Dynamic Pricing for the Crypto.tickets System

Crypto.tickets will support dynamic and differentiated pricing. It is a service based on artificial intelligence algorithms, segmentation, and historical data that can increase profits from ticket sales.

Pricing service is implemented with two main variants, both can be used simultaneously:

• Differentiating starting prices for each fixed seat or place available for such events/venues that support differentiated seating (such as theaters, a show at a venue, a sports match, etc.), taking into account historical data.

• A dynamic price which changes during the sales period with the purpose of maximizing the overall ticket sales revenue.

Pricing service uses the information indicated below:

• Historical ticket sales data for specific event types, locations, seats.

• Comprehensive segmentation based on:

a. event type;

b. location type;

c. ticket price history for similar events;

d. user type.

• Change in demand, which depends on the following:

a. price changes across ticket segments;

b. organizer marketing activity

  - related to intentions for changing ticket prices;
  - related to launching marketing campaigns (advertising);

c. competition between various events;

d. stage of sales lifecycle (the beginning or the end of sales period).

• Information about restrictions being introduced by event organizer:

a. minimum and maximum prices for specific tickets;

b. occupancy limit;

c. restrictions on revenue gained from the event;

d. possible restrictions for the frequency and range of price changes.

The pricing model utilizes deep reinforcement learning algorithms when offering ticket prices to maximize event profitability taking into account current situation and sales dynamics.

More specifically, the system uses deep reinforcement learning (Q-learning, where Q is a function of profit prediction from ticket sales, depending on the selected pricing strategy) adopted for continuous states and actions. The Q function is expressed by the following formula:

The specific structure of the neural network or the learning specifics is a trade secret.

The block diagram of the common usage of the pricing model is shown below.

To learn more about our project, please follow us:

Crypto.ticket
Facebook
Whitepaper