Attention networks and the forecasting of digital-asset volatility
Why sequence models outperform classical GARCH on crypto volatility, and where they still fail badly.
What our data scientists and market analysts are reading, testing and arguing about. Published when there is something worth saying, not on a schedule.
Why sequence models outperform classical GARCH on crypto volatility, and where they still fail badly.
How the CSA staff notices and provincial rules shape what a Canadian investor can actually access.
Detecting the slow decay of predictive signal, and why calendar-based refitting is not enough.
Supply mechanics, holder behaviour and what the derivatives curve is pricing in.
A study of decision quality under drawdown, and where a rules engine beats an experienced trader.
Throughput, finality and custody: the three questions institutions ask before they commit.
Treating allocation as a control problem, and the reward functions that stop it going wrong.
What a CAD-referenced stablecoin would need to be useful, and what stands in the way.
Clustering order-book anomalies to flag wash trading and spoofing without a labelled dataset.
A framework for scoring smart-contract, oracle and governance exposure before allocating to it.
The models behind these notes are the ones running in the platform.
Open an AccountRisk warning: trading cryptocurrencies and contracts for difference carries a high risk of loss, and you may lose your entire deposit. These products are not deposits and are not protected by the Canada Deposit Insurance Corporation or the Canadian Investor Protection Fund. Nyprel is not registered as a dealer or adviser with the Canadian Securities Administrators or with CIRO, and does not provide individual investment advice. Consider your own circumstances and take independent professional advice before trading. Past performance does not indicate future results. Contact: [email protected]