Systems and methods for providing customized financial advice

Abstract

A system includes one or more memory devices storing instructions, and one or more processors configured to execute the instructions to perform steps of a method for providing customized financial advice. The system may receive transaction data for a transaction associated with a customer and satisfaction data associated with the transaction. Based on the received transaction data and satisfaction data, the system may update a financial state of the customer and a financial policy for determining one or more actions to take in order to maximize a cumulative reward associated with the customer. The system may determine and output a recommended action based on the updated financial policy and customer financial state.

We claim: 
     
         1 . A system for providing customized financial advice, the system comprising:
 one or more processors; and   a memory in communication with the one or more processors and storing instructions that, when executed by the one or more processors, are configured to cause the system to:
 receive data representative of an environment comprising a plurality of financial states, the plurality of financial states comprising a first financial state and a second financial state, the first financial state corresponding with a customer financial state associated with a customer; 
 receive financial policy data representing a financial policy for selecting one or more actions from a plurality of actions to take to increase a cumulative reward associated with a customer, each of the plurality of actions being associated with one or more financial decisions; 
 receive transaction data for a transaction associated with the customer; 
 receive customer data indicative of a satisfaction communication from the customer, the satisfaction communication being indicative of a satisfaction of the customer regarding the transaction; 
 determine satisfaction data based at least in part on the customer data; 
 update, based on the transaction data, the customer financial state to correspond with the second financial state; 
 update the financial policy based on the transaction data; 
 update a relative customer satisfaction policy based on the satisfaction data; and 
 responsive to receiving, from a customer device associated with the customer, a query for a recommended action:
 select the recommended action out of the plurality of actions based on the updated financial policy, the updated relative customer satisfaction policy, and the updated customer financial state; and 
 output, to the customer device, recommendation data for display by the customer device, the recommendation data being indicative of the recommended action. 
 
   
     
     
         2 . The system of  claim 1 , wherein the satisfaction communication comprises a survey or review completed by the customer. 
     
     
         3 . The system of  claim 1 , wherein the satisfaction communication comprises a customer input received from the customer device. 
     
     
         4 . The system of  claim 3 , wherein the satisfaction communication is received via a web-enabled chat application. 
     
     
         5 . The system of  claim 1 , wherein the satisfaction data is associated with a recipient of a good or service purchased by the customer. 
     
     
         6 . The system of  claim 1 , wherein updating the relative customer satisfaction policy comprises:
 determining a reward based on the satisfaction data; and   determining whether the reward increases or decreases the cumulative reward associated with the customer, the cumulative reward representing one or more of a savings goals, a passive income goal, a debt reduction goal, or a happiness goal.   
     
     
         7 . The system of  claim 6 , wherein determining the reward comprises applying a set of rules associated with the financial policy to the second financial state and the satisfaction data. 
     
     
         8 . The system of  claim 1 , wherein:
 the transaction is a first transaction,   the transaction data is first transaction data,   the satisfaction data is first satisfaction data, and   the instructions, when executed by the one or more processors, are configured to further cause the system to:
 receive second transaction data for a second transaction associated with the customer, the second transaction being associated with a purchase, investment, or other transaction the same as or similar to the purchase, investment, or other transaction associated with the first transaction data; 
 receive second satisfaction data for the second transaction, the second satisfaction data being derived from a detection of a facial expression or body language from an image associated with the second transaction; 
 update, based on the second transaction data, the customer financial state to correspond with a third financial state of the plurality of financial states; 
 update the financial policy based on the second transaction data; and 
 update the relative customer satisfaction policy based on the second satisfaction data. 
   
     
     
         9 . The system of  claim 1 , wherein the customer data comprises data indicative of a facial expression or body language of the customer from an image associated with the transaction, and the satisfaction data is derived at least in part from the data indicative of the facial expression or body language of the customer. 
     
     
         10 . The system of  claim 1 , wherein receiving the customer data indicative of the satisfaction communication comprises:
 implementing an interactive voice response (IVR) system to produce an interaction with the customer, the interaction including receiving the customer data, the customer data comprising audio data associated with the customer.   
     
     
         11 . The system of  claim 1 , wherein updating the financial policy based on the transaction data is performed using one or more reinforcement learning techniques. 
     
     
         12 . The system of  claim 11 , wherein the one or more reinforcement learning techniques includes at least one of Q-learning, a policy gradient method, or an actor-critic method. 
     
     
         13 . A method for providing customized financial advice comprising:
 obtaining, via a network, environmental data representative of an environment comprising a plurality of financial states, the plurality of financial states comprising a first financial state and a second financial state, the first financial state corresponding with a customer financial state associated with a customer;   receiving, via the network, financial policy data representing a financial policy for selecting one or more actions from a plurality of actions to take to increase a cumulative reward associated with a customer, each of the plurality of actions being associated with one or more financial decisions;   receiving, via the network, transaction data for a transaction associated with the customer;   receiving customer data indicative of a satisfaction communication from the customer, the satisfaction communication being indicative of a satisfaction of the customer regarding the transaction;   determining satisfaction data based on the customer data;   updating, based on the transaction data, the customer financial state to correspond with the second financial state;   updating the financial policy based on the transaction data;   updating a relative customer satisfaction policy based on the satisfaction data; and   responsive to receiving, via the network and from a customer device associated with the customer, a query for a recommended action:
 selecting the recommended action out of the plurality of actions based on the updated financial policy, the updated relative customer satisfaction policy, and the updated customer financial state; and 
 outputting, to the customer device and via the network, recommendation data for display by the customer device, the recommendation data being indicative of the recommended action. 
   
     
     
         14 . The method of  claim 13 , wherein the satisfaction communication comprises a survey or review completed by the customer. 
     
     
         15 . The method of  claim 13 , wherein the satisfaction communication comprises a customer input received from the customer device. 
     
     
         16 . The method of  claim 13 , wherein the satisfaction data is associated with a recipient of a good or service purchased by the customer. 
     
     
         17 . The method of  claim 13 , wherein updating the relative customer satisfaction policy comprises:
 determining a reward based on the satisfaction data; and   determining whether the reward increases or decreases the cumulative reward associated with the customer.   
     
     
         18 . The method of  claim 13 , wherein the transaction is a first transaction, the transaction data is first transaction data, and the satisfaction data is first satisfaction data, the method further comprising:
 receiving second transaction data for a second transaction associated with the customer, the second transaction being associated with a purchase, investment, or other transaction the same as or similar to the purchase, investment, or other transaction associated with the first transaction data;   receiving, via an electronic interface, second satisfaction data for the second transaction, the electronic interface being between the customer and a system for providing customized financial advice and the second satisfaction data being derived from a detection of a facial expression or body language from an image associated with the second transaction;   updating, based on the second transaction data, the customer financial state to correspond with a third financial state of the plurality of financial states;   updating the financial policy based on the second transaction data; and   updating the relative customer satisfaction policy based on the second satisfaction data.   
     
     
         19 . The method of  claim 13 , wherein customer data comprises data indicative of a facial expression or body language of the customer from an image associated with the transaction, and the satisfaction data is derived at least in part from the data indicative of the facial expression or body language of the customer. 
     
     
         20 . The method of  claim 13 , wherein updating the financial policy based on the transaction data is performed using one or more reinforcement learning techniques including at least one of Q-learning, a policy gradient method, or an actor-critic method.