Transformer models and spotting volatility regime changes
Why attention-based networks react faster than recurrent models when a market shifts from calm to stressed.
Research
Analysis from the trading desk
Regular notes from the Thruvix research team on machine-learning models, UK regulation and the infrastructure behind digital assets.
Latest research notes
The FCA cryptoasset regime: what UK retail investors should know
A plain-English guide to the authorisation, disclosure and promotion rules taking shape in the UK.
Using language models to read market-moving headlines
How natural-language processing turns financial news and social posts into early, structured alerts.
Bitcoin mid-cycle: what on-chain data is signalling
Exchange flows, long-term holder behaviour and miner activity point to a handful of plausible scenarios.
Why rules beat reflexes in a falling market
Systematic models stick to their risk limits when fear starts driving human decisions.
Sterling stablecoins and the future of UK settlement
Plans for regulated pound-backed stablecoins could change how digital assets are paid for and settled.
Reinforcement learning for smarter position sizing
How adaptive agents scale exposure up or down in line with volatility and a fixed risk budget.
Cryptoassets and HMRC: a primer on Capital Gains Tax
Disposals, allowances and record-keeping: the essentials for UK holders before Self Assessment season.
Detecting wash trading and spoofing with anomaly models
Unsupervised learning flags manipulative order patterns long before a human analyst could.
Layer-two networks and where institutional liquidity is heading
Arbitrum, Optimism and Base are drawing growing professional flows. Here is what that means for fees and depth.
Put the research into practice
Open an account and let the Thruvix engine apply these ideas to live markets.
Create account