Calibrating Multi-Agent Market Simulators to Historical Regimes

Summary

This paper addresses how to tune multi-agent market simulators so their generated price and volume series resemble historical market conditions. Such simulators support strategy testing, and the authors emphasize the need to represent both ordinary periods and stressed regimes, including conditions observed around the start of the COVID pandemic.

The proposed approach first trains a discriminator within a self-attention GAN to distinguish real from simulated price and volume series. That discriminator is then used in an optimization framework to adjust a simulator whose agent archetypes are known, targeting a chosen market scenario. The paper reports experiments demonstrating the method’s effectiveness. The available description gives no performance measures, benchmark comparisons, or details about the specific simulator and data, so it does not establish how broadly the calibration method generalizes across markets or regimes.

Key ideas

• The method calibrates multi-agent simulators to reproduce historical market regimes in price and volume data.
• A self-attention GAN discriminator is trained to distinguish observed series from simulated series.
• An optimization step uses the discriminator to tune parameters for a simulator with known agent types.
• The authors report effective experimental results, but the abstract provides no metrics or generalization analysis.

Tags

# 2108.00664


# Learning who is in the market from time series: market participant discovery through adversarial calibration of multi-agent simulators









In electronic trading markets often only the price or volume time series, that result from interaction of multiple market participants, are directly observable. In order to test trading strategies before deploying them to real-time trading, multi-agent market environments calibrated so that the time series that result from interaction of simulated agents resemble historical are often used. To ensure adequate testing, one must test trading strategies in a variety of market scenarios -- which includes both scenarios that represent ordinary market days as well as stressed markets (most recently observed due to the beginning of COVID pandemic). In this paper, we address the problem of multi-agent simulator parameter calibration to allow simulator capture characteristics of different market regimes. We propose a novel two-step method to train a discriminator that is able to distinguish between "real" and "fake" price and volume time series as a part of GAN with self-attention, and then utilize it within an optimization framework to tune parameters of a simulator model with known agent archetypes to represent a market scenario. We conclude with experimental results that demonstrate effectiveness of our method.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.