Stock-to-Flow
No predictive value
Stock-to-flow divides the existing supply of an asset by the amount produced each year. The model asserts that Bitcoin’s market value follows this scarcity ratio, so that each halving — which doubles the ratio — implies a step change in price.
- Bitcoin price (monthly close)
- Stock-to-flow model price (ln MV = 3.3 × ln S2F + 14.6)
Bitcoin price: Coin Metrics Community Data (BTC reference rate, daily, monthly closes, through 14 August 2026). Model line: PlanB’s published formula from the March 2019 paper, evaluated per halving epoch with his own stock-to-flow values — 25 for 2016–2020 and 52 for 2020–2024, which reproduce the $55,000 figure in the paper to within $62. The 2024–2028 segment uses the ratio implied by the emission schedule (about 120) and is our extrapolation, not a figure PlanB published. The step at each halving is the model, not a data error: an exponent of 3.3 multiplies the model by 2^3.3 ≈ 9.8 whenever the ratio doubles.
How it is built
The ratio is computed from Bitcoin’s known emission schedule, and market value is regressed against it in logarithms. The published relationship is ln(market value) = 3.3 × ln(S2F) + 14.6, reported with an R² of 95 per cent. Because the emission schedule is fixed in advance, that curve produces dated price targets years ahead — which is the whole appeal, and the reason the model can be checked.
Our verdict
No predictive value. The stock-to-flow ratio itself is sound arithmetic and correctly describes Bitcoin’s emission schedule — but that schedule was public knowledge already, and the ratio adds nothing to it. Everything the model claims beyond it comes from the regression, and the regression fails on the test its own author proposed: judged over whole epochs, it was 1.7 times too high for 2020–2024, its successor was nine times too high, and its extrapolation to the current epoch sits ten times above the market.
The record
Every line below is our own calculation on the public price series — reproducible where it is our own calculation, and linked to its primary source where it is quotation.
| When | What the chart says | What the data shows | Source |
|---|---|---|---|
| 2019-03-22 | Published in “Modeling Bitcoin’s Value with Scarcity”: “The predicted market value for bitcoin after May 2020 halving is $1trn, which translates in a bitcoin price of $55,000.” PlanB called it “a great out of sample test of this hypothesis and model”. | Over that epoch — 11 May 2020 to the next halving on 20 April 2024 — Bitcoin averaged $32,843 and its median day closed at $29,938. It traded at or above $55,000 on 11 per cent of the days in the epoch. Judged over the whole cycle, which is the test its author asks for, the model sat about 1.7 times above the market. | PlanB, “Modeling Bitcoin’s Value with Scarcity”, 22 March 2019 · archived |
| 2020-04-27 | The follow-up model S2FX replaced the time series with a regression across six asset clusters, reported a 99.7 per cent R², and stated: “Market value = exp(12.7598) * 56 ^ 4.1167 = $5.5T. This translates into a BTC price (given 19M BTC in 2020–2024) of $288K.” | Bitcoin never reached $288,000, in that epoch or since. Its highest price to date is $124,824, on 6 October 2025. The same author’s two models put the same four years at $55,000 and $288,000 — a factor of five apart, from the same scarcity input. | PlanB, “Bitcoin Stock-to-Flow Cross Asset Model”, 27 April 2020 · archived |
| 2021-12-24 | Asked about missed targets, PlanB drew the line himself: “Let me be clear: 1) Nov98K & Dec135K miss = FLOOR MODEL FAIL 2) Dec100K miss: I have said earlier that if BTC<$100K Dec2021, then S2F is “off the rails” “invalidated” “dead”. Tbh that was an ill attempt to clarify “$100K average this cycle” & regression analysis. S2F MODEL INTACT” | The separation is fair and this page keeps it: the monthly floor numbers were a different model, and their failure is not S2F’s. What the same message also shows is the falsification condition being withdrawn after the fact — a stated threshold for calling the model dead, reclassified as an “ill attempt” once price fell through it. | PlanB on Twitter, 24 December 2021 · archived |
| 2020-06-30 | Scarcity drives value, and gold is the benchmark that shows it. | Nico Cordeiro of Strix Leviathan checked the benchmark. Gold’s stock-to-flow ratio has sat near 60 for about 115 years while its market capitalisation ranged from roughly $60bn to $9trn. The same ratio, three orders of magnitude of value: at the reference asset, the claimed relationship is absent. | Nico Cordeiro, “A Chameleon Model — Why Bitcoin’s Stock-to-Flow Model is Fatally Flawed”, Strix Leviathan, 30 June 2020 · archived |
| 2026-08-14 | Each halving doubles scarcity, so each halving implies a step change in price. | Extending the published formula to the current epoch’s stock-to-flow of about 120 gives a model price near $806,000. Bitcoin has averaged $84,272 since the 2024 halving and stood at $62,925 on 14 August 2026 — about a tenth of the model. This extrapolation is ours, not PlanB’s; the exponent of 3.3 is what makes each halving multiply the model by roughly ten, and that multiplication is built into the functional form rather than found in the data. | Coin Metrics Community Network Data — BTC reference rate and hash rate, daily |
Why the method does not hold
01
Both series trend strongly upward over the sample, which is the classic setup for a spurious regression: two rising lines fit each other well without either explaining the other. A 95 per cent R² between two trending series is the expected result, not evidence.
02
The stock appears on both sides of the equation. The left-hand term is market value, which is stock × price; the right-hand term contains stock ÷ flow. Since stock grows monotonically, the two sides correlate before scarcity has explained anything.
03
The model contains no demand term at all. It cannot account for why the same fixed emission schedule accompanied $67,542 in November 2021 and $16,524 in December 2022 — the schedule was identical in both. Whatever explains that difference lies outside the model, and it is most of what moves the price.
04
The sample is tiny. Bitcoin has had four halvings; the effective number of observations for “what follows a halving” is three or four, not the number of monthly data points, which are highly autocorrelated within each epoch. S2FX makes this worse rather than better: a 99.7 per cent R² across six chosen points and two parameters is an overfitting signal, not a seal of quality.
05
Because flow falls towards zero, the ratio rises without bound and the model price diverges. A functional form that goes to infinity is not interpretable outside a narrow window, and nothing in the model says where that window ends.
The strongest case for it
The fair version of the defence separates two things that are usually conflated, and its author separated them first: the short-horizon “floor” numbers were a different model, and their failure is not evidence against S2F. On S2F itself the defence is that it was never a monthly price ticker but a statement about whole halving epochs, and that its qualitative core — scarcity steps up, price steps up over cycles — has held across 2012, 2016, 2020 and 2024. That is a real argument, and the epoch test is the right one to run. It is also the test the model loses: over its own headline epoch the market averaged 60 per cent of the model price, and the successor model missed by a factor of nine. And the weak version of the claim — that Bitcoin has risen over multi-year spans — needs no model at all. What was sold was the number.
What to look at instead
Supply is knowable; demand is the part that moves. ETF flows and stablecoin supply measure money arriving, which is the side stock-to-flow leaves out entirely.
Questions
What did the stock-to-flow model actually predict?
Two published numbers, both dated. The March 2019 paper predicted a $1 trillion market value after the May 2020 halving, “which translates in a bitcoin price of $55,000”. The April 2020 S2FX follow-up predicted $288,000 for the same 2020–2024 epoch. Bitcoin averaged $32,843 over that epoch and peaked at $73,082 within it.
Is the stock-to-flow model dead?
Its arithmetic still works and its author still defends it, so “dead” is the wrong word — the useful question is whether it forecasts. Tested the way its author asks, over whole halving epochs rather than single months, it does not: 2020–2024 came in at about 60 per cent of the model price, S2FX at about a tenth of its target, and the current epoch is running at about a tenth of what the formula implies.
Didn’t PlanB admit the model failed?
He admitted a different one failed, and the distinction is real. On 24 December 2021 he wrote that the missed November and December targets were a “FLOOR MODEL FAIL” while S2F itself was “INTACT” — those monthly numbers came from a separate floor model, and treating their failure as a refutation of S2F would be unfair. The same message is also where he withdrew the threshold he had set for calling S2F invalid, describing it as an “ill attempt” at clarification.
Why is a 95% R² not good enough here?
Because both sides of the regression rise over the sample, and any two rising series fit each other well — the standard warning about spurious regression. It is worse than that here: market value is stock × price and stock-to-flow is stock ÷ flow, so the same growing stock sits on both sides. The fit is partly the model regressing against itself.
Does scarcity not matter at all?
It matters, but it cannot set a price on its own — something has to want the scarce thing. The clearest evidence is the model’s own benchmark: gold has held a stock-to-flow ratio near 60 for roughly 115 years while its market value moved between about $60bn and $9trn. Same scarcity, three orders of magnitude of value. Whatever drove that came from the demand side, which stock-to-flow does not contain.
Sources & method
Every figure on this page comes from one public price series and arithmetic anyone can repeat. Where a claim would need a source we could not verify, it is not on this page.
- Coin Metrics Community Network Data — BTC reference rate and hash rate, daily — the price series every actual-price figure on this page is computed from
- PlanB, “Modeling Bitcoin’s Value with Scarcity”, 22 March 2019 · archived — the original model: the formula, the 95 per cent R², and the $55,000 forecast in the author’s own words
- PlanB, “Bitcoin Stock-to-Flow Cross Asset Model”, 27 April 2020 · archived — the S2FX model, the $288,000 target, and the author’s own caveat that it “has not yet been replicated and reviewed by others”
- PlanB on Twitter, 24 December 2021 · archived — the author separating the floor model from S2F, and withdrawing his own invalidation threshold
- Nico Cordeiro, “A Chameleon Model — Why Bitcoin’s Stock-to-Flow Model is Fatally Flawed”, Strix Leviathan, 30 June 2020 · archived — the gold counter-example: constant stock-to-flow across 115 years, market value spanning three orders of magnitude
- Moonkelp’s own analysis — the epoch averages and the model line — method described above, reproducible from the series
The other charts
Perspectives, not investment advice. This page criticises a method, not the people who publish it — and it says so where a construction does something real. How Moonkelp works
This study is reviewed quarterly; corrections normally ship within days. Corrections:contact@moonkelp.com