Deep Q-learning market makers in a multi-agent simulated stock market

Fernández, Vicente, Oscar; Rebollo, Fernando Fernández; Polo, Francisco Javier García

doi:10.1145/3490354.3494448

Cited by 3 publications

(1 citation statement)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Khairalla et al used a combination of ARIMA and other forecasting algorithms for different time series [9]. With the development of hardware, methods such as machine learning, deep learning, and reinforcement learning can be used in the prediction [10][11][12]. By constructing different technical indicator variables, Zhang et al achieved high accuracy in classifying stock prices' increasing and decreasing for the coming day [13].…”

Section: Related Workmentioning

confidence: 99%

Enhanced Bollinger Band Stock Quantitative Trading Strategy Based on Random Forest

Yan¹,

Wang

2023

Artificial Intelligence Evolution

View full text Add to dashboard Cite

Constructing applicable automated stock trading strategies has become one of the best ways that people can earn profits from their underlying assets' investments now. Automated stock trading, also called quantitative trading, contains sets of human-defined rules, which are written in codes to make decisions to go long or short on stocks on a computer. Investment banks, brokerages, private equity funds, and other financial institutions around the world are keen on investigating and developing quantitative trading strategies with sustainable profitability to yield higher returns than the normal market. This research aims to observe the trading performance and profits of financial banking stocks in the Hong Kong stock market by building a quantitative trading strategy named Enhanced Bollinger Band Strategy based on Random Forest and Bollinger Bands. In experiments, the Random Forest algorithm is applied to predict the Weighted Moving Average the next day. Meanwhile, Bollinger Bands are the trading signals used to make decisions on going long or short positions based on the historical moving average lines and standard deviation. Performances of the Enhanced Bollinger Band Strategy are evaluated by test sets of ten financial banking stocks. We also compare the performance of the Enhanced Bollinger Band Strategy and Traditional Bollinger Band Strategy and find that the Enhanced Bollinger Band Strategy can earn 10-30% profits on a variety of stocks although these stocks are losing 10-50% original amount of investment in Traditional Bollinger Band Strategy and basic investment. Therefore, a combination of Random Forest and Bollinger Bands in the quantitative trading strategy generates higher returns than simply investing in stocks.

show abstract

Section: Related Workmentioning

confidence: 99%

Enhanced Bollinger Band Stock Quantitative Trading Strategy Based on Random Forest

Yan¹,

Wang

2023

Artificial Intelligence Evolution

View full text Add to dashboard Cite

show abstract

Automated market maker inventory management with deep reinforcement learning

Fernández

Garcı́a

2023

Appl Intell

View full text Add to dashboard Cite

Stock markets are the result of the interaction of multiple participants, and market makers are one of them. Their main goal is to provide liquidity and market depth to the stock market by streaming bids and offers at both sides of the order book, at different price levels. This activity allows the rest of the participants to have more available prices to buy or sell stocks. In the last years, reinforcement learning market maker agents have been able to be profitable. But profit is not the only measure to evaluate the quality of a market maker. Inventory management arises as a risk source that must be under control. In this paper, we focus on inventory risk management designing an adaptive reward function able to control inventory depending on designer preferences. To achieve this, we introduce two control coefficients, AIIF (Alpha Inventory Impact Factor) and DITF (Dynamic Inventory Threshold Factor), which modulate dynamically the behavior of the market maker agent according to its evolving liquidity with good results. In addition, we analyze the impact of these factors in the trading operative, detailing the underlying strategies performed by these intelligent agents in terms of operative, profitability and inventory management. Last, we present a comparison with other existing reward functions to illustrate the robustness of our approach. Graphic Abstract

show abstract