Research

We study optimal Market Making in a realistic LOB simulated via a Hawkes process, capturing key microstructural features. Framing the problem as an impulse control task with discrete interventions, we face a complex, high-dimensional HJB-QVI. To address this, we propose a two-network PPO-based RL approach with self-imitation learning, achieving Sharpe ratios above 30 with limited training. This draft supports our SIAM FM 2025 talk, with full results to follow.
Preprint : Link

Jain, Konark and Firoozye, Nick and Kochems, Jonathan and Treleaven, Philip, An Impulse Control Approach to Market Making in a Hawkes LOB Market (July 01, 2025). Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5353913 

 A key contribution of this study is the identification of several stylized facts, which are used to differentiate between large, medium, and small tick stocks, along with clear metrics for their measurement. We provide cross-asset visualizations to illustrate how these attributes vary with relative tick size. Further, we propose a Hawkes Process model that accounts for sparsity, multi-tick level price moves, and the shape of the book in small-tick stocks.

Preprint : Link

Jain, Konark, Jean-François Muzy, Jonathan Kochems, and Emmanuel Bacry. "No Tick-Size Too Small: A General Method for Modelling Small Tick Limit Order Books." arXiv preprint arXiv:2410.08744 (2024).

We propose a novel methodology of using Compound Hawkes Process for the LOB where each event has an order size sampled from a calibrated distribution. The process is formulated in a novel way such that the spread of the process always remains positive. Further, we condition the model parameters on time of day to support empirical observations. We make use of an enhanced non-parametric method to calibrate the Hawkes kernels and allow for inhibitory cross-excitation kernels.

Published in Finance Research Letters 2024: Link

Jain, Konark, Nick Firoozye, Jonathan Kochems, and Philip Treleaven. "Limit Order Book dynamics and order size modelling using Compound Hawkes Process." Finance Research Letters 69 (2024): 106157. 

 In this review we examine the various kinds of LOB simulation models present in the current state of the art. We provide a classification of the models on the basis of their methodology and provide an aggregate view of the popular stylized facts used in the literature to test the models. We additionally provide a focused study of price impact's presence in the models since it is one of the more crucial phenomena to model in algorithmic trading. Finally, we conduct a comparative analysis of various qualities of fits of these models and how they perform when tested against empirical data.

Preprint: Link

Jain, Konark, Nick Firoozye, Jonathan Kochems, and Philip Treleaven. "Limit Order Book Simulations: A Review." arXiv preprint arXiv:2402.17359 (2024). 

We address the inefficiency of parallel tempering MCMC for large-scale problems by combining parallel computing features with surrogate assisted likelihood estimation that describes the plausibility of a model parameter value, given specific observed data. Hence, we present surrogate-assisted parallel tempering for Bayesian neural learning for simple to computationally expensive models. Our results demonstrate that the methodology significantly lowers the computational cost while maintaining quality in decision making with Bayesian neural networks. 

Published in EAAI Journal 2020: Link

Chandra, Rohitash, Konark Jain, Arpit Kapoor, and Ashray Aman. "Surrogate-assisted parallel tempering for Bayesian neural learning." Engineering Applications of Artificial Intelligence 94 (2020): 103700. 

First, parallel tempering MCMC sampling method is used to explore multiple modes of the posterior distribution and implemented in multi-core computing architecture. Second, we make within-chain sampling scheme more efficient by using Langevin gradient information for creating Metropolis–Hastings proposal distributions. We demonstrate the techniques using time series prediction and pattern classification applications.

Published in Neurocomputing Journal 2019: Link

Chandra, Rohitash, Konark Jain, Ratneel V. Deo, and Sally Cripps. "Langevin-gradient parallel tempering for Bayesian neural learning." Neurocomputing 359 (2019): 315-326. 

In this study, we propose a handcrafted model based on three-dimensional sequential skeleton data. The human body skeleton movement over a frame is computed through joint positions in a frame. The joints of these skeletal frames are projected into two-dimensional space, forming a “movement polygon.” These polygons are further transformed into a one-dimensional space by computing amplitudes at different angles from the centroid of polygons. The feature vector is formed by the sampling of these amplitudes at different angles. The performance of the algorithm are compared with similar state-of-the-art and show superior performance.

Published in ETRI 2022: Link

Vishwakarma, Dinesh Kumar, and Konark Jain. "Three‐dimensional human activity recognition by forming a movement polygon using posture skeletal data from depth sensor." ETRI Journal 44, no. 2 (2022): 286-299.