strategy beats the returns for buy-and-hold and has significantly lower drawdowns. Even so, there will be drawdowns along the way. Backtests and the resulting performance metrics can help up set realistic expectations. The current drawdown is in line with prior drawdowns and the hypothetical portfolio is down to four positions (UUP, DBE, XOP, XLE) after the June decline (28% invested). More positions will be added when the Trend Composites turn positive and the Composite Breadth Model turns bullish again.
The differences between the daily signals and trading on the last day of week (StochClose H5) are negligible. This is interesting because it suggests that we can achieve similar results by trading just one day a week (Fridays). As such, I will look into developing a signal and ranking table for trading this strategy with the All Weather list.
The usual disclaimers apply for trend-following and the analysis on TrendInvestorPro. Past performance does not guarantee future performance. You and you alone are responsible for your investment and trading decisions so do your own due diligence.