Tactical Composite Trend Model Update
The goal of today's Tactical Composite Trend Model update is to share some insight into the model-driven process and provide a current-day outlook. In my introduction email to subscribers, I shared some remarks about the Tactical Composite Trend Model. I want to start today's note by expanding upon my comments with some further thoughts and details.
- The TCTM is not a black-box model.
- I use a hybrid approach - man + machine.
- While the model follows some general rules, I believe in having as much flexibility as possible.
- When I set out to build the model, I did not intend to avoid every drawdown. i.e., the risk, recession, and long-term trend models identify significant risks.
- The long-term trend indicator acts as the default market exposure model in the absence of any composite model signals.
- I take a regular and methodical approach to incrementally improve the model.
- The TCTM should act as a complementary input into your research process and nothing more.
Please note, the team is working on adding the TCTM to the website. As of now, you can find some general information in the knowledge base.
Before I review the current status of the TCTM, let's examine the 2018-19 cycle as it provides an excellent example of the model-driven process.
Tactical Composite Trend Model - 2018 Example
- 9/20/18 - S&P 500 Closing High
- 10/1/18 - Risk Warning Model Count = 50% (signal threshold level)
- 10/8/18 - Risk Warning Model Signal - Raise cash (SPX -1.58% from high)
Comment: The model did not trigger on 10/1/18 because it contains a condition that requires the Index to exhibit negative momentum before activating.
- 12/17/18 - Long-Term Trend Model Signal (bear trend) Raise cash optional
Comment: When the long-term trend model turns negative, I check the washout and recession model signal counts. If the washout signal count is rising and the recession count is zero, I will most likely take no action. Historically, by the time the long-term trend model turns negative, the SPX is already down on average 12% from a peak. I don't want to raise more cash in what is most likely a growth scare correction. If the washout model count is zero and the recession count is rising, raise more cash.
- 12/24/18 - S&P 500 closing low
- 12/26/18 - Composite Washout Model Count = 50% (signal threshold level)
Comment: If one has short exposure, cover half.
- 1/4/19 - Composite Washout Model Signal - Decrease cash and cover any additional short exposure.
Comment: The model did not trigger on 12/26/18 because it contains a condition that requires the Index to exhibit positive momentum before activating. I never want to catch a falling knife. The signal overrides the bear trend condition for a window period with a stop.
- 1/18/19 - Composite Thrust Model Signal - Decrease cash
Comment: The signal overrides the bear trend condition for a window period with a stop. One can add leverage on a thrust signal for a window period.
- 4/10/19 - Long-Term Trend Model Signal (bull trend)
Comment: The goal is to be 100% invested once the thrust model signal occurs. However, if a thrust signal did not trigger, one could decrease cash on the long-term trend model turning positive.
- 4/12/19 - Composite Confirmation Model Signal
Comment: The goal is to be 100% invested once the thrust model signal occurs. However, if a thrust signal did not trigger, one could decrease cash on a CCM signal.
Note: A window period is a time after a composite signal when one ignores another model condition. Signal statistics determine the window period.
2018 Chart Example
Tactical Composite Trend Model - Pandemic Crash to Current Day Review
- 2/19/20 - S&P 500 Closing High
Comment: The Composite Risk Warning Model signal count reached 20% when the S&P 500 Index peaked on 2/19/20. The risk-off signal threshold is 50%.
- 3/16/20 - Composite Washout Model Count = 50% (signal threshold level)
- 3/26/20 - Composite Washout Model Signal
- 3/31/20 - Composite Recession Model Signal
Comment: The composite recession model signal triggered after the washout signal and against the backdrop of a positive trend condition. Given the washout signal, one would ignore the recession signal for a window period.
- 6/03/20 - Composite Thrust Model Signal
- 8/26/20 - Composite Confirmation Model Signal
- 11/11/20 - Composite Thrust Model Signal (follow-on signal)
Current Day Chart Example and Comment
As the chart below shows, the market is following the normal TCTM cycle with the Washout, CTM and CCM signals since the low in March 2020. Please note, it's unusual to see the long-term trend model (green line in SPX chart) stay positive after such a large drawdown. However, because the V-shaped crash and recovery happened so fast, the model did not have time to adjust and turn to a bear condition.
Composite Washout Signal Performance
As the table below shows, performance has been robust since the washout signal on 3/26/20. Results finished above the median on all but one timeframe, with the 1-year still to be determined.
Composite Recession Model
Typically, the composite recession model does a pretty good job of registering a signal when the S&P 500 is still within spitting distance of the highs. The forced shutdown of the economy resulted in an untimely recession warning. As I stated above, one would have ignored the recession warning for a window period as the composite washout model had just registered a signal.
Composite Thrust Model
As the table below shows, 80% of the composite thrust model components registered a signal post the pandemic low.
Composite Thrust Model Signal Performance
The pandemic crash and subsequent speedy recovery to new highs wreaked havoc on several indicators as prices did not have time to adjust down like they usually do when markets go through a bottoming process. As a result, signal performance lagged historical measures in the 2-4 week timeframe as the S&P 500 had already appreciated by a significant amount. The market needed some time to digest the gains. Performance has since picked up and is above the median.
Composite Confirmation Model
As the table below shows, 90% of the composite confirmation model components registered a signal post the pandemic low. The one holdout is a model that uses a 252-day calculation. Once closing prices start to roll down in late February, I suspect the model will trigger.
Composite Confirmation Model Signal Performance
As was the case with the composite thrust model, the confirmation model's signal occurred after the S&P 500 had appreciated a significant amount. One should not be surprised to see the weak performance in the 2-8 week timeframe as the market digested the gains. I suspect the results look better in the six and twelve-month timeframe when they closeout later this year.
Composite Risk Warning Model
As the table below shows, 20% of the composite risk warning model components registered a signal before the pandemic peak. The risk warning model contains indicators that typically only trigger around significant market tops when market participation is weak.
Composite Risk Warning Model Current Count
Now that the market is firmly in a new cyclical uptrend, my attention will focus on the risk warning model. As the chart below shows, the model signal count currently resides at 10%, with one out of ten components on a signal. The sentiment component is the only indicator with a warning signal, which should not be a surprise.
Conclusion: The market, as measured by the S&P 500, remains firmly in a new cyclical uptrend. While the sentiment backdrop is overly optimistic, market participation is firm. Therefore, it is unlikely that the risk warning model will initiate a new signal in the near-term.