A mechanism-first look at Technical Indicator Networks, which turn classical trading indicators into trainable neural architectures without throwing interpretability into the usual black-box bonfire.
A mechanism-first reading of how Wasserstein QGANs may generate synthetic financial time series with distributional realism and partial temporal structure.
A mechanism-first reading of Lévy-driven option pricing when the market has no clean risk-free asset to lean on.
A comparison of analytical volatility calibration and signature-based rough-path models, showing where speed, structure, flexibility, and model risk actually trade off.
Bitcoin’s complexity is not just higher volatility; it is scale-dependent structure that changes how risk, forecasting, and model validation should be read.
MountainLion shows how multi-agent LLM systems can compress crypto research workflows, but its strongest value is interpretability and decision support rather than proven autonomous trading alpha.
A mechanism-first reading of finance as extended reciprocity, with implications for AI-agent markets, decentralized trust, and simulation design.
A mechanism-first look at how regime-switching-induced price shocks change mean-variance portfolio design from parameter tuning into shock-aware control.
A network-first reading of how MST filtering, FEVD influence, VaR weighting, and neural forecasts produced a concentrated equity strategy that beat the S&P 500 in one gross backtest.
A mechanism-first reading of why trade-volume fluctuations can make familiar Markowitz variance either understate or overstate portfolio risk.