Leveraging the Power of AI to Navigate Drawdowns
Is the strategy to simply buy and hold stocks good for everyone? Buy and hold has been a staple of investing over the last couple of decades. Thanks to many pioneers in the market, many have accepted that outperforming the market is challenging to do consistently and that their best bet is to just buy the market as cheaply as possible. Yet financial markets witness far more tail events than a normal distribution would suggest. For an investor with a 30-year investment horizon, this may not be an issue - but for many who have a shorter time horizon, this poses a potential significant risk to their financial goals. On the other hand, increasingly volatile equity markets can make it difficult for even those with a 30-year horizon to keep on track towards their goals.
Despite the advisory community preaching the mantra of staying invested, many investors understand that, if possible, it is prudent to reduce exposure to risk assets when there is risk of large drawdowns. However, this is easier said than done. While numerous rule-based systems in the market purport to mitigate downside risk, we feel they may not be nimble enough and contain too many assumptions from past markets to provide timely and accurate insight. To address this problem, Qraft has set out to harness the power of deep learning artificial intelligence and bring an ETF to market that seeks to bring a new approach to this problem.
The problem with drawdowns
Avoiding large drawdowns1 in the market can have a significant impact on an investor’s portfolio. Many investors don’t realize that the return required to recover from a drawdown is not equal to the loss incurred- as seen in the chart below, the amount required to recover from a drawdown grows exponentially as losses increase.
While negative annual returns in the market are rare, intra-year drawdowns are quite common. From 2002 to 2021, the S&P 500 index saw >10% intra-year drawdowns 10 out of 20 years, meaning 50% of the time(1). Many investors understand this and look to move to cash or other safe assets during these times. Most of the time they do it when it is too late, and conversely get back into the market when it is too late. That is why Qraft has built an AI model seeking to solve this issue and released it to investors in our ETF, Qraft AI-Enhanced Large Cap Dynamic Beta and Income (AIDB).
Why AI
Hard to Keep up
If 2022 and 2023 have taught us anything, it is that understanding the economic and equity market environment is very challenging. In today’s market, there is no end to the metrics, indices, indicators, benchmarks, and LinkedIn opinion columns to grasp what is going on. Most investors do not have the time nor the resources to learn from and analyze all the market data to make informed decisions. However, we now have a tool, artificial intelligence (AI) (2), that can do just that.
By utilizing machine learning, Qraft’s AI engine is capable of learning and adapting from massive data sets without being explicitly programmed to do so. Qraft uses algorithms and statistical models to analyze and draw inferences from data, but more importantly this advanced computing capability looks at data and the relationships between data from 360 degrees and on multiple planes. The ability to parse a signal (good) from the noise (bad) can allow Qraft’s AI algorithms to navigate today’s dynamic markets, potentially giving one a better handle on avoiding drawdowns.
Forward Looking Risk Mitigation
We believe that one area where AI can be effective is market risk mitigation. While many models and indicators used are based solely on historical data, Qraft’s AI model is forward looking. Qraft’s machine-learning(3) techniques absorb, analyze, and process a wide range of macro and market data in real time, with the AI predicting the imminent risk environment in a manner that is instant, automatic, and actionable. By leveraging the analytical power of AI, the model seeks to nowcast market risk in the coming week, and then adjust the ETF equity and cash exposure accordingly. The end goal is to provide investors with early defensive positions to protect from drawdowns, as well as quickly get back into the market when the model feels the risk is lower.
While the ETF launch is new in the U.S. Qraft has been running its risk signal with Hana Life Insurance in Korea since 2019. Below you can see the historical cash weighting for the model through different market periods.
*The Hana Life Product has a constraint of a maximum 50% cash level due to Korean regulations.
Not only was the model dynamic in its allocation to cash but can be seen raising its cash level before market drawdowns before the 2020 and 2022 drawdowns.
How the AI and ETF Works
Qraft’s Risk model uses over 70 macroeconomic and market data points, most of which consist of indices and ETF data of momentum, volatility, and correlation among asset classes. Taking in these statistical metrics, the model outputs a score for the level of cash/cash equivalents to be held in the fund. For example, if the model gives a score of 80%, it will hold 20% in equities and 80% in cash/cash equivalents. By providing this in an ETF, investors may only need to invest in the fund, and need not worry about implementing the cash level themselves. As such, the fund seeks to reduce drawdowns, lower volatility, and provide better risk-adjusted returns than that of broad-based large-cap indices.
Who is this for?
This product fits as part of the core portion of a portfolio. It is ideal for clients with asymmetric risk aversion or those with a shorter investment horizon needing higher returns without increasing their drawdown risk.
Conclusion
The topic of Artificial Intelligence is sweeping the world by storm. Many wonder how the technology will be applied and its long-term implications on the financial industry and society as a whole. By applying our proprietary technology to help investors reduce drawdowns, we hope to leverage the computing power of AI to help investors reach their goals, so they can focus on the things that matter most.
Drawdown: A term used in finance to describe the decline in value of an investment from its peak to lowest point before it recovers.
Artificial Intelligence (AI): Computer systems that can perform tasks that would typically require human intelligence, such as understanding natural language, recognizing patterns, and learning from experience.
Machine Learning (ML): A subset of AI that involves the use of algorithms and statistical models to enable computer systems to improve their performance on a specific task by learning from data, without being explicitly programmed.