Natural Algorithms Research is a systematic investment firm in Herzliya, Israel. We research market structure, build our own execution and risk infrastructure, and deploy our own strategies against it.
Crowded trades compress edge, and correlated positions create fragility that surfaces only in the worst moments.
We treat individual strategies as disposable and invest in the layer beneath them — the research, validation, and risk machinery that outlives any single edge.
We designed, built, and operate the research infrastructure in-house. Market data capture, normalisation, and storage sit at the base of it. Above that run distributed compute for large-scale research workloads and the backtesting and forward-testing engines that every hypothesis passes through.
The machine-learning pipeline spans feature engineering, signal research, and risk overlays, and it runs against the same data path that trades.
We test against historical data and then forward, and we hold every experiment to the same standard on leakage and overfitting that we hold the systems themselves to. Hypotheses that do not survive are discarded, continuously.
Execution infrastructure and order routing are ours. Orders leave through the same codebase the research runs on, which is why we do not separate the two.
Risk overlays are part of the pipeline rather than a review that happens beside it: limits are enforced in the execution path, not by an operator watching a screen.
We build the research infrastructure. Strategies are its outputs.
We generate investment hypotheses continuously, across markets and data sources.
We validate every hypothesis against historical and forward data before we trust it.
We deploy what survives, monitor it live, and retire it when it stops working.