Artificial Intelligence in Asset Managementminute read
Interview in the Tagesanzeiger
24 January 2022
Artificial Intelligence in Asset Management
In the Interview: Dr Miró Mitev
A personal investment strategy with the least risk for each customer: This is possible through artificial intelligence in wealth management. Artificial intelligence has now also arrived in wealth management and offers particular benefits for people looking for security when investing. In an interview, Dr Miró Mitev, Founder & CEO of Smart Wealth AG, explains what artificial intelligence means in this area, what benefits arise and what is already possible at Smart Wealth AG.
Artificial intelligence has arrived in many areas of our society today, including wealth management. What does artificial intelligence mean in this field?
At Smart Wealth AG, artificial intelligence is embedded in many different levels, such as the creation of forecasts for the future development of stocks, bonds, and commodities, as well as the optimization of customer portfolios. In general, we understand artificial intelligence as mathematical models that operate independently based on predefined self-learning algorithms. Artificial neural networks or genetic algorithms are included in this. One trains such a mathematical model on the basis of financial data in so-called supervised learning in order to gain insights. These insights are then applied in the further process so that the model can continuously learn new things and become better and better. Neural networks mimic the human brain through mathematical functions. The neurons of the model are linked with non-linear mathematical functions that process the data sets.
The goal is to process the information in the input data until a model output is generated. This model output is compared with the provided target output of the training data until the deviation is minimized by supervised learning. This is how one trains an algorithm.
However, one must ensure that there is no rote learning (overfitting), but that the algorithm also deals with unknown data. So also corresponding prediction models work. Through the described supervised learning, predictions for stock prices can be created. In order to minimize the prediction error as much as possible, the AI model must be continuously optimized. Therefore, it is important to provide meaningful inputs into the prediction models on a regular basis.
Prediction models are one of the key achievements of AI in wealth management. Describe in a few sentences how the prediction model works at Smart Wealth.
Our forecasting models are designed to handle various variables such as macroeconomic, fundamental, and technical inputs with different frequencies. The goal is to use the latest information available on the market and evaluate it accordingly, providing current forecasts. For example, we speak of early indicators from macroeconomic variables. Such macroeconomic variables are, for example, the currently widely discussed inflation or unemployment rate. Such data is usually available monthly.
Decision-making for our forecast model is also based on fundamental data. These are data that concern a specific sector or company. The fundamental data also describes the relationships between different asset classes. This includes the impact of exchange rates or interest rates on the stock and bond market. To these fundamental data, technical indicators are added such as trends and volatility. These are fed into the forecast model on a daily basis. All these inputs mentioned influence the forecast models and make detailed forecasts possible through AI.
Smart Wealth promotes knowledge rather than speculation. What advantages does AI bring to the consumer in wealth management?
Every purchase and sale in our AI-controlled investment process is based on data and facts. This makes these processes objective, systematic and 100% model-based and not dependent on human decisions. This has the advantage that emotions such as fear or impatience do not play a role. Such emotions are harmful to long-term investment success. This allows us to ensure that our customers receive a portfolio with the highest possible return and lowest risk. This portfolio is scientifically and fact-based and not based on human emotions, which significantly reduces the risk.
Through AI, customized investment strategies with the appropriate investment vehicle for customers seem possible at Smart Wealth. How can I imagine that as a customer?
For a tailor-made investment strategy, we need to know our customers' personal preferences. The customer can determine which industries should be included. For example, themes such as sustainability can be taken into account and industries that do not act sustainably can be excluded. Through AI and genetic portfolio optimization, we create the optimal portfolio with the lowest risk for each customer. In genetic portfolio optimization, the goal is to find the optimal portfolio with the highest return for a given risk. Therefore, a personal, optimal, and risk-efficient portfolio can be developed based on given restrictions through AI.
Private customers can choose their suitable investment strategy for achieving their wealth goals through our AI-based calculator. Afterwards, they can quickly and easily open their securities account through our fully automated onboarding process and have the selected strategy implemented by our AI-investment process.
At Smart Wealth, we can implement a wealthy private customer's personal investment strategy in a personal, actively managed certificate (Actively Managed Certificat AMC) with a Swiss ISIN in about two weeks. Unlike a fund, there is no need to wait a long time for approval. We specialize in Actively Managed Certificates AMCs, which are becoming increasingly popular, even outside of Switzerland. AMCs have many advantages as they are very flexible, safe, cost-efficient and can be implemented quickly. The AMCs are issued by so-called fully segregated special purpose vehicles (SPVs). This practically eliminates the issuer default risk. The AMCs can be seen as a real alternative to funds.
Wealth management is more prevalent than ever, yet many people are afraid to invest and prefer to keep their money in savings accounts. How can AI reduce the fears of potential customers?
Generally speaking, one must always be invested in order to achieve long-term success. It only makes sense to reduce investments during times of recession or major market downturns. Long-term success can only be achieved by planning sustainably and systematically implementing the investment strategy. In recent years, we have found that investing with the help of AI and based on tried and tested algorithms is very successful in the long term. Human components such as gut feeling are more difficult in long-term investments. Therefore, investing with a system is more promising in the long term.
AI detects many developments before humans do. If the forecasts indicate that expectations for the future are not good, our AI reduces positions in riskier assets, invests the capital in cash or fixed-income securities and reinvests when the forecasts for assets are positive.
The year 2020 is a good example of a recession. Our AI or our forecasts have already detected the negative changes by the end of February. That means we had already reduced our position in stocks by the end of February, and as a result, we lost less. However, the AI also understood that the global economy, with the support of central banks and governments, would continue to roll on despite the pandemic, and it is not comparable to the financial crisis of 2008. Therefore, we were able to re-enter the market in April/May and achieve a top-performance worldwide. In such difficult phases, we reduce losses by more than 50% compared to the benchmark, and when the market turns and the economy recovers, the AI quickly re-invests. The effectiveness of this approach has been clearly demonstrated in recent years.
Are there also risks that AI represents in wealth management?
There are risks in any wealth management. It has been scientifically proven that 8 out of 10 active managers achieve a performance that is below their benchmark. That is also where the success of ETFs is based. Of course, there are also risks when using AI in wealth management. An AI is a mathematical model that is data-based and not a crystal ball. This can lead to uncertainties in the data and there are also faulty data that increase the risk of forecasting errors and thus the risk of false signals. Due to our long experience, our models have a high hit rate of almost 60%. This is very good when it comes to track record predictions.
This enables us to achieve long-term above-average performance that is far above the benchmark. It is very difficult for outsiders to specify the risks and dangers of the various AI systems in wealth management with the different providers. When training an artificial neural network, there are many ways to make mistakes, such as overfitting.
The development and maintenance of models based on artificial intelligence are very complex. It requires many years of experience and preparation as well as regular adjustments throughout the entire process, in order for everything to be automated and coordinated.
What developments can we look forward to in the future?
As a team, we have been working with artificial intelligence for 20 years. Being involved in models and predictions based on AI at that time was very advanced, and we provided predictions for the first AI-based products from very reputable banks.
There will be many developments in the future. If I compare the current state of AI and prediction models to what it was 20 years ago, the progress is enormous. We are still in an early stage and very few people have even understood the progress that has happened in the last 20 years. The time is ripe for AI. There will be many new things coming because once someone uses AI, they will immediately recognize the benefits and will not want to do without it.
Once I have driven a fully automatic car with a navigation system, I do not want to do without it completely. Many things are better with AI and much simpler and therefore will prevail. So we can look forward to many things in the field of AI, including in wealth management, also at Smart Wealth.
Created: January 24, 2022, 7:00 AM