Trend Prophecies: March 25, 2024
Hi reader
Welcome to another edition of the Trend Prophets Academy Newsletter!
In this edition, we look at:
- Why volatility is such an important measure in portfolio construction and why we need exposure to volatility for long-term wealth creation.
- How Trend Prophets uses volatility to generate signals for buying and selling.
Volatility is a key financial metric that investment professionals use ubiquitously. Almost every investor, from beginners to seasoned professionals are aware of what it is, albeit to varying degrees. It is also a term that the media will blast at the public, often penning some variation of the phrase “market volatility has increased”. No one likes volatility. It tends to mean the market is going through periods of turbulence and there are many days potentially experiencing some steep losses. In fact, almost every market crash and correction are associated by periods of high volatility. But there are several points to consider here:
- Not all volatility is bad volatility. It works in both directions: up and down. Most people are not normally concerned with upside volatility since they are making money.
- If you want to achieve your financial goals many years down the road, you will need to embrace a certain level of volatility. Volatility is needed for asset growth.
- Volatility is one of the most important inputs in many financial models, especially in machine learning and AI.
Figure 1: Detecting the volatility regimes in SPY using machine learning. Source: EODHD, Trend Prophets
Volatility in the markets is like the weather: Just as it can cycle between calm and sunny to stormy in a single day, investment prices can fluctuate from stable to highly unpredictable in short periods of time.
Figure 1 shows the volatility states of the S&P 500. This is created by using a type of AI driven model that identifies the hidden states of a system. In this case, the state is volatility. This is one of the models that we use in the Trend Prophets system (we use multiple models for volatility identification, and we weight the results).
We can see that going back to 2000, all periods associated with higher volatility in red. We can also see that many of them tend to coincide with extreme down-market events.
We can also see that some of the periods of growth, normally the rebounds after the down periods are in a higher volatility state. This emphasizes that avoiding volatility is not possible. It is also not desirable.
Figure 2: Simulating the return distributions for two investments with the same average return but different volatilities. Source: Trend Prophets
Now imagine holding these investments for 30 years with an initial investment of $10,000 (figure 2). We can run 1,000 simulations for the second case and get a wide range of potential terminal values. We cannot know for certain how much our initial investment will be worth in the second case. The first case has the same outcome every time because it is a guaranteed rate of return (red vertical line).
The problem is that we cannot get a guaranteed 7% per year return. That simply doesn’t exist. So, we need to accept this uncertainty, and we need to accept that bad market events will happen. That’s the only way we can hit our financial goals.
Since volatility is needed for investment growth, and because volatility is associated with extreme market events, it is a key ingredient in most financial models. And this is especially true for machine learning models that try to predict future market conditions.
Since the world is currently obsessed with large language models such as ChatGPT, many of you are not familiar with how we can use volatility in AI applications in finance. Here are a few examples:
- Predictive Power: Volatility is a strong predictor of future market movements. Machine learning models that include volatility as a feature can better understand and predict price movements and trends. For example, models used in algorithmic trading can adjust their strategies based on predicted volatility, improving the chances of making profitable trades.
- Risk Assessment: Machine learning models often use volatility to assess the risk associated with different assets or portfolios. By understanding the potential range of price movements, these models can help manage risk more effectively, suggesting when to reduce exposure to certain assets or when to diversify investments further.
- Anomaly Detection: Volatility patterns can help identify anomalies or unusual market conditions. Machine learning models trained on historical data can recognize when current volatility patterns deviate significantly from past trends, potentially signaling a need for a strategic review or adjustment. Or they can possibly identify structural changes within various markets (something I am researching right now).
- Market Sentiment Analysis: Some machine learning models analyze news and social media to gauge market sentiment. Volatility is a key factor in these analyses because significant news events often lead to increased market volatility. By correlating sentiment with volatility, these models can predict how news might influence market movements.
These are just a few examples. Identifying volatility regimes is incredibly powerful and it is for this reason that it is a key factor in the Trend Prophets system.
If we can identify volatility regimes and identify periods where the possibility of sharp losses increases, we can use machine learning to help protect us. And this is what our system does. By significantly reducing the frequency of sharp losses, we increase long-term wealth. We do not want to eliminate volatility. As shown above, a certain level of volatility is needed, so we should embrace it.
What we want to do is manage it and use it to help guide our investment decisions. And while human beings are terrible at timing the market, machine learning algorithms are much better at it. I will back up this statement with the analysis in the following section.
But first, let me clarify something: I am in no way saying computers can time everything. That is just not possible. But machine learning algorithms are amazing at recognizing patterns within a staggering amount of data that the human brain can simply not deal with. And all we are trying to do, is increase the odds of not being exposed to the really bad events and stack the odds in your favor.
If you try and avoid every volatile event, you will do more harm than good.
To demonstrate how well machine learning when combined with signal processing techniques works, we will use a drawdown analysis on the Nasdaq-100 ETF (QQQ). A drawdown is defined as a period where an investment declines from a peak value, hits a bottom, and then climbs back to retake that previous peak. Analysts use this type of “underwater” chart to assess how long periods of extreme losses can last so that we can understand investment behaviour.
Figure 3 shows a drawdown graph for QQQ since 2010. I have drawn a horizontal line when the losses associated with a drawdown exceed 10% (the decline from the peak value is now a 10% loss). This is run using monthly data to make it a little cleaner and easier to understand. We can see that there are 5 periods where losses exceeded a 10% drawdown from a peak value. The drawdown ends when the line regains the zero axis (this is why it is called an underwater graph).
Figure 3: A drawdown, or “underwater” graph for QQQ showing all periods of declines or losses from a peak value. Source: EODHD, Trend Prophets.
Returning to my claim that machine learning algorithms can do a great job at identifying these events as they unfold and take corrective action, I will switch to daily data. This is what we experience in real-time as the markets unfold each day (although for the sake of your sanity, you are much better off looking at monthly data for many performance tests).
The following table identifies every time QQQ experienced a 10% loss from any given peak value. This occurred 14 times, and the table shows the beginning and end dates for each drawdown period. Remember that the drawdown ends when the previous peak is reclaimed.
Table 1: The drawdown results for QQQ.
The first column shows the Trend Prophets performance using our signals of when to get in and out of QQQ. The QQQ_index column is the buy and hold case, where you don’t follow our signals and just own the ETF the entire time. As an example, the pandemic drawdown began on 2020-02-19 (the peak value date) and ended on 2020-06-03 (when it erased the full loss).
This means that during this period, the index fell by more than 10% and recovered in about 4 months and regained all the losses. In those 4 months, your return would have been 0.95%. Had you followed Trend Prophets, you would have gained 22.4% in that same 4 months. Why? Because we would have taken you out of the investment as the system identified the danger. The buy signal would have occurred once the bottom was hit, and the market started to go up again. That is a 21.45% difference and has a staggering effect on your long-term wealth due to compounding over time.
Trend Prophets performed significantly better in 11 out of 14 of the periods. That is a close to 80% accuracy rate. Again, forget about achieving 100% accuracy rates of outperforming in every bad period. That simply doesn’t exist and is not realistic when dealing with chaotic systems. Examining all the returns shown in the table will give you an idea of how well our system uses volatility as an indicator to enhance long-term investment results. You will see similar results for all strategies we offer. And the more volatile the strategy, the better our strategies do at improving drawdown results.
The key thing to take away from this, is that understanding volatility, identifying periods of high volatility, and taking the necessary steps to protect your capital is possible with machine learning. And this is what we can do for you.
Don’t forget about our March promotion. Join today to get a free month!
Don’t wait. Subscribe today and see what we can do for you.
That’s it for this edition of the Trend Prophets newsletter! Please contact us at info@trendpophets.com for any questions.
Cordell L. Tanny, CFA, FRM, FDP
President & Founder
Join the Trend
Disclaimers: Past performance is no guarantee of future results. This newsletter should not be considered as investment advice and is intended for information purposes only. Please see our Terms and Conditions for all disclaimers.