Monolith Research is a systematic-research practice — led by its founder and built by a small, growing team — held to a standard most ideas do not survive. The short version: a live, independently-scored model, a validation framework most desks would recognise, and a funding-grade research book — with the code, the working papers, and the way to reach me at the end.
Monolith Research is an institution of few — a systematic-research practice founded and led by Bilal Malik and built by a small team, held to a bar hard enough that most ideas do not clear it. Its output is method, evidence, and a small number of records I can stand behind. This page is the honest short version, and then it points you at the proof.
Four things worth knowing before you read further.
Most track records show you the survivors and ask you to trust the selection. I built the instrument that decides whether a survivor is real, and I open-sourced it. It scores a strategy the way a sceptical desk would: a Probabilistic Sharpe ratio that asks whether the record is long enough to trust, a Deflated Sharpe ratio that subtracts the edge you get simply from trying many ideas, an estimate of the probability the backtest is overfit, costs stressed well beyond the live figure, and holdouts the search never sees.
You can run it yourself — the interactive version lives on the gate page, and the code is public. A number that survives this is worth more than a higher one that did not.
What survives is deliberately small, and I would rather show you a few results that held up than a wall of curves that did not. The research book is built from five low-correlation sleeves — distinct sources of return that do not rise and fall together — combined by equal risk contribution so no single sleeve runs the book. Out-of-sample and after costs, they combine to a Sharpe near 1.64.
These are research results, not a live trading record, and I keep that line bright. The value here is a defensible book and a process that documents what it tests and retires what does not earn its place.
Two things are genuinely live. A model scored independently on Numerai, writing an out-of-sample record I cannot retouch. And the infrastructure itself: a research operating system of thousands of files — data pipelines, a validation layer, execution and monitoring tooling — designed, built, and run in-house.
The engineering is not a sideline to the research; it is the reason the research can be trusted, because every figure on this site is reconstructible from a versioned notebook and a dated data snapshot.
Monolith Research was founded by Bilal Malik, who leads the research, and is built by a small and growing team. He read Finance at Lancaster and took a high distinction in the International Business Tournament; together, the team placed 15th of roughly 2,200 in IMC Prosperity. We work across statistics, machine learning, and the systems engineering that holds a research operation together. The fuller picture of the team is on the founder page.
Monolith Research is a research practice. It is not asset management, does not manage client capital, and nothing here is investment advice or a performance promise.
Monolith Research