Risk-averse reinforcement learning for algorithmic trading
Authors:
- 1 March 2014
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 23808454,p. 391-398
Abstract
We propose a general framework of risk-averse reinforcement learning for algorithmic trading. Our approach is tested in an experiment based on 1.5 years of millisecond time-scale limit order data from NASDAQ, which contain the data around the 2010 flash crash. The results show that our algorithm outperforms the risk-neutral reinforcement learning algorithm by 1) keeping the trading cost at a substantially low level at the spot when the flash crash happened, and 2) significantly reducing the risk over the whole test period.Keywords
This publication has 13 references indexed in Scilit:
- Risk-Sensitive Reinforcement LearningNeural Computation, 2014
- Risk-Sensitive Markov Control ProcessesSIAM Journal on Control and Optimization, 2013
- The market impact of a limit orderJournal of Economic Dynamics and Control, 2012
- The Flash Crash: The Impact of High Frequency Trading on an Electronic MarketSSRN Electronic Journal, 2011
- Stochastic Root Finding and Efficient Estimation of Convex Risk MeasuresOperations Research, 2010
- Optimal execution strategies in limit order books with general shape functionsQuantitative Finance, 2009
- Robust Preferences and Robust Portfolio Choice2009
- DISTRIBUTION‐INVARIANT RISK MEASURES, INFORMATION, AND DYNAMIC CONSISTENCYMathematical Finance, 2006
- Reinforcement learning for optimized trade execution2006
- Optimal control of execution costsJournal of Financial Markets, 1998