Studio Farzulla
@farzulla.com
Adversarial Systems Researcher | Part-time poet | Over-caffeinated & sleep deprived
Socials: https://farzulla.com
Research: https://farzulla.org
Dissensus AI: https://dissensus.ai
My homelab: https://resurrexi.io
Socials: https://farzulla.com
Research: https://farzulla.org
Dissensus AI: https://dissensus.ai
My homelab: https://resurrexi.io
Submitted to SSRN for review! Expected: papers.ssrn.com/abstract=578..., next step would be to submit to the journal of Digital Finance.
papers.ssrn.com
November 22, 2025 at 8:26 PM
Submitted to SSRN for review! Expected: papers.ssrn.com/abstract=578..., next step would be to submit to the journal of Digital Finance.
Preprint: doi.org/10.5281/zenodo.17677682
Code: github.com/studiofarzul...
Paper + code + data, CC-BY-4.0/MIT.
Code: github.com/studiofarzul...
Paper + code + data, CC-BY-4.0/MIT.
Market Reaction Asymmetry: Infrastructure Disruption Dominance Over Regulatory Uncertainty
Cryptocurrency markets respond 5.7× more strongly to infrastructure failures than regulatory announcements (2.385% vs 0.419% volatility impact, p=0.0008, Cohen's d=2.753). Using TARCH-X models across ...
doi.org
November 22, 2025 at 1:04 PM
Preprint: doi.org/10.5281/zenodo.17677682
Code: github.com/studiofarzul...
Paper + code + data, CC-BY-4.0/MIT.
Code: github.com/studiofarzul...
Paper + code + data, CC-BY-4.0/MIT.
Methods: TARCH-X quasi-MLE, Bayesian (BF>10 for 4/6 assets), k-means clustering, VAR network connectedness, regime-switching, bootstrap (10k). 50 events, 6 cryptos, 2019-2025.
November 22, 2025 at 1:04 PM
Methods: TARCH-X quasi-MLE, Bayesian (BF>10 for 4/6 assets), k-means clustering, VAR network connectedness, regime-switching, bootstrap (10k). 50 events, 6 cryptos, 2019-2025.
Sentiment evidence: TARCH-X with GDELT achieves superior AIC (5/6 assets) despite weekly aggregation and 100% negative raw bias. Markets so sentiment-driven even noisy proxies capture behavior.
November 22, 2025 at 1:04 PM
Sentiment evidence: TARCH-X with GDELT achieves superior AIC (5/6 assets) despite weekly aggregation and 100% negative raw bias. Markets so sentiment-driven even noisy proxies capture behavior.
Near-integrated volatility: All six assets hit stationarity boundaries (α+β≈0.999), half-lives >100 days vs 5-20 in equities. Crypto volatility shocks are quasi-permanent.
November 22, 2025 at 1:04 PM
Near-integrated volatility: All six assets hit stationarity boundaries (α+β≈0.999), half-lives >100 days vs 5-20 in equities. Crypto volatility shocks are quasi-permanent.
Crisis amplification: Infrastructure effects multiply 5× during stress periods (COVID-19, FTX collapse), explaining 67% of extreme volatility despite 18% sample coverage.
November 22, 2025 at 1:04 PM
Crisis amplification: Infrastructure effects multiply 5× during stress periods (COVID-19, FTX collapse), explaining 67% of extreme volatility despite 18% sample coverage.
Network topology surprise: ETH is the primary systemic risk driver (eigenvector centrality 0.89) over BTC (0.71). Spillover analysis shows ETH is the volatility transmission hub.
November 22, 2025 at 1:04 PM
Network topology surprise: ETH is the primary systemic risk driver (eigenvector centrality 0.89) over BTC (0.71). Spillover analysis shows ETH is the volatility transmission hub.
Mechanism: decentralized protocols enable jurisdictional arbitrage (regulations are geographically constrained), while infrastructure failures cascade globally with no escape route.
November 22, 2025 at 1:04 PM
Mechanism: decentralized protocols enable jurisdictional arbitrage (regulations are geographically constrained), while infrastructure failures cascade globally with no escape route.