Essential Market Microstructure Papers for Relaxation at the Beach
In the dynamic world of finance, artificial intelligence (AI) is making significant strides in stock market operations. One of the latest developments comes from the research of Guillaume Maitrier and Jean-Philippe Bouchaud, who argue that trading itself, rather than new information reaching the market, is the primary cause of volatility. They have found that a crucial parameter, the long-range correlation between metaorders, plays a decisive role in ensuring that trading creates volatility.
Their approach encodes the limit order book into a feature vector, which is used to predict the next state of the market. This method outperforms other techniques, especially when the limit order book is compressed using order-flow imbalance. The researchers provide an algorithm that allows practitioners to apply their research on lit order books.
Another notable study is by Álvaro Cartea and Leandro Sanchez-Betancourt, who focus on the problem of market makers and brokers being left with inventory that is about to rapidly change in price due to informed traders. They deploy a pair of 'Siamese twin' deep learning networks to the bid and ask side of the order book. Their study provides a solution to this problem over an infinite time horizon.
Meanwhile, researchers such as Zhang et al. and Li and Wang have described using long short-term memory (LSTM) networks for trading Chinese stocks. Their approaches generally outperformed traditional machine learning methods like SVM and random forests in prediction accuracy and profit metrics. Jiahao Yang, Ran Fang, Ming Zhang, and Jun Zhou have tackled the challenge of high-frequency trading of Chinese stocks using LTSM networks.
Notably, AI has been benchmarked against Bloomberg's TCA formula using agent-based modelling, but details about this research were not provided in the given text. The AI trader, using a technique called deep reinforcement learning, can choose trading strategies in response to market conditions. The AI trader is assigned a reward function that measures how well the strategies perform.
A paper by Jedrzej Maskiewicza and Pawel Sakowski demonstrates the successful application of AI to trading. Agent-based modelling, pioneered by Doyne Farmer, outperforms traditional efficient market frameworks in market microstructure theories, even when agents possess 'zero intelligence'.
In conclusion, the application of AI in trading is proving to be a game-changer. From understanding the causes of market volatility to developing strategies that adapt to market conditions, AI is revolutionising the trading landscape. As research continues to advance, we can expect to see even more innovative applications of AI in the world of finance.
