Building a Meta Strategy – Technical vs Fundamental Indicators

As ingredients for the meta strategy (aimed at scaling exposure to different assets and strategies according to market conditions), I will combine reliable fundamental and technical market timing tools to build a model that is more robust and nuanced than using just one method on its own. More on the foundations and objectives of the strategy in the previous articles of this series.

The Pros and Cons of technical and fundamental models

Fundamentals paint an accurate big picture

While fundamental data can give us a very accurate overview of what is actually happening in the economic world around us, it is a rather blunt tool when it comes to timing the market. 

Very frequently stock prices lead or lag fundamental signals significantly, resulting in poorly timed execution points. This is especially noticeable in strong corrections (-15% or larger), where sensitive indicators are often late to show economic deterioration, signaling to exit near the market bottom. That is why using less sensitive leading economic indicators in a strategy usually yields better backtested results – they simply sit through most larger corrections and exit only during major recessionary bear markets.

For example, when comparing strategies using composite leading indicators like the monthly LEI with a more sensitive weekly composite – WLEI – results in a reduction in annual return from S&P 500 buy & hold by 0,24% for the above mentioned LEI strategy and a 1,75% lower return for WLEI (the strategy is out of the market when WLEI is negative) – both strategies reduce maximum drawdown, but frequently go through high mid-range drawdowns around -25%.

This chart compares the warning signals, triggered when the WLEI index (blue) turns negative (red arrows), to the S&P 500 (yellow, recessions shaded in grey) and the entry (green vertical lines) and exit (red vertical lines) of a 60/275-day moving average crossover strategy.

The latest market drop late 2018, which would have certainly been worthwhile to avoid, is a good illustration: leading fundamental data pointed to a slowdown in economic growth, but did not indicate a negative GDP in the near future. Following the economic argument in the last post, we should conclude, that any market decline and technical sell signals should prove to be temporary and not lead to a strong bear market (say beyond a 25% to 40% decline) in the current economic environment – examples for a similar development would be the market drops of 1998 or 2016.

In addition of suffering frequent deeper drawdowns than comparable technical strategies, fundamental indicators are often late to catch on when equities begin a new sustained rise. The first leg of a bull market very often turns out to be the continuation of a bear market rally that begins under very negative economic conditions that slowly start to improve.
Their great advantage is that a high percentage of fundamental warning signals are correct, when compared to technical strategies. They signal fewer false positives, that tend to cause return-reducing whipsaws in practice.

 

Technical indicators can serve as a safety measure

Technical indications can help to navigate an uncertain future as they give timely, price based warning signals. To be sure, these signals are very often wrong due to the strong upward bias of the equity markets over the long term – statistically the probability of the S&P 500 to be up year over year at any time, the base rate, is about 73%. 

The best framework is to view these technical indications as safety measures, designed not to be right more often than not, but to avoid the disastrous, low probability events, that occur rarely – with all the more devastating effects.

Historically (1950 to present) a long-term trend strategy using a moving average crossover (I tested the 60/275-day SMA cross as an arbitrary example) to time entry and exit in the S&P 500 reduced yearly returns compared to buy & hold by 0,78% with a much lower maximum drawdown of 33,24% (vs 56,78% for B&H). But only 38% of the exit signals resulted in a better re-entry price – meaning it would have actually been better to stay in the position the majority of the time – and still the strategy has much higher risk-adjusted returns than buy & hold. 

On the other hand 85% of the entry signals turned into winning trades – the golden cross is a very reliable entry indicator, which seldom gives a wrong signal before a bear market is actually over.

The key difference to the fundamental strategies above is that even the reduced maximum drawdown of 33,24% is an extremely rare outlier event (happening only once in 70 years during Black Monday crash of 1987). The vast majority of drawdowns is stopped at a level of -10% to -15% by the technical timing model (see yellow rectangle in the drawdown chart of the strategy below) – this is the most useful insight when building investment strategies that include a long-term trend model

The exact drawdown depth is determined by the speed of the decline and the sensitivity of the indicator (e.g. the length of the moving average), but the crucial point is, that an exit protecting from ruin will be triggered eventually – there is no way the strategy can fail by missing a crucial exit as a fundamental model potentially could. 

The main problem are frequent false warnings, that often lead to frustrating, return-reducing whipsaws in and out of equities. Behaviorally these frequent pointless moves can quickly lead to an abandonment of such a strategy, just when its protection is needed most.

Combining technical with fundamental indicators

One possibility to improve the low percentage of correct technical signals and the weak timing of fundamental indicators, would be to use fundamental data as a filter for the technical signal: we only use the technical trend strategy to go to cash, if our fundamental criteria signal a weakening economy or possible recession. 

Unfortunately this means that since late 2018 we would still be sitting things out – an insecure situation, that I would certainly prefer to avoid. And historically, as an example, the 1937 and 1945 recessions were missed by pretty much every recession indicator on the books – if what we are seeing right now should turn out to be similar, we will be forced to sit through most or all of the drawdown following such a model. 

An essential safety measure, that kicks in, if the fundamental model turns out to be wrong, is missing.

Combining indicators in such a way will also quite often lead to problematic binary decisions. The results can often frustratingly whipsaw us between being entirely in or out of the market. A better solution would be to use a good fundamental model to assign a higher probability of a technical signal being right and adjust our investments more gradually.

For example, if either our technical or our fundamental model indicates an exit, we could decide to reduce exposure by 50%. Only if both turn negative would we move to 100% cash or other safe assets.

At the moment, we would thus be exposed to the risk of sitting through the majority of a bear market with only 50% of assets invested. Conversely, if equity markets quickly return to bull mode, we would only get whipsawed with 50% of our portfolio – both alternatives would be much easier to stomach than the possibility of being all-out wrong.

As this is a fairly rough method, I will next highlight an idea of how to implement a more gradual shift in portfolio weights when we begin to see a possible change in market conditions. 
Then I will look at the exact indicators, that I want to use in practice and how to objectively judge their significance.

One comment

Submit a comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s