Stock-to-Flow

The chart plots Bitcoin's daily close against the model price implied by the regression PlanB introduced in 2019. The model takes circulating supply (“stock”) divided by annual issuance (“flow”) and maps that ratio to a fitted price through a power law. Two lines: realised daily close, and the dashed S2F-implied model price computed from the live coefficients. The whole indicator is the distance between them.

Today's S2F ratio is 122.0, yielding a model price of $213,346.58 against a spot of $65,837.03 — a residual of −69.1%. The current fit's coefficients are a = -1.2821, b = 2.8210, with R² = 0.9317. That high R² is the trap this whole page exists to unpack: it is a property of the model's structure, not evidence that it forecasts.

Inputs: the daily Bitcoin USD close history and a deterministic supply schedule keyed off the four canonical halvings (block heights 210,000 / 420,000 / 630,000 / 840,000 with subsidies 25 / 12.5 / 6.25 / 3.125 BTC respectively). For every day t:

stock(t) = circulating supply at day t
flow(t) = subsidy_t × 144 × 365
S2F(t) = stock(t) / flow(t)

The model is then a log-log OLS regression of price on S2F:

ln(price_i) = a + b · ln(S2F_i)

with implied model price price_model = exp(a) · S2F^b. PlanB's original 2019 fit reported b ≈ 3.3 and a ≈ 14.6 on a market-cap regression; our nightly refit publishes daily coefficients with the latest cycle included. Full derivation lives on the methodology page.

The key structural fact is that S2F is monotone: stock only accumulates, and flow only falls — in halving-sized steps. So the model price never reacts to the market; it climbs by construction and jumps at each halving. We use 144 blocks per day (the protocol target, not the realised ~147/day), which trims roughly 2% off the absolute flow figure but preserves the doubling step at every halving — the one feature the model is built to highlight, and the one that drives every forward forecast it produces.

The single read on this chart is the residual: the gap between spot and model price. PlanB's regime thresholds were never formally codified, but the convention used in most replications is “extremely undervalued” below −50%, “fair value” within ±25%, and “extremely overvalued” above +100%. By that convention today's residual is in the extremely undervalued regime. The caution: those labels were calibrated on the first decade, when the model tracked. Since 2021 the residual has spent its time deep in the “undervalued” bands not because Bitcoin is cheap, but because the model line ran away from price. The bands below carry the historical context; read them as a record of past behaviour, not a current verdict.

Sampling the canonical cycle anchors against today's fit shows the model's behaviour across cycles. The 2013 and 2017 tops printed far above the model; the 2015 and 2018 cycle lows printed near it. The 2021 tops are not where the sign turns — both the April $63,576.68 peak and the November $67,145.37 ATH printed sharply above the model line (+156.9% and +163.7%), with S2F at the post-halving step of ≈ 57. The sign only goes negative at the cycle lows: the 2022 post-FTX trough at $16,304.08 printed −39.1%. Read down the residual column and every cycle top scores positive while only the lows go negative — the opposite of a line that leads price.

For the two years following PlanB's March 2019 publication, S2F's model line and Bitcoin's realised price moved in close lockstep. PlanB explicitly forecast in the original piece: “The predicted market value for bitcoin after May 2020 halving is $1trn, which translates in a bitcoin price of $55,000” (Medium, 22 Mar 2019). Bitcoin reached $55,000 in late February 2021, broadly on schedule. PlanB then followed with the “Stock-to-Flow Cross-Asset” (S2FX) variant in 2020, which set higher post-2020 average targets.

What happened next is the model's defining failure — though not in the direction the original regression predicted. PlanB's 3.3/14.6 market-cap fit placed the post-2020-halving step near Bitcoin's actual 2021 highs, so the realised April $63,576.68 peak and November $67,145.37 ATH were broadly in line with that original target. This site's nightly refit tells a different story: it scores those same tops far above its own line (+156.9% and +163.7%), then scores the 2022 cycle trough at $16,304.08 −39.1% — a swing of nearly 200 points of residual across a single cycle. PlanB's later S2FX variant set still higher post-2020 targets that realised price never approached.

The model has not recovered. The 2024 pre-halving high of $73,097.77 printed +156.4% against the current fit's line — another top the model misread. The April 2024 halving roughly doubled S2F again (the ratio now sits near 122.0), which lifts the fitted model price to $213,346.58 — far above spot, leaving the largest negative residual the chart has carried. That is the reason most thoughtful operators no longer treat the model as a price target.

Stock-to-Flow has attracted heavier econometric critique than any other mainstream Bitcoin valuation model. The technical objections were published before the model began breaking from price, and the data since has borne them out.

Nico Cordeiro (Strix Leviathan), 30 June 2020. “A Chameleon Model — Why Bitcoin's Stock-to-Flow Model is Fatally Flawed” (Strix Leviathan). Cordeiro's central claim: “The model's accuracy will likely be about as successful at forecasting Bitcoin's future price as the astrological models of the past.” The structural critique — that S2F's flow term is a step function of time, so any model regressing price against it is implicitly regressing price against time — has aged exceptionally well.

Level39 (via Harold Christopher Burger, 27 March 2022). Burger's “What's up with S2F?” (hcburger.com) catalogues the strongest mathematical critique to date — the “Stock is a function of Stock” framing originally posted by the pseudonymous Level39 — that the regression is structurally circular and that adjusting for the look-ahead drops R² toward zero. Burger himself, the author of the power-law corridor, runs a separate diminishing-returns critique: S2F predicts roughly equal-magnitude rallies from each successive halving, while the empirical data shows the opposite (each cycle's peak multiple shrinks).

Eric Wall's catalogue, 1 July 2020. “A list of the greatest blows to the S2F model” (Medium) catalogues the formal critiques: Sebastian Kripfganz's cointegration analysis (no statistical evidence for the relationship); Marcel Burger's spurious-regression walkthrough; and Nick Emblow's BTConometrics work showing the S2F correlation is “entirely spurious.” Emblow's pseudonym is a play on “Bitcoin econometrics,” and his S2F work was published under btconometrics.com — an independent site, often miscredited to institutional analytics shops in secondary sources.

The 2021 cycle is the documented break. Against this site's nightly fit, the realised cycle peaks printed far above the model line — +156.9% in April 2021 and +163.7% in November 2021 — while the 2022 trough swung to −39.1% and the 2024 pre-halving high printed +156.4% above the line again. A fit that lands so far from price at every turn, and now puts spot −69.1% under its line, is what distinguishes a broken model from a mispriced market — no narrative dressing covers a line that misses every cycle top by triple digits.

The regression is mathematically circular. Stock and flow are both functions of time (cumulative issuance and current-block-subsidy respectively), so the regression of price on S2F is nearly a regression of price on time with extra steps. The high R² is structural, not predictive. Burger's 2022 walkthrough at hcburger.com catalogues the Level39 framing of this circularity (“Stock is a function of Stock”) and shows the adjusted R² collapses toward zero. Cordeiro's June 2020 critique made the same point with a different methodology and arrived at the same conclusion.

The 2024 halving did not fire. If scarcity set price deterministically, the April 2024 halving — roughly doubling S2F from the high-50s to the current ≈ 122 — should have driven a multi-fold step. Instead Bitcoin spent the months after the halving range-bound near and below its pre-halving high, exactly as a “halving-priced-in” market would. Supply changes are known years in advance, so an efficient market forward-discounts them; the swing variable for Bitcoin's price is demand from speculative and institutional flows, which S2F does not model at all. This is the cleanest live test the model has faced, and it failed it.

The residual now points the wrong way. Because the model line climbs mechanically while price does not, the residual sits in “deeply undervalued” territory more or less permanently — currently −69.1%. Treating that as a buy signal is exactly the error the model invites: a perpetually cheap reading is not a signal, it is a model that has detached from its target. Treat S2F as a historical curiosity and a cautionary tale in how a high-R² regression breaks in real time, not as a forecast.

If you accumulate on a schedule, the S2F regime labels are noise. The scarcity argument is a fine long-run framing of Bitcoin's monetary properties, but the specific regression here has not forecast forward returns since 2020. Treat the chart as context for how the scarcity narrative played out, not as operating guidance.

If you are timing the cycle, the residual is at most one column in a panel, and a contrarian one. The deep-undervalued print at the 2022 trough was, in retrospect, an extreme buy — but it was extreme because price had collapsed, not because the model is sound. Never size off the model price. Read S2F alongside instruments that fit to time rather than to a circular scarcity term: the Power-Law Corridor and the Rainbow describe the same long-run shape without claiming the halving sets price.