Akademik Məlumat İdarəetmə Sistemi
Journal of Applied Mathematics, vol.2026, no.1, 2026 (ESCI, Scopus)
This study assesses whether Bitcoin’s linkage with AI equities remains robust after accounting for equity risk sentiment. To this end, the study employs the multiscale quantile-on-quantile correlation (MSQQC) and multiscale quantile-on-quantile partial correlation (MSQQPC) approaches, using daily data covering 02/01/2019–16/06/2025. The results indicate that BTC–AI comovement is strongly state- and frequency-dependent rather than stable across the joint distribution or across horizons. In the high-frequency band, dependence is weak and only intermittently significant, with localised negative regions around BTC ≈ 0.20 with AI ≈ 0.30–0.50 and BTC ≈ 0.30 with AI ≈ 0.70. In the mid-frequency band, significance concentrates in the tails, showing negative dependence under downside stress conditions such as BTC ≈ 0.10–0.30 with AI ≈ 0.10, alongside sign changes when BTC is in upper-tail states. In the low-frequency band, dependence becomes broadly positive and significant across most quantile combinations, with limited decoupling when AI is highly elevated (≈ 0.80–0.90) and BTC is also in upper quantiles (≈ 0.70–0.90). Importantly, conditioning on VIX and VVIX does not materially alter these patterns, suggesting that sentiment influences segments of short-run dependence but does not overturn the longer-run BTC–AI linkage. The study derives policy recommendations from these findings.