نشر
Bluechip
·
--
$BTC Hidden Edge: Hard to Trade but Powerful to Hold (3/3)
People say “you can’t predict Bitcoin.”
That’s half true and very bullish.
The key metric is the Hurst exponent (H), which measures market memory:
H = 0.5 → random walk
H > 0.5 → persistence (trends tend to continue)
H < 0.5 → mean reversion (moves tend to fade)
Bitcoin’s rolling 120-day Hurst has ranged from ~0.36 to ~0.91 in my tests.
That means the game keeps changing:
trend regime, chop regime, near-random regime.
I tested momentum, mean-reversion, and random strategies across 2,000+ days:
Momentum hit rate: 52–55%
Mean reversion: 45–48%
Random: ~50%
Short-term edge is thin and unstable.
Most people are trying to force consistency in a regime-shifting market.
Zoom out.
Across ~17 years, Bitcoin’s long-run power-law fit is around R² ≈ 0.96 (in-sample).
That does not mean perfect day-to-day prediction.
It means the long-horizon structure has been strong.
With full-sample Hurst around ~0.57–0.61, the picture is consistent: persistence dominates over longer windows.
What the data suggests on horizon (approximate):
• <3 months: mostly regime noise
• 3–12 months: mixed, path-dependent
• 12–18+ months: trend signal starts to dominate
• Multi-year (3+ years): strongest structural predictability
The same math that makes short-term trading hard is the math that statistically supports long-term holding.
(This is how my predictions are made)
People say “you can’t predict Bitcoin.”
That’s half true and very bullish.
The key metric is the Hurst exponent (H), which measures market memory:
H = 0.5 → random walk
H > 0.5 → persistence (trends tend to continue)
H < 0.5 → mean reversion (moves tend to fade)
Bitcoin’s rolling 120-day Hurst has ranged from ~0.36 to ~0.91 in my tests.
That means the game keeps changing:
trend regime, chop regime, near-random regime.
I tested momentum, mean-reversion, and random strategies across 2,000+ days:
Momentum hit rate: 52–55%
Mean reversion: 45–48%
Random: ~50%
Short-term edge is thin and unstable.
Most people are trying to force consistency in a regime-shifting market.
Zoom out.
Across ~17 years, Bitcoin’s long-run power-law fit is around R² ≈ 0.96 (in-sample).
That does not mean perfect day-to-day prediction.
It means the long-horizon structure has been strong.
With full-sample Hurst around ~0.57–0.61, the picture is consistent: persistence dominates over longer windows.
What the data suggests on horizon (approximate):
• <3 months: mostly regime noise
• 3–12 months: mixed, path-dependent
• 12–18+ months: trend signal starts to dominate
• Multi-year (3+ years): strongest structural predictability
The same math that makes short-term trading hard is the math that statistically supports long-term holding.
(This is how my predictions are made)
إخلاء المسؤولية: تتضمن آراء أطراف خارجية. ليست نصيحةً مالية. يُمكن أن تحتوي على مُحتوى مُمول. اطلع على الشروط والأحكام.