Location: London preferred, strong remote candidates considered.

About Synth

Synth is a frontier financial research company and trading firm.

Synth is powered by an open, adversarial machine learning competition where independent researchers and models generate future price distributions across financial markets. These forecasts are evaluated continuously and applied to our live trading and market-making systems.

Our goal is to build the most advanced synthetic state-price engine in finance: a system that can model not just where markets may go, but the full probability distribution of future outcomes.

We are looking for a Quant Researcher to improve the quality of Synth's competition, strengthen our internal machine learning research capabilities and collaborate with leading external research teams and institutions.

The goal is to establish Synth as the benchmark for machine learning in financial forecasting.

The role

You'll improve the research quality, scoring methodology, benchmark models, miner tooling, and evaluation systems that power the Synth competition. This is a research-heavy role focused on making Synth the best environment in the world for testing machine learning models on live financial forecasting tasks.

You'll work on the full research loop: designing better forecasting tasks, building stronger baselines, improving scoring rules, evaluating miner performance, detecting failure modes, and identifying which modelling techniques produce real predictive edge. The ideal candidate has experience in quantitative research, machine learning, probabilistic forecasting, financial modelling, or systematic trading research.

Responsibilities

• Improve the design and quality of the Synth forecasting competition.
• Research better methods for evaluating probabilistic forecasts across multiple assets, horizons, and market regimes.
• Build and maintain benchmark models that miners must outperform.
• Develop stronger scoring frameworks using proper scoring rules such as CRPS.
• Identify whether miners are improving from real predictive skill, overfitting, exploitation, regime luck, or benchmark weakness.
• Research new competition formats for volatility forecasting, path forecasting, distribution markets, options pricing, and event-based markets.
• Work with engineering to improve miner tooling, validator logic, evaluation pipelines, data quality, and leaderboard transparency.
• Create research reports explaining model performance, competition improvements, and emerging forecasting techniques.
• Help define the roadmap for new Synth assets, horizons, scoring systems, and research tracks.