One of the AI mechanisms based on an allocation model for wealthy clients forms an investment strategy, taking into account the preferences and financial capabilities of a particular client. The algorithm builds medium-term forecasts of the profitability of the bank's investment products related to assets such as stocks, bonds, precious metals and investment diamonds, real estate, etc. Based on the model's forecasts and client's constraints, which include risk limits and weight limits for individual products, a task is formed, the solution of which is to create an optimal portfolio. The AI assistant offers an optimal investment strategy, which involves regular rebalancing of the portfolio taking into account current forecasts for instruments and the client's current portfolio, after which the client can either follow the AI recommendations or make their own decisions.
The second algorithm is based on longer-term strategies for managing a client's portfolio – from three years. The starting point is the goals for the planned increase in capital. Based on the specified parameters and financial capabilities, as well as the current situation in various markets, the algorithm recommends making certain investment decisions. The modeling is based on the use of a scenario approach, scenarios of asset behavior in the future are generated, the model increases the probability of achieving the client's financial goals over a given horizon, for all scenarios. As a result of the model's work, for each realized scenario, an optimal action plan is developed, a portfolio for achieving the client's goals.
Another already implemented algorithm allows predicting the amount of savings in term deposits, savings accounts, metal accounts and investment products. And this is not just about forecasting income based on the product rate and the current amount in the account (this analytical algorithm has long existed), but specifically about forecasting balances taking into account new savings and investment products that the client may potentially open in the bank over the forecasting horizon.
The pool of developed models is used as part of CRM campaigns aimed at increasing portfolios of savings and investment products. The operation of the algorithms is based on a comprehensive mathematical analysis of various types of data – socio-demographic, transactional, information about credit history, client communications with the bank, as well as information about the client's current and historical product ownership.
«Artificial intelligence algorithms take the personalization of services that the bank provides to various clients to a fundamentally new level – when it comes to deposits, salary accounts, credit offers, and other bank services. The more effective artificial intelligence algorithms work, the better service our clients will receive,» said Maxim Konovalikhin, Senior Vice President, Head of Data Analysis and Modeling Department of VTB Bank.