The Bitcoin Halving Explained: How It Works, What Happened After Each One, and What Research Says

Start hereBitcoin's code says halvings will occur approximately every four years, yet the gaps so far ran 1,319 to 1,440 days. Prices rose a lot around each one, yet a 2025 study of cryptocurrency halvings found a negative cumulative average abnormal return around halving dates. So what does the halving do, and what is still unknown?

The Bitcoin halving is a rule in Bitcoin’s code. Every 210,000 blocks, the number of new coins paid for each block is cut in half. Four halvings have happened so far, in 2012, 2016, 2020 and 2024. By our estimate the next one comes around the first half of 2028.

Prices moved a lot around each halving. Research does not agree on whether the halving caused those moves. Past patterns are not predictions, and nothing here is financial advice. This page explains how the rule works, what happened each time, and where you can check it yourself.

011. What the halving is

New bitcoins are paid out to whoever adds the next block to the chain. The whitepaper says the first transaction in each block starts a new coin owned by the creator of the block.

Bitcoin’s code calls that payment the block subsidy. Miners also keep the fees from the transactions in their block. The halving cuts only the subsidy, never the fees.

The subsidy started at 50 bitcoins per block. Since April 2024 it has been 3.125 bitcoins. In the 2024 halving block itself, fees came to about 37.6 bitcoins, roughly 12 times the new subsidy (our calculation from the block’s data). Our reading: that was one busy block, not a trend.

022. Where the rule comes from

It is easy to assume the halving is in the whitepaper.

Doesn't hold up The halving rule is not in the whitepaper. The paper was announced on 31 October 2008. It says that once a predetermined number of coins have entered circulation, fees can take over. We searched its full text and found no mention of a halving, of 210,000 blocks or of 21 million coins.

The schedule is spelled out in the launch announcement of 8 January 2009. It says total circulation will be 21 million coins. It also says the amount paid for new blocks would be cut in half every four years.

Today the rule sits in Bitcoin Core’s source code. The code sets the halving interval at 210,000 blocks. A comment beside the subsidy rule says the halvings will occur approximately every four years.

Partly A cap of 21 million bitcoins is close, but not exact. A note in the code says the total “currently happens to be less than 21,000,000 BTC for various reasons”.

033. Why “every four years” is only roughly true

Doesn't hold up Halvings do not come exactly every four years. They are counted in blocks, not in days.

The code aims for one block every ten minutes. It measures the pace over a two-week timespan. The whitepaper explains the idea: if blocks are generated too fast, the difficulty increases.

In practice, blocks came a little faster than ten minutes. Between halvings they averaged about 9.0, 9.6 and 9.9 minutes, in our calculation from block times. So the gaps were 1,319, 1,402 and 1,440 calendar days, against 1,461 days in four calendar years.

044. The four halvings at a glance

	1st	2nd	3rd	4th	Next (our estimate)
Block	210,000	420,000	630,000	840,000	1,050,000
Date (UTC)	28 Nov 2012	9 Jul 2016	11 May 2020	20 Apr 2024	first half of 2028
New coins per block	50 → 25	25 → 12.5	12.5 → 6.25	6.25 → 3.125	3.125 → 1.5625

Dates are each block’s own timestamp, read from the mempool.space block explorer.

On 19 October 2026 the chain stood at block 972,140, which left 77,860 blocks to go. At about ten minutes a block, that is roughly 540 days, or around April 2028. The real date depends on how fast blocks are found.

055. What happened to the price around each one

The price rose before and after every halving so far, and later fell hard each time. The figures below are our calculation from daily US dollar prices published by Coin Metrics. “A year before” and “a year after” mean 365 days either side of the halving day.

Halving	A year before	On the day	A year after
28 Nov 2012	about $2.76	about $12.33	about $1,007
9 Jul 2016	about $286	about $652	about $2,506
11 May 2020	about $6,943	about $8,592	about $56,612
20 Apr 2024	about $27,292	about $64,908	about $85,102

In the year before each halving, the price rose by about 347%, 128%, 24% and 138%. In the year after, it rose by about 8,069%, 284%, 559% and 31%.

Each time, the price reached a high within two years and then fell a long way.

• After 2012: a high on 4 December 2013, then a low about 85% below it on 14 January 2015.

• After 2016: a high on 16 December 2017, then a low about 84% below it on 15 December 2018.

• After 2020: a high on 8 November 2021, then a low about 77% below it on 9 November 2022.

• After 2024: a high on 5 October 2025, then about 53% lower by 30 June 2026, in daily prices published by FRED, the St. Louis Fed’s data service.

Our reading: the climb from each halving to the next high got smaller every time, from about 9,100% to about 92%. The falls got smaller too, so far.

The window you pick also changes the story. Between the 2020 and 2024 halvings, the highest price came on 13 March 2024, before the next halving. On 4 March 2024 the price also rose above its 2021 high, about seven weeks before the halving. In the three earlier cases, the price on halving day was still below its previous record.

Mining power dipped too, in our calculation from Coin Metrics’ hash-rate estimates. The network’s average computing power in the two weeks after each halving was about 5%, 4%, 21% and 3% below the two weeks before. The data does not say why.

The US Securities and Exchange Commission’s investor site puts the general point plainly: “past performance cannot predict how an investment strategy will perform in the future.”

066. What research says about cause

Still argued Whether the halving causes the price moves is still argued. The studies point different ways:

• For: a 2019 paper concluded that halving the reward “leads to an increased market value of the cryptocurrency”. It illustrated this with a rank correlation. Our reading: a correlation alone cannot show cause.

• Against: a 2025 study in the Pacific-Basin Finance Journal looked at halving events in cryptocurrencies, not Bitcoin’s alone, and found a negative cumulative average abnormal return around halving dates.

  In plain words, prices did worse than expected around those dates. It found these negative returns more pronounced in the Bitcoin ecosystem and around first-time halvings. The authors trace this to attention-driven buying before halvings, followed by a reversal.

• Mixed: a 2025 preprint, not peer-reviewed when we read it, built a synthetic Bitcoin that had no halving. It found evidence of a positive effect of the 2024 halving on the price three months later. For the 2020 halving it could not obtain a statistically significant, robust estimate of the effect.

Three other papers say less about cause. One, from 2024, studied the 2012, 2016 and 2020 halvings to forecast prices; we do not repeat its forecast.

Another, from 2025, surveyed 294 people in a crypto community in Bali, so it shows what people believe, not what prices did.

A third, from 2025, collected data on the 2012, 2016, 2020 and 2024 halvings and compared abnormal returns, focusing on the last two. Its authors suggest that returns and volatility during halvings are decreasing as the market becomes more regulated and draws more institutional investors and governments.

Our reading: the evidence leans towards unproven. Four halvings are too few to separate cause from coincidence, and the studies disagree. Each halving’s block number is also fixed years ahead, so people can act before it arrives.

077. Crypto and the stock market

In 2017–19, returns on Bitcoin and the S&P 500 had a correlation of 0.01, an IMF blog reported. That rose to 0.36 for 2020–21. In those years the two increasingly moved together. Those are two past windows, and correlations change.

A 2023 study for the Bank for International Settlements looked at who gained and lost.

It used data on major crypto trading platforms from August 2015 to December 2022. It concluded that a majority of crypto app users in nearly all economies made losses on their bitcoin holdings.

That is an estimate. The authors assumed users bought bitcoin on the day they downloaded an app. They also ran simulations in which each new user bought $100 of bitcoin in the month of the download and every month after. They measured the losses at December 2022.

After two big crypto failures in 2022, it found “large and sophisticated investors selling and smaller retail investors buying”.

088. What regulators warn

Regulators publish plain warnings about crypto. Here are a few, in their own words.

• UK, Financial Conduct Authority: cryptoassets “are all high risk and speculative as an investment”. If you invest, “you should be prepared to lose all your money, for any one of a variety of reasons”. It adds that “crypto is largely unregulated in the UK, so it is highly unlikely you will be covered by the Financial Services Compensation Scheme”.

• EU supervisors, 2022: consumers “face the very real possibility of losing all their invested money if they buy these assets”. They should be “particularly wary of promised fast or high returns, especially those that look too good to be true”.

• EU supervisors, 2025: “legal protection, if any, may be limited depending on which crypto-assets they invest in”. They ask consumers to “check the provider of crypto-asset services is authorised in the EU”.

• US, SEC and Commodity Futures Trading Commission alert: claims such as “risk-free”, “zero risk”, “absolutely safe” and “guaranteed profit” are “hallmarks of a fraud”. Fraudsters “may try to create a false sense of urgency”.

Our reading: a halving date is known well in advance, so it is easy to build a sense of urgency around it. The red flags above apply to any pitch that does that.

099. What you can do with this

• Check the countdown yourself. Open a block explorer, read the current block height and take it away from 1,050,000. That is how many blocks remain.

• Read the rule at its source. Search Bitcoin Core’s validation.cpp file for GetBlockSubsidy. The halving takes a few lines.

• Question any “halving cycle” chart. Ask how many halvings it rests on, where its window starts and ends, and whether it shows the falls. The SEC’s investor site suggests questioning any performance presentation that does not cover “both up and down markets”.

• Look up a firm before trusting it. Regulators keep public registers, such as the one linked from ESMA’s MiCA page below.

For why four data points can mislead, see our guide to numbers that mislead. For the stories we remember and the ones we forget, see survivorship bias.

1010. Where to go next

• The Bitcoin whitepaper: the original paper, nine pages long.

• The January 2009 launch announcement: where the 21 million schedule is set out.

• Bitcoin Core on GitHub: the source code, including the subsidy rule.

• mempool.space: an open-source block explorer where you can see the current block height.

• FRED Bitcoin price series: a free daily price series hosted by the St. Louis Fed.

• BIS Bulletin 69: central-bank research on retail crypto losses.

• FCA, Crypto: the basics: the UK regulator’s plain-English risk page.

• ESMA, MiCA page: links the EU’s interim register, including crypto-asset service providers.

• Investor.gov crypto fraud alert: the SEC and CFTC list of red flags.

• Investor.gov, Performance Claims: how to read past-performance figures.

11Quick answers

12Use this

13Quick quiz

How we checked this · Last checked October 2026

• We read the whitepaper PDF, the October 2008 and January 2009 mailing-list posts and Bitcoin Core's source code on 19 October 2026.
• Halving dates are each block's own timestamp, read from the mempool.space API. Day counts, minutes per block and the next-halving date are our calculations.
• Prices are our calculations from Coin Metrics daily data, which ended on 23 May 2026 when we downloaded it, and from FRED for the 2025 high and the 2026 low. The two series differ by a few per cent on some days, so each figure names its source.
• For the research we read the abstracts. The 2025 synthetic-control paper was a preprint with no journal reference when we read it. One 2024 paper publishes a dated price forecast, which we do not repeat.
• The IMF and FRED pages block scripts, so we read them with a page-fetching tool. Regulator pages were read on 19 October 2026 and may change.

How we check every article →

14Sources