Artificial Intelligence (AI) has been exceeding expectations for years and is advancing in every industry. It has been proved that the best AI units are more than a thousand time smarter than the smartest human. Decisions can be made on the automatic learning that the machine has done based on all former decisions that have been made. AI is used in healthcare, finance, operations but it lacks in the investment industry. If AI is so helpful, why haven’t we seen a large adoption of AI when making investment decisions?
With high stakes in the investment industries, the risks are high and so are the potential opportunities. Until now, investment decisions never assured positive returns. Stock returns are the most accessible asset class and therefore using AI to predict stock returns should be possible, right? Unfortunately, financial time series are incredibly advanced and is has been found that the signal to noise ratio for investments are very low and therefore is it not possible for AI to create an algorithm for predictions on stocks. At this moment, that is to an extend that the benefits do not outweigh the costs of implementing AI.
Applying an off-the-shelf data set on a new algorithm often creates large generalization errors in AI and therefore makes the AI in investment decision making hard to implement. At this moment, investment companies are trying to include new information in algorithms, to find out what can influence and improve AI predictions on investment decisions. At the same time, the inconclusiveness of AI might mean that people, who have been the base of the decision-making dataset that AI learns from, have never made its decisions based on continuous factors and therefore I feel that making an investment may actually be the largest gamble there is.
An article of PWC makes it clear that there is another reason that AI is not in the industry. Investment companies try to implement AI, but fail to implement is as they have not created a AI strategy. Without this strategy for understanding how to use it, you take a lot of risk of implementing AI. When AI is implemented in the right way, this can increase returns in your investment companies and make more efficient decisions.
The investment industry has thus far not succeeded to find out what are the crucial factors in predicting stock returns. However, will the investments in AI ever give returns to the companies, or will it be useless to even try to implement the technology in the investment industry? And if it is impossible to implement, how safe are the safest investments when not even the smartest technology ever existed cannot even closely predict your returns?
Sources:
PWC (2019), Available at: https://www.pwc.com/us/en/industries/financial-services/library/artificial-intelligence-investing.html
Mike M (2019), Available at: https://medium.com/swlh/investing-with-ai-does-it-actually-work-58f2cb3c3a10
From the investment perspective, the return such as stock returns could be affected by several factors, including market factor which is systematic risk, and the idiosyncratic factor which refers to a specific risk. it is common sense that all investment has its risk, and risk cannot be eliminated. however, in order to reduce the risk maximum, the trader comes up with a solution which is known as diversification. An investor could put different finance products in one funding pool, additionally, the correlation between those products should be less one. In that way, the investor could diversify their portfolio and reduce the idiosyncratic risk. Technically, the idiosyncratic risk could be reduced to 0 by diversification. However, it is impossible to eliminate systematic risk, and it is also hard to predict the systematic risk correctly since the finance market is volatile and fluctuating. An unexpected event could easily influence the market, thus affect the value of asset and investment return. Therefore, I think the best way that ai could be used to optimize the investment return is by diversifying the portfolio and risk control. As a matter of factor, there are some tech already been used in this filed. For example, simulation. A simulation like Monte Carlo is a good way to examine the loss and risk, with the introduction of machine learning, the simulation could be more accurate, thus increase the predictability. As for the investment side, robo-advising is an excellent idea. Since the rise of more advanced technology that can analyze various portfolio options 24/7, financial institutions have adapted to offer online robo-advising services. Compared to the human, a machine could make a more rational and sound decision.
However, it is never safe to make an investment since the risk could never be avoided, the investor should acknowledge that high return comes along with high risk, and hedging is necessary to reduce risk.