TL;DR. Four halvings now exist (2012, 2016, 2020, 2024). The "halving → 18-month top" template held in three earlier samples — 371, 526, 548 days respectively. Cycle 4 topped at $126,199.63 on 2025-10-06, exactly 534 days post-halving — uncannily on schedule but at a +94% return versus roughly +9,100% / +2,940% / +700% in cycles 1–3. The "schedule still works" headline disguises a "magnitude collapsed" reality. Below we walk through what the four cycles actually share, where the Stock-to-Flow model broke, the statistical-significance problem with n=3, and the three structural changes (spot ETF, pre-halving ATH, large ETF / corporate / sovereign holders) that make cycle 4 different.
1. Four halvings — the numbers on the table.
| Cycle | Halving date | Block reward | Price at halving | Cycle top | Top date | Days post-halving |
|---|---|---|---|---|---|---|
| 1 | 2012-11-28 | 50 → 25 BTC | ~$12.33 | ~$1,135 | 2013-12-04 | 371 |
| 2 | 2016-07-09 | 25 → 12.5 BTC | ~$651.9 | $19,798.68 | 2017-12-17 | 526 |
| 3 | 2020-05-11 | 12.5 → 6.25 BTC | ~$8,592 | $69,000 | 2021-11-10 | 548 |
| 4 | 2024-04-20 | 6.25 → 3.125 BTC | $64,940.59 | $126,199.63 | 2025-10-06 | 534 |
Returns side-by-side (halving-day price → cycle top; halving prices for cycles 1–3 are Coin Metrics daily reference rates and cycle 4 is the Binance close; tops for cycles 2–4 are Binance BTCUSDT intraday highs, and for cycle 1, before Binance existed, the highest Coin Metrics daily reference rate — background: Wikipedia: Bitcoin halving history):
| Cycle | Halving price | Top price | Gain | Multiple |
|---|---|---|---|---|
| 1 | $12.33 | $1,135 | +9,100% | ~92× |
| 2 | $651.9 | $19,798.68 | +2,937% | ~30× |
| 3 | $8,592 | $69,000 | +703% | ~8× |
| 4 | $64,940.59 | $126,199.63 | +94% | ~1.9× |
2. Stock-to-Flow: the model that worked, until it didn't.
PlanB's Stock-to-Flow model treats BTC's price as a function of scarcity, defined as S2F = stock ÷ annual issuance (Binance Academy: Stock-to-Flow model explained). Halvings double S2F. PlanB's 2019 paper fitted market value = e14.6 × S2F3.3, so each doubling of S2F implies roughly 23.3 ≈ 9.8× more model market value. The S2F numbers across halvings (stock from Coin Metrics supply on halving day; annual issuance = block reward × ~144 blocks × 365):
| Time | BTC stock (approx) | Annual new BTC | S2F ratio |
|---|---|---|---|
| 2012 halving (before → after) | ~10.5M | ~2.63M → ~1.31M | ~4 → ~8 |
| 2016 halving (before → after) | ~15.75M | ~1.31M → ~657k | ~12 → ~24 |
| 2020 halving (before → after) | ~18.4M | ~657k → ~329k | ~28 → ~56 |
| 2024 halving (before → after) | ~19.7M | ~329k → ~164k | ~60 → ~120 |
| 2028 post-halving (projected) | ~20.05M | ~82k | ~240 |
Against that ~9.8× per halving, cycles 1 and 2 ran far ahead (about 92× and 30× from halving to top) and cycle 3 came in slightly under (about 8×) — close enough that the model "looked right" in hindsight. Cycle 4 broke it: the formula implied another ~9.8× (a model price on the order of $800,000 at ~19.7M coins), and BTC delivered about 1.9× (+94%). The model's explanatory power collapsed when the asset's market cap exceeded ~$1T and structural buyers (spot ETFs) changed the demand curve.
| S2F evaluation dimension | Cycles 1-3 | Cycle 4 |
|---|---|---|
| Formula-implied move per halving | ~9.8× model market value | ~9.8× model market value |
| Actual (halving → cycle top) | ~92× / 30× / 8× | ~1.9× (+94%) |
| Versus formula | Two far above, one slightly below | Far below |
| Explanatory power | "Looks right" | Clearly failed |
3. The "halving + 18 months → top" rule has an n=3 problem.
Three samples — 371, 526, 548 days — averaging ~482 days with a standard deviation of ~96 days. Cycle 4 came in at 534 days, well inside the ±1σ band. The schedule appears to repeat. But three samples is statistically inadequate: even assuming normally distributed data, with n=3 the 95% confidence interval for the true standard deviation runs from about 0.52× to 6.3× the sample figure (chi-square, 2 degrees of freedom) — anywhere from roughly 50 days to more than 600.
Even after cycle 4, n=4 is far too few to infer a rule. Each cycle's macro context has also been radically different: cycle 1 was pre-Mt. Gox collapse, cycle 2 was pre-ICO mania, cycle 3 was the pandemic stimulus cycle, cycle 4 is the spot-ETF cycle. Treating the day count as a forecasting tool is calendar superstition, not analysis.
4. Three structural changes in cycle 4.
(1) The spot ETF channel rewrote the demand structure. The January 10, 2024 SEC approval of eleven spot BTC ETFs (BlackRock IBIT, Fidelity FBTC, ARK ARKB, Grayscale GBTC after conversion from GBTC trust, Bitwise BITB, Invesco BTCO, Franklin EZBC, VanEck HODL, Valkyrie BRRR, WisdomTree BTCW and Hashdex DEFI) opened a regulated allocation channel that had been blocked for a decade. In 2024 alone the funds took in about $35.2B of net inflows (sum of Farside Investors daily totals) — a demand source structurally different from the retail FOMO that drove cycles 1–3. Cycle 4 ran on RIA allocation, institutional portfolios and corporate treasuries rather than on retail exchange accounts alone.
(2) BTC broke its prior cycle ATH before the halving — a first. On 2024-03-08 BTC traded at $69,990 on Binance, above the 2021 intraday high of $69,000, a full 43 days (about six weeks) ahead of the halving; on 2024-03-14 it went on to $73,777. In the two prior cycles with a meaningful prior peak (2016, 2020), the break came roughly 7 months after the halving (+229 and +203 days). The 2024 sequence inverted that template entirely. Spot ETF inflows pulled the new high forward by weeks, and the post-halving period through August 2024 actually saw a sharp drawdown to $49k (the August 5 flash crash) before the next leg up. Any model expecting "halving → consolidation → breakout into new highs" was structurally wrong from the outset, because the new high arrived first.
(3) ETFs, corporate treasuries and sovereigns became large holders. MicroStrategy (rebranded to Strategy in early 2025) held 640,031 BTC as of 2025-09-30, a week before the cycle-4 top (per its 8-K filing), and Tesla, Block (formerly Square), Marathon Digital and Riot Platforms maintain corporate holdings. By 2026, BitcoinTreasuries counts more than 1 million BTC held by public companies alone, against about 20.0M BTC in circulation (Coin Metrics, May 2026). The spot ETFs hold their own large slice, and sovereign holders include El Salvador, Bhutan and the US Strategic Bitcoin Reserve created in March 2025. Together these holders take a meaningful slice of the float out of day-to-day trading — check the funds' and companies' own disclosures for the current figures. That did not exist in cycles 1–3. It may have helped keep the cycle-4 top free of a retail blow-off, but it did not prevent a deep drawdown: after the 2025-10-06 top, BTC fell to $60,000 on Binance on 2026-02-06, about 52% below the high.
| Cycle | Dominant macro narrative | Primary capital source | Compliant channels |
|---|---|---|---|
| Cycle 1 (2012) | Early network effects, Cyprus banking crisis | Geeks and early speculators | Mt. Gox and other grey-market venues |
| Cycle 2 (2016) | ICO wave, blockchain narrative emergence | Retail + ICO issuers | Huobi, OKCoin, early Coinbase |
| Cycle 3 (2020) | COVID + unlimited Fed QE | Retail + early corporate treasuries | Coinbase IPO, Binance global expansion |
| Cycle 4 (2024) | ETF approval + systematic institutional allocation | Traditional institutions + ETF rails | Eleven spot ETFs + GBTC conversion |
The pre-cycle-top ATH breach is the single most distinctive structural marker of cycle 4. The table below confirms that earlier cycles all needed time post-halving before breaking the prior peak, while cycle 4 inverted the sequence entirely.
| Cycle | Date of first prior-cycle-ATH breach | Days from halving | Months from halving |
|---|---|---|---|
| Cycle 2 | 2017-02-23 | +229 days | +7.5 months |
| Cycle 3 | 2020-11-30 | +203 days | +6.7 months |
| Cycle 4 | 2024-03-08 | −43 days | −1.4 months (before halving) |
Cycle 2 uses Coin Metrics daily reference rates against the 2013 daily high (Binance did not exist yet); cycles 3 and 4 use Binance BTCUSDT intraday highs against the prior intraday top ($19,798.68 and $69,000).
5. Miner economics — how much supply actually leaves the market?
The block reward fell from 6.25 to 3.125 BTC per block on 2024-04-20 (Bitcoin whitepaper: block reward and supply curve). At ~144 blocks per day, daily new issuance dropped from ~900 BTC to ~450 BTC — about a $29M per day reduction in marginal supply at the ~$65k halving-day price. Annualized, that is roughly 164,000 BTC of new supply per year versus 329,000 BTC in the prior cycle. Compare that with spot ETF demand (Farside Investors daily net flows converted to BTC at Binance closes): across 2024 the funds averaged about 2,000 BTC of net buying per trading day, and in the week of 2024-03-11 to 03-15 about 7,100 BTC per trading day — roughly 16× the ~450 BTC of daily issuance.
| Time | Daily new BTC | Annual new BTC | Share of circulating supply |
|---|---|---|---|
| Post-2020 halving | ~900 BTC | ~328,500 | ~1.75% |
| Post-2024 halving | ~450 BTC | ~164,250 | ~0.85% |
| Change | −450 | −164,250 | Halved |
| Flow source | Daily BTC flow | Versus miner issuance in the same period |
|---|---|---|
| Miner new supply (before the 2024-04-20 halving, 6.25 BTC × ~144 blocks) | ~900 BTC/day | — |
| Miner new supply (after the 2024-04-20 halving, 3.125 BTC × ~144 blocks) | ~450 BTC/day | — |
| ETF average daily net inflow (2024, per trading day) | ~2,000 BTC (≈1,400 spread over calendar days) | ~2.4× (calendar days: 1,400 ÷ 585; 2024 average issuance ≈ (110 days × 900 + 256 days × 450) ÷ 366 ≈ 585) |
| ETF single-day peak (2024-03-12, before the halving) | ~14,600 BTC (about $1.05 billion that day) | ~16× (14,600 ÷ 900) |
ETF figures: Farside Investors daily net inflows (USD) ÷ Binance BTCUSDT close on the same day, approximate. Miner issuance is estimated as block reward × ~144 blocks per day and compared with ETF flows from the same period.
The ~16× ratio on the peak day, 12 March 2024 (14,600 ÷ ~900 BTC of pre-halving daily issuance; The Block), is the structural fact that makes cycle 4 different from every prior cycle. When demand exceeds miner supply by an order of magnitude on the most active days, the marginal price-setter is no longer the miner — it is the ETF authorized participant. This shifts the analytical framework from "miner capitulation = bottom" toward "ETF flow regime = primary driver".
Miner economic vulnerability also shifted. Public miners (Marathon, Riot, CleanSpark — all SEC filers, all with quarterly 10-Q disclosures available on EDGAR) lean more on capital markets — equity issuance, convertible notes, loans — alongside or instead of selling mined BTC, and several now hold most of their production. That means the classic "miners as the marginal seller" mental model is partially broken — financing obligations introduce flow patterns that respond to interest-rate and capital-market conditions rather than purely to BTC price. Miner outflow series from on-chain data providers are the place to check how quickly post-halving selling fades.
One emergent risk worth tracking: if BTC enters a sustained drawdown while financing costs remain elevated, BTC-collateralized loans can trigger forced liquidations across multiple lenders simultaneously — a 2022-summer-style cascade but with miners rather than Three Arrows Capital as the trigger. Miner balance-sheet leverage is worth watching as a separate cycle-stage indicator alongside on-chain Glassnode reads.
| Miner economics indicator | Post-2020 halving | Post-2024 halving |
|---|---|---|
| Daily total miner revenue (30-day average after the halving) | ~$9M | ~$33M |
| Block reward share | ~90% | ~85% |
| Transaction fee share | ~10% | ~15% |
| Hash price (USD per TH/s per day), 30 days before → 30 days after | ~$0.132 → ~$0.089 | ~$0.107 → ~$0.055 |
Method: Coin Metrics community data — block subsidy in USD plus fees (BTC × daily price), divided by network hash rate for hash price.
Across the 2024 halving, hash price roughly halved (about −49% comparing the 30 days before and after), versus about −33% in 2020, when the price recovered quickly after the halving. BTC more than doubled within 12 months of each of the first three halvings; after the 2024 halving it was up only about 31% a year later. That is the underlying reason miner margins were structurally thinner this cycle.
On 2024-04-20 at block 840,000, the BTC block reward dropped from 6.25 to 3.125 BTC. Marathon Digital's Q2 2024 10-Q (filed 2024-08-01) reported total revenue of $145.1M, down about 12% from $165.2M in Q1, even though BTC's average price was higher in Q2 (about $65,700 versus $53,600, Coin Metrics daily average). Riot's Q2 2024 10-Q (filed 2024-07-31) reported $70.0M of revenue versus $79.3M in Q1, down about 12%. Both filings are publicly available on EDGAR.
What this case shows: the halving's effect on miner economics is real and measurable in audited financial filings, not just on-chain inference. For analysts trying to validate the on-chain story, the SEC filings of public miners are an underused cross-check in audited dollars.
6. The shrinking-returns pattern vs cycle 4's actual gain.
The log-decay structure is clear: ~92× → ~30× → ~8× → ~1.9×. Each cycle, returns compress by roughly half-an-order-of-magnitude. That tracks how much bigger the base has become: BTC's market cap at each top was about $14B (2013), $330B (2017), $1.27T (2021) and $2.49T (2025-10), per Coin Metrics. Each successive cycle requires absorbing larger absolute capital inflows to produce a percentage move, which is exactly what an efficient-market thesis predicts as the asset's market cap rises relative to global capital pools.
| Cycle | Gain | Log10 | Decay ratio vs prior cycle |
|---|---|---|---|
| Cycle 1 | +9,100% | ~1.96 | — |
| Cycle 2 | +2,937% | ~1.48 | 0.32× (~68% decay) |
| Cycle 3 | +703% | ~0.90 | 0.24× (~76% decay) |
| Cycle 4 | +94% | ~0.29 | 0.13× (~87% decay) |
Log10 column is the log of the multiple (e.g. 92× → 1.96).
The decay itself is getting steeper: about 68%, 76%, 87%. A mechanical extrapolation would put cycle 5 (post-2028 halving) well below cycle 4's +94% — but with four data points that is a direction, not a forecast. The implication for analysts is that the four-year cycle metaphor itself may be approaching the end of its useful life as the asset matures.
Cycle 4's +94% slots into that decay curve, but two competing readings of "is this cycle done" matter. Argument that cycle 4 is not over: the ETF channel could re-accelerate if RIA platforms expand allocation; macro liquidity (Fed rate path, dollar strength, global M2) remains the largest swing factor and a 2026 rate-cutting cycle would mechanically lift risk assets; MVRV Z-Score computed from Coin Metrics community data was about 2.5 on 2025-10-06, versus about 9.4 at the 2017 top and 3.5 at the 2021 top, which some read as room for one more leg; total USD stablecoin supply (DefiLlama) kept expanding through the print, from about $285B in early September to about $301B on 2025-10-06.
Argument that cycle 4 is near or past peak: $126,199.63 top on 2025-10-06, followed by a sharp drop to a $112,775 daily close four days later (Binance); spot ETF flows were strong into the high (about $3.2B and $2.7B of net inflows in the two calendar weeks up to and including the ATH week, per Farside) and then turned to a roughly $1.2B net outflow the following week; and a persistently negative Coinbase premium (the spread of Coinbase spot price vs Binance) would be the classic US-side distribution flag — check it on current data rather than taking it from an archive. By 2026-05-19 BTC was near $76,800, about 39% below the high, after touching $60,000 on 2026-02-06 (Binance) — so far the second reading has fit better. The responsible move is still reading both narratives, then sizing for the bear-case rather than betting on the bull-case.
7. Cycle-top objective signals that are firing.
Through the late-2025 high zone, the classic cycle-top gauges that we can verify ran noticeably cooler than in past cycles. Binance BTCUSDT funding did not rise above the 0.01% baseline in any of the 120 settlements from September 1 to October 10, 2025 (Binance funding history). Daily RSI(14), computed from Binance daily closes, peaked at about 72.8 on the ATH day and never reached 80. The Fear & Greed Index (alternative.me) peaked at 74 on October 5. On-chain gauges such as MVRV Z-Score and STH-SOPR, and the Coinbase premium, are best read on their providers' own charts.
None of these is decisive alone, and this time the verifiable ones did not line up the way they did at the 2021 highs. Critical caveat: the 2024 cycle has shown that traditional cycle-top signals fire at lower magnitudes than in earlier cycles because the institutional capital base both dampens and prolongs price moves. Treat the signals as probability shifts, not as triggers. When several line up at once — RSI divergence, funding compression, a valuation gauge in its historical top zone, a turn in CME basis — that has tended to arrive before a top rather than at it, but how much warning it gives varies from cycle to cycle and is not a number worth anchoring on.
| Top indicator | 2017-12 top | 2021-04/11 top | 2025-10 cycle 4 ATH |
|---|---|---|---|
| Daily RSI(14) peak, final month before the top (Binance closes) | 93.5 (2017-12-07) | 67.5 (2021-04-13) | 72.8 (2025-10-06) |
| Binance BTCUSDT funding (8h peak) | — (Binance USDT-M futures launched in 2019) | +0.1486% (2021-04-10) | +0.01% (never left baseline, 2025-10-01 to 10-15) |
| F&G peak | — (index starts 2018-02) | 95 (2021-02) | 74 (2025-10-05) |
The readings you can check on free public data moderated from cycle to cycle: funding and sentiment ran far cooler into the 2025 high than in 2021, and momentum never reached the extremes of 2017. On-chain gauges such as MVRV Z-Score, the Puell Multiple and long-term-holder supply share are worth checking on their providers' own charts.
8. Seven objective signals to size against if you are still in.
For position management at this cycle stage, watch these seven signals together. (1) BTC.D drops from 60%+ to below 50% sustained for 4 weeks — classic altseason setup and frequently a late-cycle distribution marker as capital rotates from BTC into smaller-cap alts. (2) Funding rate sustains above +0.05% per 8h cycle (about 55% annualized) across BTC and ETH perpetuals — leverage saturated and the carry trade is paying historically extreme rates. (3) Daily RSI breaks 80 and holds 5+ sessions on the daily timeframe — momentum extremely extended. (4) MVRV Z-Score in its historical top zone on the provider's own chart (Glassnode and others) — read it against that provider's history rather than a fixed number. (5) Fear & Greed sustains 80+ for 6 days (alternative.me archive) — sentiment in extreme greed territory persistently. (6) Spot ETF net inflows unusually large for several consecutive days (Farside Investors) — institutional FOMO accelerating into the high. (7) Coinbase premium stays persistently negative — US-side distribution flag as Coinbase trades at a discount to Binance suggesting US sellers active. Three or more of these together at any time = high-confidence late-cycle environment that warrants exposure reduction. All seven together = full distribution regime; size down hard, tighten stops, do not chase rebounds.
| Signal | Top threshold | May 2026 reading | Verdict |
|---|---|---|---|
| BTC.D | Below 50% sustained 4 weeks | Check a dominance chart (TradingView, CoinGecko) | Not assessed here |
| Funding rate 8h | Sustained +0.05%+ | May 1–19: 57 settlements between -0.0092% and +0.0072%, mean about +0.0005%, none reaching +0.01% (computed from Binance BTCUSDT perpetual funding history) | Not triggered |
| RSI 1d | > 80 for 5+ days | About 44.5 on May 19 (Binance closes) | Not triggered |
| MVRV Z | Historical top zone | Check the provider's chart | Not assessed here |
| F&G | 80+ for 6 days | Daily values between 25 and 50, May 1–19 (alternative.me) | Not triggered |
| ETF net inflows | Unusually large, several days running | Check Farside Investors | Not assessed here |
| Coinbase Premium | Persistently negative | Check Coinbase vs Binance prices | Not assessed here |
On 2025-10-06 BTC reached $126,199.63 on Binance — the cycle 4 ATH. Checking the signals we can verify against that print: Binance BTCUSDT funding stayed at or below the 0.01% baseline into the high (Binance funding history); daily RSI(14) peaked at about 72.8 that day and never reached 80 (computed from Binance daily closes); F&G peaked at 74 on 2025-10-05 and never entered the 80+ zone (alternative.me historical API). None of these three fired at its classic threshold. For ETF flows, MVRV Z-Score and the Coinbase premium, check Farside Investors, Glassnode and Coinbase data directly.
The lesson cuts against the neat story: a real cycle high can arrive with the popular top signals well short of their thresholds. Use the stack to size down when several signals do fire together, but do not read "no signal" as "no top".
9. Three things not to do.
(1) Do not max-leverage a "halving + 18 months top" calendar bet. n=4 is statistically inadequate, and a calendar window says nothing about how deep the path gets before or after the top. (2) Do not trust S2F as a forecast. The model looked right in hindsight on cycles 1–3, but its formula implied about 9.8× per halving and cycle 4 delivered about 1.9×. Using S2F as a target is mistaking storytelling for data. (3) Do not treat "halving" as a religion. The block-reward cut is an arithmetic fact; the price response depends on the contemporary demand-side macro context. ETF inflows, dollar strength, rate path, regulatory tone and stablecoin supply all matter as much as or more than the issuance schedule.
10. Closing — what to keep, what to discard.
The halving still produces a real, measurable cut in new supply — that part is not myth. The cycle-4 evidence shows that magnitude is shrinking faster than the day count is shifting. The shape persists at the level of "approximately 18 months between halving and major top"; the slope of expected returns has collapsed from about 92× → 30× → 8× → 1.9×. Position accordingly. Discard rigid S2F price targets; discard "halving means buy" reflexes; keep the structural framework — supply cut + demand cycle + leverage saturation — but update each input with current data from Farside Investors (ETF flow), Glassnode (on-chain), Coinglass (derivatives) and Coinbase / Binance (spot) rather than backtest-fitted formulas.
Combine this with our funding rate guide, six on-chain metrics, and open interest guides for the full late-cycle SOP.
FAQ.
Does the halving still matter in the ETF era?
Yes for supply mechanics — new issuance still halves — but the demand side now dominates pricing. In 2024, US spot ETF net inflows averaged roughly 2,000 BTC per trading day, or about 1,400 per calendar day (Farside daily totals converted at Binance closes). That is about 2.4× the year's average daily new supply of ~585 BTC (900 a day before the April halving, 450 after), and on the busiest day, 12 March 2024, still before the halving, inflows ran about 16× the ~900 BTC of daily issuance, so the supply cut is less of a price catalyst than in earlier cycles.
Why did the 2024 cycle break the prior ATH before the halving?
Spot ETF approval in January 2024 opened a new demand channel: about $12.3B of net inflows arrived before the April 2024 halving (sum of Farside daily totals through 2024-04-19). BTC first traded above its 2021 high of $69,000 on 2024-03-08 (Binance), 43 days before the halving, inverting the classic “halving then breakout” template.
Is Stock-to-Flow still useful?
As a way of describing scarcity arithmetic, yes. As a price-target framework, no — PlanB's 2019 formula implies roughly 9.8× more market value each time scarcity doubles, while cycle 4 rose only about 1.9× from halving to top.
Will cycle 5 (2028 halving) repeat the same structure?
Unknown. The dominant variables in 2028 will likely be macro liquidity, regulatory state, and stablecoin supply — not the halving itself. Use the supply schedule as background, not as a forecast.
How do I track ETF flow in real time?
Farside Investors publishes daily aggregate flow tables; SoSoValue maintains a per-ETF dashboard. CME basis and Coinbase premium offer faster-moving proxies.
What to read next.
Crypto assets are volatile and not suitable for every investor. This page is editorial analysis, not financial advice.