Quantitative research
Numerics, stocks, derivatives etc. - all quantitative, not using the traditional qualitative approaches.
Independent practice · London · October 2025 – August 2026
An independent research programme in futures, derivatives and equities. The organising question was simple to state and hard to answer: how do you detect a regime shift without overfitting to the last one?
Lines of work.
- Non-stationarity. Most published edges are stationary-regime artefacts. I was interested in signals whose derivation survives a change in the data-generating process, not signals that happened to survive a backtest.
- Structural constraints. Market impact, inventory, liquidity and latency are not frictions to be assumed away; they are boundary conditions. A strategy specified without them has not been specified.
- Signal and noise. Physics-inspired separation methods: the same problem as pulling a faint astrophysical signal out of an instrument’s systematics, with an adversary added.
- Stochastic control. Hamilton–Jacobi–Bellman (HJB) formulations for execution and inventory, Monte Carlo for path-dependent problems, and Bayesian methods for parameter uncertainty.
The execution problem, formally. Liquidating inventory \(Q_t\) at rate \(\nu_t\) against a mid-price that your own trading moves is a stochastic control problem. With permanent impact \(b\) and temporary impact \(k\),
\[dS_t = -b\,\nu_t\,dt + \sigma\,dW_t, \qquad dQ_t = -\nu_t\,dt, \qquad d X_t = \nu_t\left(S_t - k\,\nu_t\right)dt\]and the value function obeys a Hamilton–Jacobi–Bellman equation, the supremum running over the admissible trading rate:
\[\partial_t V + \frac{1}{2}\sigma^{2}\,\partial^{2}_{SS}V + \sup_{\nu}\Big\{\nu\left(S - k\nu\right)\partial_X V - \nu\,\partial_Q V - b\,\nu\,\partial_S V\Big\} = 0, \qquad V(T,\cdot) = -\alpha\,Q_T^{2}\]The terminal penalty \(-\alpha Q_T^{2}\) is the point. Impact, inventory and the closing bell are boundary conditions rather than frictions bolted on afterwards, and a strategy that omits them is solving a different problem from the one the market poses.
Background. Securities Education Certificate (Distinction, Imperial College Investment Society); Finance Accelerator, London; member of Imperial’s Algorithmic Trading and Investment societies.
A note on the dual track. I do not treat physics and quant as a hedge against each other. They are the same discipline - build a model of a process you cannot fully observe, quantify what you do not know, and act on the result - applied to data that pays differently.
Work in progress; outputs will appear here and on Publications.
Neutrino Physics @ Imperial College London, DUNE & NOvA Collab || Machine Learning || Quantum Computing || Quant Researcher || Multidisciplinary Scientist || Physics × ML × QC × Quant || 🇬🇧 London-based (2023–Present)
Imperial College London | Imperial College London
View profile