A Popular Bitcoin Price Model Predicted Six Figures Years Early. Here's What the Stock-to-Flow Model Actually Got Wrong
6 min read
Loading...
6 min read
In 2020, a chart with a near-perfect fit to Bitcoin's price history went viral. It said Bitcoin was worth about $288,000 within a year or two. Bitcoin's actual high that cycle was $68,789. The chart came from the stock-to-flow model, one of the most shared, and most argued-about, price models in crypto. Its own creator has since admitted parts of it failed. Here is what it predicted, what actually happened, and by how much.
Stock-to-flow is a scarcity ratio borrowed from commodities like gold and silver. It divides an asset's existing stockpile (the "stock") by how much new supply gets produced each year (the "flow"). A high ratio means an asset is hard to inflate away, since new supply is small next to what already exists.
In March 2019, a pseudonymous analyst who goes by PlanB published "Modeling Bitcoin's Value with Scarcity" on Medium. It applied stock-to-flow to Bitcoin, whose supply schedule is fixed and public: a halving event roughly every four years cuts new coin issuance in half. PlanB's regression fit Bitcoin's historical market cap against its stock-to-flow ratio and reported a very high statistical fit (an R-squared of 0.95), then used that fit to project forward. The model's first headline call: Bitcoin would be worth roughly $55,000 on average across the 2020-2024 halving cycle.
The 2020-2024 cycle has now fully played out, and PlanB has given his own verdict. In a follow-up post on X, he wrote that the average bitcoin price across that cycle came in at about $34,000, versus the model's original $55,000 call, a shortfall of roughly 38%. He described that as "a bit below" the prediction and still within a range he considered normal.
That is a meaningfully more modest result than the model's reputation suggests, and it came from the person who built the model, not a critic.
The bigger miss came from a follow-up version. In 2020, PlanB published S2FX, a "cross-asset" variant that added silver and gold's own scarcity data into the same regression. S2FX pointed to a fair value near $288,000 for Bitcoin during the 2021-2024 window.
Bitcoin's actual all-time high for that entire cycle was $68,789, reached on 10 November 2021. The S2FX call missed by roughly 76%. Bitcoin never got within striking distance of $288,000 during the window the model named.
A third, narrower model made an even more specific and more public call. PlanB described it as a "worst case" floor, built from price and on-chain data rather than stock-to-flow directly, though it was published and read as part of the same broader forecast. It set a monthly floor of $98,000 for November 2021 and $135,000 for December 2021, meaning Bitcoin should not close below those levels in those months.
Bitcoin's November 2021 high was $68,789, already short of the $98,000 floor, and price kept falling from there. By January 2022, Bitcoin had dropped below $40,000. On 2 January 2022, PlanB tweeted: "floor model failed, I will not use it again."
Here is the part that keeps the model's fans engaged: Bitcoin did eventually cross $100,000. It happened for the first time on 4 December 2024, roughly three years after the floor model called for $98,000 to $135,000 in November and December 2021.
Three years is not a rounding error. An investor who trusted the 2021 timeline and treated $100,000 as imminent would have watched Bitcoin spend all of 2022 and 2023 trading well below that level, including a fall to roughly $15,479 around the FTX collapse in November 2022, a drawdown of about 77% from the 2021 peak.
After the actual 2020-2024 cycle underperformed his own 2019 call, PlanB refit the stock-to-flow model on updated data. The new call: an average price near $500,000 for the 2024-2028 halving cycle, and $4,000,000 for the cycle after that.
The 2024 halving landed on 19 April 2024, with Bitcoin trading around $63,762 that day. The cycle's peak so far came on 6 October 2025, at roughly $126,198, already about 75% below the $500,000 average the refit model calls for across the full cycle. Bitcoin has since pulled back further. As of 17 September 2026, it trades at roughly $76,000, according to CoinDesk, a figure corroborated by Coinbase and Bybit pricing the same morning.
| Prediction | Model | Window | Called price | What actually happened |
|---|---|---|---|---|
| Cycle average, 2020-2024 | Original S2F (2019) | 2020-2024 | ~$55,000 | ~$34,000 average (PlanB's own figure) |
| Cycle high, 2021-2024 | S2FX (2020) | 2021-2024 | ~$288,000 | $68,789 high (10 Nov 2021) |
| Monthly floor | Floor model | Nov-Dec 2021 | $98,000 / $135,000 | $68,789 high, then a fall below $40,000 by Jan 2022 |
| Cycle average, 2024-2028 | Refit S2F (2024) | 2024-2028 | ~$500,000 | ~$126,198 peak so far (6 Oct 2025); ~$76,000 now |
Every miss above points at the same gap. Stock-to-flow describes new supply, which is genuinely fixed and predictable. It says nothing about demand: who actually buys Bitcoin, when, and why. Spot Bitcoin ETF flows, interest rate policy, regulatory decisions and macro risk appetite all move price on timescales the model has no input for. Ethereum co-founder Vitalik Buterin was among the public critics who mocked the model's track record after the 2021-2022 miss became hard to ignore.
None of this means Bitcoin's price is unpredictable or that scarcity is irrelevant. It means a model built on one variable, supply, will keep missing whenever demand moves independently of it, which is most of the time.
A model that predicted six figures three years early, and a floor its own creator abandoned, is a reasonable case for treating any single bitcoin price target with real skepticism, no matter how confident the chart looks.
If you hold Bitcoin or are weighing a position, Bitcoin's own price and volatility history is a better starting point than any single forecast. The Explore screener also lets you compare Bitcoin's actual volatility and drawdown history against other assets before you size a position, rather than anchoring to a price target a model's own author no longer stands behind.