What Actually Moves Bitcoin? Geopolitics vs. Policy vs. Macro
March 2026
What Actually Moves Bitcoin?
Geopolitics vs. Policy vs. Macro
A Bayesian posterior update analysis of three competing hypotheses — using real event data from 2022–2026.
With U.S.-Iran conflict active, Middle East escalation ongoing, and Trump's crypto policy regime reshaping the regulatory landscape simultaneously, Bitcoin is being pulled by three distinct force vectors at once. This post applies Bayesian posterior updating, event-study abnormal return analysis, and rolling correlation estimates to answer a single question: which driver has the highest posterior probability of being the dominant influence on Bitcoin price in the current macro regime? The answer may surprise you — and the data is unambiguous.
Three Competing Hypotheses
When a major event hits — missiles over Dubai, a White House executive order, a Fed rate decision — financial media scrambles to assign causation. But Bitcoin is simultaneously exposed to all three force vectors. To disentangle them, we need a probabilistic framework, not headlines.
We define three mutually non-exclusive hypotheses about what drives Bitcoin price:
Direct military conflict triggers risk aversion. Bitcoin either falls (risk-off) or rises (safe-haven / dollar hedge). The "digital gold" narrative is tested every time missiles fly.
Executive orders, SEC posture shifts, ETF approvals, and strategic reserve policy directly alter Bitcoin's regulatory discount rate and institutional accessibility.
Fed rate decisions, M2 money supply expansion, and the dollar's reserve currency credibility set the liquidity floor that Bitcoin trades within as a high-beta risk asset.
The Posterior Update Model
We apply Bayes' Theorem to update our belief in each hypothesis given the observed empirical evidence from 2022–2026. The core question: given what we've actually observed in Bitcoin price data across documented events, what is the posterior probability that each hypothesis represents the dominant driver?
Where Hᵢ ∈ {Geopolitical, Policy, Macro Liquidity} and "data" = observed abnormal BTC returns in the 24–72hr event window.
Prior Probability Assignment
Before looking at recent data, we set priors based on the academic literature on Bitcoin price drivers (pre-2024). The dominant finding prior to the Trump regulatory era was that macro liquidity and risk-on/risk-off dominated, with geopolitical events producing short-lived, mean-reverting shocks:
These priors are informed by the ARDL model findings (Buthelezi 2025, MDPI) showing U.S. monetary policy as the dominant BTC driver pre-2024, and by the factor regression result that Bitcoin's β relative to the S&P 500 is approximately 0.97, confirming its risk-asset nature.
The Event Evidence (2022–2026)
We now compute the likelihood P(data | Hᵢ) for each hypothesis by reviewing documented price events. The key question: did Bitcoin's observed behavior match the prediction each hypothesis would make?
| Date | Event | Type | Predicted Δ | Actual 24–72hr Δ | H₁ Score | H₂ Score | H₃ Score |
|---|---|---|---|---|---|---|---|
| Nov 2022 | FTX collapse | POLICY | Large drop | −30% | — | ✓ Strong | — |
| Jan 2024 | SEC approves 11 spot BTC ETFs | POLICY | Large rally | +20% (weeks) | — | ✓ Strong | — |
| Apr 2024 | Iran retaliatory strikes on Israel | GEO | Drop then recovery | −4% → recovery | ~ Partial | — | — |
| Sep 2024 | Fed cuts rates 50bps | MACRO | Rally (risk-on) | +15% (weeks) | — | — | ✓ Strong |
| Nov 2024 | Trump election victory | POLICY | Large rally (reg. signal) | +30% (1 week) | — | ✓ Strongest | — |
| Jan 2025 | Trump inauguration + EO on digital assets | POLICY | Rally | +2.9% abnormal | — | ✓ Confirmed | — |
| Feb 2025 | CPI 3.0%, Fed rate pause confirmed | MACRO | Slight negative | −2% (week) | — | — | ✓ Confirmed |
| Mar 2025 | BTC Strategic Reserve EO signed | POLICY | Rally | +1.1% abnormal | — | ✓ Confirmed | — |
| Apr 2025 | Trump "Liberation Day" tariffs | POLICY | Drop (risk-off) | −5.7% | — | ✓ Confirmed (↓) | ✓ Shared |
| Apr 2025 | BTC-S&P500 correlation spike (tariffs) | MACRO | High correlation | ρ = 0.73 | — | — | ✓ Confirmed |
| Jun 2025 | Israel-Iran escalation (June) | GEO | Crisis drop | −4%, then recovery | ~ Partial | — | — |
| Jun 2025 | BTC-S&P500 corr. during ME conflict | MACRO | High correlation | ρ = 0.90 | — | — | ✓ Strong |
| Oct 2025 | BTC ATH $126K after Fed rate cut | MACRO | Rally | +ATH $126K | — | — | ✓ Strong |
| Feb–Mar 2026 | U.S. launches "major combat" in Iran | GEO | Crash? | −4% → partial recovery | ~ Partial | — | — |
Computing P(data | Hᵢ)
We now calculate the likelihood of observing the above data pattern under each hypothesis. The critical diagnostic is hit rate: what fraction of the time did the observed BTC price move match the direction and magnitude predicted by each hypothesis?
H₁ — Geopolitical: Likelihood Score
Across the five major geopolitical shock events (Iran-Israel Apr 2024, Jun 2025, Israel-Gaza Oct 2023, Russia-Ukraine Feb 2022, US-Iran Mar 2026), Bitcoin's response was consistent but shallow: initial drops of 3–4%, followed by recovery within 48–72 hours in four of five cases. The Jun 2025 conflict produced only a 1.27% 24-hour move despite simultaneous missile strikes on Dubai, Kuwait, and Bahrain. The pattern is mean-reverting and subsiding — classic noise, not structural.
P(data | H₁) ≈ 0.22 — Geopolitical events produce a predictable but shallow, mean-reverting signal. Bitcoin behaves like a risk asset during the initial shock (falling), then decouples within 72hrs. The "digital gold" safe-haven thesis receives no empirical support across these events. Weak likelihood of being the dominant driver.
H₂ — Policy: Likelihood Score
The policy hypothesis produces the strongest and most consistent evidence. The Trump election produced +30% in one week — the largest single-trigger move in the event dataset. The SEC ETF approval produced a multi-week +20% rally. Each major regulatory win (EO, strategic reserve, SEC enforcement withdrawal) generated positive abnormal returns. The Columbia Law School event study (Dec 2025) found a +5.63% abnormal return when Trump met with crypto industry leaders — statistically significant at the 1% level. Critically, policy events produce durable price shifts, not mean-reverting spikes.
P(data | H₂) ≈ 0.68 — Policy events show the highest hit rate, largest magnitude, longest duration, and most consistent directionality in the event dataset. The regulatory risk premium is a real, quantifiable, and currently dominant force in Bitcoin pricing. Strongest likelihood estimate of the three hypotheses.
H₃ — Macro Liquidity: Likelihood Score
The macro hypothesis performs well in correlation-based analysis. The BTC-S&P500 rolling correlation hit 0.73 post-tariff announcement and 0.90 during the June 2025 Middle East conflict — confirming Bitcoin's risk-asset identity. The Fed's September 2024 rate cut preceded the rally to the $126K ATH. However, macro effects are permissive rather than catalytic: they set the risk environment that policy signals operate within. When macro and policy align (loose money + pro-crypto policy), Bitcoin surges. When they conflict (tight money + pro-crypto policy), the policy signal is muted.
P(data | H₃) ≈ 0.55 — Strong evidence in correlation-based metrics, but macro factors function as the baseline environment rather than the primary catalyst. Macro is a modulator, not a trigger. Second-strongest hypothesis but structurally subordinate to policy in the current regime.
Bayesian Posterior Update
Applying Bayes' theorem with our priors and computed likelihoods:
Normalized posteriors sum to 1.0. Z is the marginal likelihood (normalizing constant).
Why the Geopolitical Signal Fails
This is the most counterintuitive finding — and the most important for trading. During active warfare involving U.S. forces and Iran, Bitcoin has consistently produced a 3–4% initial drop followed by recovery. Not a safe-haven surge. Not a sustained crisis premium. This behavior has repeated across five separate geopolitical escalation events since 2020.
The mechanism is clear: Bitcoin is still primarily a risk asset, not digital gold. Its correlation with the S&P 500 and Nasdaq during geopolitical stress regularly exceeds 0.85, compared to gold's near-zero or negative correlation with equities during the same periods. When institutional investors need liquidity, they sell Bitcoin alongside equities. The geopolitical signal is overwhelmed by the macro signal.
"The most severe military escalation in the region in years moved Bitcoin just 1.27% in 24 hours. That is not a crisis. That is a market taking the news like it takes the weather."
The key insight from the event study literature: Bitcoin has no geopolitical memory. It absorbs the initial shock, normalizes within 72 hours, and then reverts to the dominant policy and macro narratives. In the current regime, those narratives are: (1) Is the Trump administration adding or removing regulatory risk premium? (2) Is the Fed tightening or loosening the liquidity environment?
Bayesian Regime Switching
The posterior probabilities are not static — they shift with the macro regime. Here is how the dominant driver changes conditionally:
- Policy signal amplified by loose money
- H₂ posterior rises to ~60%
- Example: Nov 2024 – Jan 2025
- BTC behavior: sustained multi-week rally
- Geopolitical noise: irrelevant
- Macro overrides policy signal
- H₃ posterior rises to ~55%
- Example: Apr 2025 tariff period
- BTC behavior: drops with equities despite good policy
- Geopolitical noise: amplifies macro fear
- Geopolitical produces initial shock only
- H₁ posterior temporarily rises to ~25%
- Example: Jun 2025, Mar 2026
- BTC behavior: −3% to −5% then V-shaped recovery
- Resolution: reverts to H₂/H₃ within 72hrs
- Macro becomes the primary driver
- H₃ posterior rises to ~65%
- Example: Sep–Oct 2025
- BTC behavior: ATH at $126K
- Note: policy tailwind was prerequisite
Rolling Pearson Correlation — Cross-Driver
As a supplementary check on the Bayesian results, we review rolling 30-day Pearson correlation estimates across the three driver types using available reported data from 2025:
* ρ ≈ 0.71 for policy events is an aggregate estimate from event-study abnormal returns methodology (Columbia Law School, Dec 2025). The negative geopolitical correlation reflects mean-reversion tendency post-initial shock.
What This Means for Positioning
The Bayesian analysis yields four actionable conclusions for Bitcoin investors operating in the current March 2026 regime:
1. Ignore geopolitical headlines as a primary signal. With a posterior probability of only 8.3% as the dominant driver, and a demonstrated mean-reversion pattern across five separate escalation events, the U.S.-Iran conflict does not fundamentally alter Bitcoin's price trajectory. Use the initial −3% to −5% shock as a potential entry, not an exit.
2. Watch U.S. policy with maximum attention. With a posterior of 44.9%, policy events are the most reliable catalysts. The specific variables to monitor: any new executive orders, SEC enforcement actions or withdrawals, congressional crypto legislation progress, and strategic reserve expansion announcements. Each of these produces abnormal returns of +1% to +5.6% in the 24–72hr window — larger and more durable than any geopolitical event in the dataset.
3. The macro regime is the multiplier. Policy signals are amplified in risk-on environments and muted in risk-off ones. The current March 2026 state — active war in the Middle East, oil price spike, potential inflation resurgence — is a risk-off macro environment. This means policy tailwinds may be temporarily suppressed until macro stabilizes. The Fed's next move is the conditional variable.
4. Sequential Bayesian updating is required. The current geopolitical shock is not a regime-changer — it is noise within an ongoing policy and macro-driven cycle. As the situation evolves, the relevant posterior update is: does the conflict produce a change in U.S. monetary policy (inflationary oil shock → rate hike expectations)? If yes, the macro posterior rises and becomes temporarily dominant. That is the signal worth tracking.
"Bitcoin has no geopolitical memory. It has a policy memory and a liquidity memory. Learn to read those, and the missile launches become background noise."
Priors are subjective estimates based on literature synthesis — a sensitivity analysis across prior ranges of ±0.10 does not change the rank ordering. Likelihood estimates are calibrated from event-study data but are not formally computed from a regression model on this dataset — they are expert elicitation priors informed by reported empirical findings. The regime-switching analysis is qualitative. A formal Hidden Markov Model or time-varying coefficient DCC-GARCH would provide more rigorous estimates. The N of geopolitical events (5) is small; wider confidence intervals apply to H₁ estimates.
Comments
Post a Comment