BTC Macro Fair Value Model

BTC's macro fair value. Today: $63,026 vs $116,787 fair — 46.0% below, cheaper than 94% of 4,175 days.

R² — · 7 macro factors · refit daily

Actual vs Fitted

Power-law overlay

Numbered shaded bands ① ② ③… mark past “deeply discounted” stretches (Macro PR ≤ 10) — the same episodes tabulated in “What happened last time” below. Hover a band for its dates and forward return. Most predate the 4Y window, so switch to All to see them.

Where BTC sits in the macro regime — 0 cheap, 100 rich. Today: 6.0 — deeply discounted.

BTC spot
$63,026
2026-06-08
Model fair value
$116,787
BTC is 46.0% below
Macro percent rank
6.0
Deeply discounted
100 = richest, 0 = cheapest
PMI
54.0
Neutral cycle
expansion (>50) · z = +0.24

Right now: BTC at $63,026 sits 46.0% below the macro model's fair value of $116,787. PMI 54.0 (expansion, z = +0.24) — the cycle amplifier is neutral, not adding or subtracting from fair value. Macro percent rank 6.0 (Deeply discounted).

Descriptive, not predictive. The model reads where BTC sits relative to its macro factors today. It does not forecast where any of them go next.

The power-law model says fair value is $111,255. The macro model says $116,787. They agree — the macro cycle and BTC's native clock both read the same fair value.

What happened last time

BTC has sat in “Deeply discounted” territory 7 times in 4,175 days. Median window: 47 days. Median 12-month return after entering: +124%.

Window startWindow endDurationTrough PRBTC at start12-month return24-month return
2015-01-142015-03-0147 days0.8$178+142%+360%
2015-03-202015-07-18121 days0.6$262+58%+303%
2015-08-102015-09-2244 days1.8$264+124%+1,179%
2017-03-172017-04-0217 days3.2$1,100+620%+266%
2018-09-072018-10-0731 days3.9$6,467+63%+60%
2018-11-202019-05-06168 days0.0$4,452+80%+318%
2020-09-022020-10-2049 days5.7$11,414+332%+75%

“Deeply discounted” = Macro PR ≤ 10 for at least 14 consecutive days. Returns measured from the day BTC entered the band.

What we're watching

The equation

ln(BTC) = -3.740 + (1 + 0.0143·PMI_z) · { 1.641·ln(IGV) + 0.424·ln(Hash) + 2.035·ln(SPX) + 0.266·ln(IWM) + 0.830·ln(Copper:Gold) }

Multivariate regression of ln(BTC) on a small set of macro factors, with the ISM Manufacturing PMI acting as a multiplicative cycle amplifier.

Best-subset AIC over 7 candidate factors. Current selection: IGV, hash, SPX, IWM, gold, copper_gold, nasdaq_gold. R² = 0.9752 · AIC = -10209 · n = 4,175.

R² 0.9752 means the model explains 97.5% of the variance in log-BTC over roughly ten years. AIC -10,209 is the model's information-theoretic fit penalty; lower is better. Turn the PMI amplifier off and the same factors score -9,734, 475 points worse, so the cycle term earns its place. n = 4,175 daily observations from Jan 2015 to today.

Supporting data

Each factor in the model, plotted alongside BTC over the same window. Both axes are log scale (left = BTC USD, right = factor in its native units). The dotted light-orange line is the Bitcoin power-law Q50 trend, on BTC's axis, for reference. Below each panel is a raw log-ratio oscillator — ln(BTC) − ln(factor) — green above its average, red below. The PMI panel's oscillator is referenced to the power-law Q50 instead, so its zero line is "BTC at the power-law median": you can see BTC oscillate above and below it across cycles.

BTC vs S&P 500 — β = +2.035
$200$500$1,000$2,000$5,000$10,000$20,000$50,000$100,0002,0005,000Jul '15Jan '18Jul '20Jan '23Jul '25BTCSPX
Log ratio · BTC ÷ S&P 500 (raw)
avg+3.46-2.84
Risk appetite proxy. Co-moves with BTC during macro regime shifts.
BTC vs IGV (software ETF) — β = +1.641
$200$500$1,000$2,000$5,000$10,000$20,000$50,000$100,00020.0050.00100.00Jul '15Jan '18Jul '20Jan '23Jul '25BTCIGV
Log ratio · BTC ÷ IGV (software ETF) (raw)
avg+7.46+1.94
Software/tech equities. Highly correlated with risk-on liquidity regimes.
BTC vs Copper / Gold ratio — β = +0.830
$200$500$1,000$2,000$5,000$10,000$20,000$50,000$100,0000.020Jul '15Jan '18Jul '20Jan '23Jul '25BTCcopper_gold
Log ratio · BTC ÷ Copper / Gold ratio (raw)
avg+16.66+8.43
Cyclical metals ratio. Proxy for industrial growth vs flight-to-safety.
BTC vs Hash Rate — β = +0.424
$200$500$1,000$2,000$5,000$10,000$20,000$50,000$100,000500,0001M2M5M10M20M50M100M200M500M1.0BJul '15Jan '18Jul '20Jan '23Jul '25BTChash
Log ratio · BTC ÷ Hash Rate (raw)
avg-6.20-10.25
Network security. Strongest single correlation with BTC; structural rather than cyclical.
BTC vs Russell 2000 (IWM) — β = +0.266
$200$500$1,000$2,000$5,000$10,000$20,000$50,000$100,000100.00200.00Jul '15Jan '18Jul '20Jan '23Jul '25BTCIWM
Log ratio · BTC ÷ Russell 2000 (IWM) (raw)
avg+6.85-0.04
Small-cap risk premium. Sensitive to liquidity and credit conditions.
BTC vs PMI amplifier — θ = +0.0143 (multiplicative)
$200$500$1,000$2,000$5,000$10,000$20,000$50,000$100,00045.0050.0055.0060.00Jul '15Jan '18Jul '20Jan '23Jul '25BTCPMI
Log ratio · BTC ÷ power-law Q50 — 0 = at the power-law median
Q50+2.16-1.24
ISM Manufacturing PMI. Below 50 = contraction. Used as a multiplicative amplifier — when PMI is above its mean, all factor contributions get scaled up; below mean, dampened.

Coefficients

ParameterValueInterpretation
α (intercept)-3.740Baseline log-price level when all factors are at unit value and PMI is at its mean.
θ (PMI amplifier)0.0143Each 1σ rise in PMI scales the combined factor contribution by ≈ 1.43%.
βIGV · IGV (software ETF)+1.641Software/tech equities. Highly correlated with risk-on liquidity regimes.
βhash · Hash Rate+0.424Network security. Strongest single correlation with BTC; structural rather than cyclical.
βSPX · S&P 500+2.035Risk appetite proxy. Co-moves with BTC during macro regime shifts.
βIWM · Russell 2000 (IWM)+0.266Small-cap risk premium. Sensitive to liquidity and credit conditions.
βcopper_gold · Copper / Gold ratio+0.830Cyclical metals ratio. Proxy for industrial growth vs flight-to-safety.

Honest caveats

Per-factor R² is univariate. "Hash rate alone explains 0.93" doesn't mean it adds 0.93 of explanatory power on top of the other six. Adding all seven factors only takes the joint R² from ~0.93 (best single factor) to 0.975 — the marginal value of each additional factor is small because they're all heavily collinear with the same time trend.
Coefficients are partial elasticities. "1% rise in X adds Y% to fair value" assumes the other six variables are held constant. In reality M2/SPX/IGV/hash all move together, so "all else equal" is fictional. The coefficients describe a model fit, not isolated cause-and-effect.
M2 was tested and dropped. Despite the article framing, M2 doesn't appear in the live model — too collinear with SPX/IGV/hash to add explanatory power. The IGV software ETF acts as the "growth + liquidity" proxy in its place.
The PMI amplifier is fitted on ~2 cycles. n = 4,175 daily points sounds large, but the cyclical structure has only 2 expansions and 2 contractions in the dataset. The amplifier coefficient θ = 0.0143 carries wide confidence bands — a small effective sample even with abundant daily data.
Descriptive, not predictive. The model fits historical relationships. Out-of-sample forecasts implicitly assume the same regime continues. PMI rolling over hard, regulatory shock, or BTC adoption regime change all break the fit.

How it works

What is "macro fair value"?

It's what a statistical model says Bitcoin "should" be worth given the state of the macro economy — liquidity, interest rates, equities, software-growth appetite, and Bitcoin's own network activity. It is not a price target. When BTC trades above the line it's rich relative to those drivers; below the line it's cheap. The gap, not the level, is the point.

How is the fair value calculated?

A multivariate regression of log-BTC on a small, best-subset selection of macro factors, with the ISM Manufacturing PMI acting as a multiplicative cycle amplifier — when the manufacturing cycle is hot, the model lets the macro factors push fair value harder; when it's cold, it dampens them. The fit is re-estimated daily on roughly a decade of history. The exact equation, factor list, and coefficients are shown in "The equation" and "Coefficients" above.

What is the Macro percent rank, and the strip below the chart?

It's where today's discount-or-premium sits versus its own history, on a 0–100 scale: 0 = the cheapest BTC has ever been relative to macro fair value, 100 = the richest. The strip under the main chart plots that rank through time, banded into Deeply discounted / Discounted / Fair / Stretched / Euphoric zones. It answers "is this gap unusual?" rather than "how big is the gap?".

What do the numbered shaded bands on the chart mean?

Each band marks a past stretch where BTC was deeply discounted to macro (bottom-decile Macro percent rank for at least two weeks). The numbers tie 1-to-1 to the rows in "What happened last time" below, and hovering a band shows that episode's dates and the forward 12- and 24-month return. Most predate the default 4-year view — switch the range toggle to All to see them.

Why does it disagree with the power-law model?

By design. The power-law model fits BTC only to its own age (block height) — its native clock. This macro model fits BTC to external conditions. When they agree, the cycle and the long-run trend are telling the same story; when they diverge, one of them is missing something about today. Reading them side by side is more informative than trusting either alone.

Does it predict the price?

No. It estimates a fair-value level and a dispersion band around it from history. It says nothing about where price goes next week — only how far today sits from the macro-implied centre, and how that gap has resolved in the past. Treat it as context, not a signal.

How often does it update?

BTC spot and fair value refresh daily; the regression is re-fit on each refresh. PMI is monthly (it prints the first business day of the month), so the cycle amplifier only steps when a new PMI release lands.

Want the full derivation, factor selection, and the business-cycle reasoning behind the PMI amplifier? Read the deep-dive article: The business cycle & the Bitcoin amplifier.

How this page connects