Back to ResearchTopic: Protocol Fundamentals and On-Chain Data
Bitcoin Halving Cycles: History & Data (2026)
Education

Bitcoin Halving Cycles: History & Data (2026)

C

Research Desk • Organizational attribution

Source Standard
6 source notes
Last Reviewed
2026-10-28
Not Advice
Educational research, not a recommendation to buy or sell

Bitcoin has halved four times: at block 210,000 on November 28, 2012, block 420,000 on July 9, 2016, block 630,000 on May 11, 2020, and block 840,000 on April 20, 2024 (UTC). Each one cut the new-coin subsidy per block in half, from 50 BTC down to today's 3.125 BTC. The fifth halving happens at block 1,050,000. At the chain height of 973,532 on October 28, 2026, that was 76,468 blocks away, which works out to roughly spring 2028 depending on how fast blocks arrive.

That is the schedule. What the schedule does to price is a separate and much weaker claim. Four events is a tiny sample, each one landed in a different macro environment, and the share of supply each halving removes gets smaller every time. This page starts from what the chain itself records: block timestamps, difficulty, issuance arithmetic, and fees. Every figure below comes from Bitcoin Core's source code, from block data pulled from the mempool.space explorer (October 21, 2026, rechecked October 28), or, for the price-cycle section, from Coin Metrics daily reference closes pulled October 28, 2026, except one ETF-flow comparison sourced to Farside Investors' daily table, and the arithmetic is shown so you can redo it.

Where the Halving Lives in the Code

The halving is not a policy decision and nobody schedules it. It is four lines in Bitcoin Core's validation code. The function that sets the block subsidy divides the block height by 210,000 (the halving interval in the mainnet chain parameters), takes the whole-number result as the count of halvings so far, starts from 50 BTC, and shifts that amount right by one binary place per halving. A right shift by one place is the same as dividing by two and discarding any remainder. For one on-chain metric computed across these cycles, with dated readings, see the MVRV Z-score guide.

Two details in that function matter more than they look.

First, the work is done in satoshis, not bitcoin. 50 BTC is 5,000,000,000 satoshis. Integer halving of that number eventually produces remainders that get thrown away, which is why the total supply is not exactly 21,000,000 BTC. Summing every era's subsidy with integer division gives 2,099,999,997,690,000 satoshis, or 20,999,999.9769 BTC. That is the protocol ceiling, and the actual spendable supply will be a little lower, because some early miners claimed less than the full subsidy and the genesis block's coins cannot be spent.

Second, the subsidy reaches zero long before the code's guard clause. The function returns zero once the halving count reaches 64, because shifting a 64-bit number by 64 places is undefined in C++. But 5,000,000,000 satoshis shifted right 32 times is 1 satoshi, and shifted 33 times is 0. So the last block that pays any subsidy is in era 32, and from block 6,930,000 onward the subsidy is zero. At ten minutes per block that is around the year 2140, which is where the familiar date comes from.

The Four Halvings, From the Blocks Themselves

The table uses each halving block's own header data. The hashrate column is not a measured number, because nobody can measure total hashrate directly. It is the hashrate implied by the difficulty in force at that block: difficulty multiplied by 2 to the 32nd power, divided by the 600-second target. That conversion is standard and it smooths out luck, but it lags real hardware changes by up to two weeks.

HalvingBlockTimestamp (UTC)Subsidy changeDifficulty at blockImplied hashrate
First210,0002012-11-28 15:2450 to 25 BTC3,438,909about 24.6 TH/s
Second420,0002016-07-09 16:4625 to 12.5 BTC213.40 billionabout 1.53 EH/s
Third630,0002020-05-11 19:2312.5 to 6.25 BTC16.10 trillionabout 115 EH/s
Fourth840,0002024-04-20 00:096.25 to 3.125 BTC86.39 trillionabout 618 EH/s
Reference971,7122026-10-16 12:37(retarget, no halving)140.58 trillionabout 1,006 EH/s

Source: mempool.space block API, pulled October 21, 2026. Hashrate computed by CryptosEyes from difficulty.

A few readings from that table that rarely make it into cycle charts:

Implied hashrate rose about 62,000-fold from the first halving to the second, about 75-fold from the second to the third, and about 5.4-fold from the third to the fourth. Growth is still large, but the multiplier shrinks every era as mining moved from hobby hardware to industrial fleets.
The fourth halving's local date depends on where you were. Block 840,000 carries a timestamp of 00:09 UTC on April 20, 2024, which was still the evening of April 19 in the Americas. Sources that say April 19 and sources that say April 20 are both describing the same block.
The October 16, 2026 retarget pushed difficulty up 5.92 percent. The next retarget falls at block 973,728, and its size is known only once that block is mined; mempool.space's running estimate changed from day to day in the week this page was updated. Implied hashrate in October 2026 was about 63 percent above its level at the 2024 halving block, despite down periods along the way, which does not fit the "miners capitulate after the halving" story told before each event.

Issuance Arithmetic, Era by Era

The schedule below assumes the protocol's target of one block every ten minutes, which is 144 blocks a day and 52,560 blocks in a 365-day year. Real blocks have mostly arrived faster than that, because hashrate growth outruns the two-week difficulty correction, so halvings tend to land earlier than a ten-minute calendar predicts. Treat the dates as estimates and the block heights as exact.

EraBlocksSubsidySupply at era startNew BTC per year (at 144 blocks/day)Annual issuance vs. starting supply
00 to 209,99950 BTC02,628,000not meaningful
1210,000 to 419,99925 BTC10,500,0001,314,00012.51%
2420,000 to 629,99912.5 BTC15,750,000657,0004.17%
3630,000 to 839,9996.25 BTC18,375,000328,5001.79%
4840,000 to 1,049,9993.125 BTC19,687,500164,2500.83%
51,050,000 to 1,259,9991.5625 BTC20,343,75082,1250.40%

Calculated by CryptosEyes from the subsidy function. Supply figures are the protocol schedule, not a count of spendable coins.

Work one row by hand. In era 4, each block pays 3.125 BTC. At 144 blocks a day, that is 450 BTC a day and 164,250 BTC a year. Divided by the 19,687,500 BTC that existed when the era began, annual issuance is 0.83 percent. When the era ends at block 1,049,999, the schedule will have issued 20,343,750 BTC, which is 96.875 percent of the eventual total.

The schedule-implied supply at block 972,564, the tip on October 21, 2026, is 19,687,500 plus 132,565 era-4 blocks at 3.125 BTC each, or about 20,101,766 BTC. The real figure is slightly lower for the reasons above, but within a rounding error for most purposes.

The shrinking size of each cut

Each halving removes half of the remaining issuance, which means each one removes less in absolute terms than the last:

2012: daily issuance fell from 7,200 BTC to 3,600 BTC, a cut of 3,600 BTC per day.
2016: from 3,600 to 1,800, a cut of 1,800 per day.
2020: from 1,800 to 900, a cut of 900 per day.
2024: from 900 to 450, a cut of 450 per day.
2028: from 450 to 225, a cut of 225 per day.

Measured against the existing stock of coins, the 2028 cut takes away new supply equal to about 0.4 percent of outstanding BTC per year. In 2012 the comparable figure was 12.5 percent. Anyone arguing that the fifth halving will move markets the way the first one did needs a mechanism other than supply, because the supply effect is now about one-thirtieth as large. For how deep and how long bitcoin's declines have been in the same years, see the measured bear market table.

What the Halving Does to Miners

Miners are paid two ways: the subsidy, which the halving cuts, and transaction fees, which the halving does not touch. The subsidy has always been the larger part. The halving is therefore a scheduled 50 percent cut to most of the industry's revenue measured in bitcoin, and miners survive it only if one or more of these happen: the bitcoin price rises, fees rise, competing hashrate leaves, or the miner's own cost per unit of hashing falls.

The fee side has a real data point. Block 840,000, the fourth halving block itself, paid 37.63 BTC in fees on top of the new 3.125 BTC subsidy, across 3,050 transactions. Fees were twelve times the subsidy in that one block, because users were competing to inscribe assets in the first block of the new era. That spike did not last. Over the month ending October 21, 2026, mempool.space's per-block data averaged about 0.0231 BTC in fees per block, which is roughly 0.7 percent of total miner revenue per block at the current 3.125 BTC subsidy.

That gap is the most important number on this page for anyone thinking about the long run. The whitepaper describes the incentive moving to transaction fees once a predetermined number of coins has entered circulation. Today, more than 99 percent of what miners receive per block is still subsidy. The fee market is real, but on most days in 2026 it is small, and each halving makes the remaining subsidy smaller.

Worked example: one exahash through the next halving

Here is an illustration with stated inputs, not a forecast. Take a miner running 1 EH/s against the roughly 1,006 EH/s implied by the October 16, 2026 difficulty. Its expected share of blocks is 1 divided by 1,006, or about 0.0994 percent. At 144 blocks a day, that is about 0.143 blocks a day on average.

InputEra 4 todayEra 5, same everything else
Subsidy per block3.125 BTC1.5625 BTC
Average fees per block (30-day, Oct 2026)0.0231 BTC0.0231 BTC
Expected blocks per day at 0.0994% share0.1430.143
Expected BTC per dayabout 0.451about 0.227

The cut is about 49.6 percent rather than exactly 50 percent, only because fees are not halved. With fees this low, the fee cushion barely registers. To keep the same bitcoin income after the halving, this miner would need its network share to roughly double, which in practice means competitors switching off, or fees per block rising by about 1.6 BTC. Neither is something a miner controls, which is why the cost side (power price and joules per terahash) decides who keeps running.

Hashprice: the metric miners watch

Miners and their lenders track hashprice, which is expected revenue per unit of hashrate per day, usually in dollars per petahash per day. It falls when difficulty rises, when the bitcoin price falls, when fees fall, and by half (all else equal) at the moment of a halving. A halving is a known date for this one variable. That is why miners plan fleet upgrades, power contracts and financing around it, and why a hashprice drop around a halving is not news. The surprise, if any, is in the other inputs. We cover the full per-machine arithmetic in the miner research on this site, and the Bitcoin dominance guide covers a market-side view of how BTC's share moves through these periods.

Price Around Each Halving, Measured From Daily Closes

Cycle charts are usually drawn by eye, which lets the person drawing them pick whichever peak suits the story. Here is the same history under one written rule, applied the same way to all four halvings, so the numbers can be checked and argued with.

The price series is Coin Metrics' bitcoin PriceUSD reference rate, daily, pulled from the Coin Metrics community API on October 28, 2026 (5,946 rows from July 18, 2010 to October 27, 2026, no missing days). Each value is the reference close at midnight UTC at the end of the labelled day. The rule:

Halving-day close is the close on the UTC date of the halving block.
Cycle peak is the highest daily close in the 24 months after the halving.
Cycle trough is the lowest daily close after that peak and before the next halving. For the 2024 halving, the window runs to October 27, 2026, so the trough is provisional.
Days from each halving to the cycle peak, and from peak to troughHorizontal timeline bars for the four halvings. Halving to peak took 371, 525, 546 and 534 days. Peak to trough took 406, 364 and 366 days in the first three cycles, and 267 days so far after the 2024 halving. Peak multiples of the halving-day close fell from 92x to 30x, 7.9x and 1.9x, and peak-to-trough declines were 84.5%, 83.8%, 76.7% and 53.1% so far.Halving to cycle peakCycle peak to cycle trough0200400600800Days after the halving block (daily closes, UTC)2012Nov 282012 halving to peak: 371 days2012 peak to trough: 406 days371 d406 d92x, -84.5%2016Jul 92016 halving to peak: 525 days2016 peak to trough: 364 days525 d364 d30x, -83.8%2020May 112020 halving to peak: 546 days2020 peak to trough: 366 days546 d366 d7.9x, -76.7%2024Apr 202024 halving to peak: 534 days2024 peak to trough: 267 days (episode still open)534 d267 d1.9x, -53.1% so far
Days from each halving to the cycle peak, and from peak to troughEach bar starts at a halving block. The label after each bar gives the cycle peak as a multiple of the halving-day close, then the decline from that peak to the cycle trough. The 2024 trough segment is lighter because the decline from the October 2025 peak has not been recovered.Source: CryptosEyes calculation from the Coin Metrics community API, bitcoin PriceUSD daily reference closes, 2010-07-18 to 2026-10-27, pulled October 28, 2026. Halving timestamps from the mempool.space block API, checked October 28, 2026.
Show the cycle table behind this chart
HalvingHalving-day closeCycle peak (close)Days to peakPeak multipleCycle trough (close)Days peak to troughDecline
2012-11-28$12.332013-12-04, $1,134.9337192.0x2015-01-14, $175.6440684.5%
2016-07-09$651.942017-12-16, $19,640.5152530.1x2018-12-15, $3,185.0736483.8%
2020-05-11$8,591.652021-11-08, $67,541.765467.86x2022-11-09, $15,758.2936676.7%
2024-04-20$64,907.992025-10-06, $124,824.455341.92x2026-06-30, $58,525.08 (so far)267 (open)53.1% (so far)

Three things stand out once the rule is fixed. The time from halving to peak has stayed in a fairly narrow band, 371 to 546 days, and the last three cycles all peaked between 17 and 18 months after the halving block. The size of the move has not stayed in any band at all: the peak was 92 times the halving-day close after 2012, 30 times after 2016, 7.86 times after 2020 and 1.92 times after 2024. And the decline from each peak has been getting shallower, from 84.5% and 83.8% in the first two cycles to 76.7% in the third and 53.1% so far in the fourth.

The rule matters, and it is worth seeing where it bends. After the 2012 halving, bitcoin closed at $231 on April 9, 2013 and then fell 71.2% by July, a full crash inside what the table treats as one rising leg. A rule that looked for the first major peak after each halving would put the 2012 cycle peak in April 2013, not December. In the 2020 cycle, the highest close before the 2024 halving was not the November 2021 peak but $73,081.58 on March 13, 2024, five weeks before the next halving block. The 24-month window keeps that pre-halving rally out of the 2020 row; a window that ran all the way to the next halving would report a 1,402-day "cycle" with a 16% trough, which describes nothing useful. A cycle chart has to make a choice like this whether or not it says so. Ask which one.

The drawdown figures on this page match the peak-to-trough table of every bitcoin decline over 50%, which uses a running all-time-high rule instead of halving dates. The two methods agree on the three completed cycle tops because each one was also an all-time high. For the price path year by year, with each year's open, close, high and low on the same series, see the bitcoin price history by year. For how often 1-year to 5-year holding windows ended below their start price, with the windows that began at each record close listed separately, see the rolling holding-period returns.

Has the Cycle Changed? What 2024 Did Differently

The fourth cycle is the one people argue about, and the argument is mostly about which column of the table you look at. The data supports both readings in part, and this page does not pick a winner.

The case that the pattern held. Timing repeated. The 2024 halving was followed by a peak 534 days later, within two weeks of the 525 and 546 days of the previous two cycles, and then by a decline of more than 50%, as after every earlier peak. Someone who had only the timing column would see four similar cycles.

The case that it changed. One fact in the data had never happened before, and two others continued a long trend to new extremes:

Measure2012 halving2016 halving2020 halving2024 halving
Previous all-time-high close regained83 days after the halving (2013-02-19)229 days after (2017-02-23)203 days after (2020-11-30)47 days before (2024-03-04)
Cycle peak vs. halving-day close92.0x30.1x7.86x1.92x
Peak-to-trough decline84.5%83.8%76.7%53.1% (open)

Calculated by CryptosEyes from Coin Metrics PriceUSD daily closes, checked October 28, 2026.

Bitcoin set a new all-time-high close on March 4, 2024, before the halving block. In every earlier cycle the old high was regained months after the halving, never before. The other two rows reached their smallest values yet in 2024, but they show a steady shrinking that predates 2024, which is a reason to treat "smaller each time" as the long-running pattern rather than a 2024 break.

The explanations offered for the change are not things a price series can prove. One often given is that US spot bitcoin ETFs, approved by the SEC on January 10, 2024, brought a buyer whose flows dwarf new issuance: the ETF flow example in the next section shows a single session's outflow equal to about two weeks of mined coins. Another is simply size. On the same Coin Metrics series, bitcoin's market value (price times current supply) was about $130 million on the 2012 halving date and about $1.62 trillion on October 27, 2026, and doubling the second takes far more new money than doubling the first did. Both are plausible and neither is testable on four data points, which is the same sample-size problem the next section describes from the supply side.

Why the "Four-Year Cycle" Is a Weak Forecasting Tool

The halving schedule is precise. Price cycles drawn around it are not, and this site does not publish price targets built on them. Four reasons to be careful with any halving-based forecast:

1.Sample size. There have been four halvings. Any pattern fitted to four observations has very wide error bars, and the four sit in wildly different conditions: a market with no regulated venues in 2012, the 2016 period before futures, a pandemic and emergency monetary policy in 2020, and spot ETFs in 2024.
2.Shrinking supply effect. As the issuance table shows, the cut in new supply relative to the existing stock has fallen from 12.5 percent per year to under 1 percent. A mechanism that worked through supply should get weaker each time, not repeat.
3.Other flows dominate. Era-4 miners produce about 3,150 BTC a week at the target pace. On October 7, 2026, US spot bitcoin ETFs recorded 484.9 million dollars of net outflows, according to Farside Investors' daily flow table (checked October 22, 2026), led by IBIT at 207.7 million, FBTC at 105.1 million and ARKB at 101.7 million. Our October 8 news story covered the session. At about 83,000 dollars per coin, that one session was roughly 5,800 BTC, close to two weeks of new supply. Our ETF flows explainer shows how to read those flows, and our ETF flow impact analysis tests whether they lead price.
4.Everyone knows the date. The halving block height has been public since 2009. A scheduled, universally known event is the hardest kind of event to profit from, because there is no one left to surprise.

None of this means halvings are irrelevant. They change miner economics on a known day and they reduce new supply in a way anyone can verify. It means the honest framing is "a scheduled supply and revenue change" rather than "a price catalyst."

When Is the Next Halving?

The fifth halving is at block 1,050,000. Its block height is fixed, but its date is an estimate that moves every day.

AssumptionAverage block timeDays from Oct 21, 2026Estimated date (UTC)
Protocol target10:00 minutesabout 538around April 11, 2028
Recent pace (mempool.space, Oct 21, 2026)about 9:31 minutesabout 512around March 17, 2028

Inputs: tip height 972,564, 77,436 blocks remaining. Illustration only: block times change with every difficulty period.

Rechecked on October 28, 2026: the mempool.space tip was block 973,532, leaving 76,468 blocks. At the ten-minute target that is about 531 days (around April 11, 2028), and at the 9:24 average block time mempool.space reported for the current difficulty period it is about 499 days (around March 10, 2028). The ten-minute estimate has stayed on April 11 for a week because blocks have kept arriving faster than the target, while the recent-pace estimate moved a week earlier.

The two rows bracket a realistic range. If hashrate keeps growing faster than difficulty catches up, blocks stay a little quicker than ten minutes and the date drifts earlier. If hashrate stalls or falls, it drifts later. Watch the countdown in blocks, not days.

What Does Not Change at a Halving

A short list of things people sometimes expect to change, and do not:

The fee rules. Transactions pay the same way before and after. Your wallet does not need to do anything.
The block time target. Still ten minutes. Difficulty still retargets every 2,016 blocks.
Coins you already hold. Nothing about existing UTXOs changes.
The supply cap. The halving is how the cap is reached, not a change to it.
Network security in the immediate term. Difficulty is set by recent blocks, so the security level on halving day is the same as the day before. The longer question is what happens to the security budget as subsidy shrinks.

Common Mistakes When Reading Halving Data

Using the 21 million figure in exact arithmetic. The ceiling is 20,999,999.9769 BTC and the real spendable supply is somewhat lower. For most purposes the difference does not matter, but it does for anything claiming precision.
Mixing calendar estimates with block heights. Halvings happen at heights, and calendar estimates based on ten-minute blocks have historically been late. Quote the height.
Reading implied hashrate as measured hashrate. It is derived from difficulty, which is set once every two weeks, and it does not show sudden outages or additions until the next retarget.
Assuming fee spikes at the halving block reflect the fee market. Block 840,000's 37.63 BTC in fees was a one-block event. Use multi-week averages.
Confusing issuance with sell pressure. Miners sell some newly mined coins and hold others. Lower issuance caps how much they can sell from new production, not what they choose to do with inventory.

A Checklist for Following the Next Halving

[ ] Track the block countdown to 1,050,000, not a calendar date.
[ ] Watch the difficulty change at each retarget in the final two months. Large increases late in the era point to hardware coming online ahead of the cut.
[ ] Compare fees per block with the subsidy over a 30-day window. Note how far fees are from covering the 1.5625 BTC subsidy that era 5 will pay.
[ ] Check public miners' disclosed fleet efficiency and power costs in their filings, and look for financing that depends on hashprice.
[ ] Ignore any article that states a post-halving price as fact. Scenarios are fine when the assumptions are stated. Forecasts presented as certainties are not.

Frequently Asked Questions

When is the next Bitcoin halving?

At block 1,050,000. From a tip height of 973,532 on October 28, 2026, that is 76,468 blocks away, or about 531 days at the ten-minute target and closer to 499 days at the late-October 2026 pace. Either way, it falls in roughly March to April 2028. The height is fixed; the date estimate moves.

How many bitcoin are created per day after the 2024 halving?

About 450 BTC a day at the target pace: 3.125 BTC per block times 144 blocks. When blocks arrive faster than every ten minutes, actual daily issuance is a little higher, and when they arrive slower it is a little lower.

What happens when the block subsidy reaches zero?

From block 6,930,000, around the year 2140, miners are paid only from transaction fees. The protocol does not need a change for that to happen. Whether fees then pay for enough hashing to keep the network secure is an open question, and in 2026 fees are well under 1 percent of miner revenue on most days.

Can the halving schedule or 21 million cap be changed?

Only if the people who run full nodes adopt software with different rules, and the rest of the economy follows. There is no administrator who can change it. The incentive against changing it is strong, since altering the supply schedule would undermine the property that makes the asset distinct.

Why did the 2024 halving happen on April 19 in some sources and April 20 in others?

Block 840,000 has a timestamp of 00:09 UTC on April 20, 2024. That was the evening of April 19 in North and South America. Both dates refer to the same block.

Does a halving make Bitcoin transactions cheaper or more expensive?

Not directly. Fees are set by competition for block space, which the halving does not change. Over the long run, a smaller subsidy means miners rely more on fees, but that is a slow shift, not a halving-day effect. Our Bitcoin transaction fees guide explains how fees are actually set.

Related Reading

The halving cuts issuance on a fixed schedule, but the network keeps block times near ten minutes through a separate mechanism. Our difficulty adjustment explainer walks through that retarget rule using the same October 2026 block data, and it is the right next step if you want to understand why hashrate growth does not make blocks permanently faster. For the basics of how blocks and proof of work fit together, start with how a blockchain works and proof of work vs. proof of stake. For the risk side, see our guide to surviving Bitcoin volatility.

CryptosEyes publishes educational research, not investment advice. Halving dates beyond the next block height are estimates, and nothing on this page is a price forecast. Crypto assets can lose most or all of their value.

Sources

Primary documents, market context, and methodology references behind this article.

Reviewed against source notes and calculations
Education
Research note: This article is educational market research, not financial advice. Crypto and public equity data can change quickly; see our methodology and editorial policy for sourcing, review, and correction standards.
Important: Educational Purposes OnlyThe data, charts, treasury tracking metrics (including mNAV and SPS), and research provided on CryptosEyes.com are for informational and educational purposes only. They do not constitute certified financial, investment, or trading advice. Digital assets like Bitcoin and Ethereum are highly volatile. Always conduct your own research and consult with a registered financial advisor before making investment decisions.