Bitcoin Monthly Returns: Historical BTC Performance by Month

Bitcoin monthly returns measure how much BTC’s price changed from the close of one calendar month to the close of the next. Laid out as a grid, with one row per year and one column per month, the history shows two things a price chart hides. The same month can produce very different results from year to year. And a handful of extreme early months shape the long-run averages. In the BTC table, single-month moves run from losses near 40% to gains of several hundred percent. Each month column also has only about sixteen observations, so the history is better at showing how wide the range is than at predicting what comes next.

What each cell measures

Each cell in the monthly table is a simple percentage change between two closing prices:

Monthly return = (closing price this month ÷ closing price last month) − 1

The quarterly table uses the same formula with quarter-end closes: Q1 runs from the December close to the March close, Q2 from March to June, and so on. The Year column runs from the prior December close to this December close.

Bitcoin trades around the clock, so no exchange closing bell ends the month. A monthly “close” is a fixed timestamp the data provider picks, often midnight UTC on the last day of the month. A series that uses a different cutoff, or prices from a different exchange, can show slightly different numbers for the same month. Those gaps are usually small next to bitcoin’s monthly swings, but they explain why two BTC return tables rarely match to the second decimal.

The row for the current year contains an in-progress cell, marked separately from the up and down cells in the legend. That cell measures the move from last month’s close to the latest available price. It is a running figure, not a result, and it changes until the month closes. The same applies to the current quarter and to the current year’s total.

Why monthly returns don’t add up to the yearly return

Returns compound, so you multiply them rather than add them:

Yearly return = (1 + r₁) × (1 + r₂) × … × (1 + r₁₂) − 1

A hypothetical example shows why this matters. Suppose BTC rises 50% one month and falls 50% the next. Adding the two gives 0%. The actual result is 1.5 × 0.5 − 1 = −25%. A $10,000 position would grow to $15,000 and then fall to $7,500. Because losses and gains are not symmetric, a year with several large down months can end deeply negative even when the sum of its monthly percentages looks close to flat. The Year column already does the compounding, so use it, not a sum of the row, for full-year performance.

The same arithmetic lets you turn a return back into a price. If you know a month’s opening reference price and its return, the closing price is start price × (1 + return). A month that opened at a hypothetical $40,000 and returned −12% closed at $35,200.

How to read the table and summary cards

The grid rewards reading in three directions.

• Down a column shows how one calendar month performed across all the years in the sample. This is the seasonality view: you can see whether a month’s results cluster or scatter.
• Across a row shows the rhythm of a single year, month by month, ending with its full-year return. Laying the current row beside past rows at the same point shows whether this year is tracking a strong or weak pattern so far.
• The summary rows at the bottom compress each column into an Average and a Median. They summarize the history; they don’t forecast anything.

The Month and Quarter views use the same layout at different resolutions. Quarterly figures smooth out some single-month noise, but each quarter still has only one observation per year, so the sample stays small.

Above the table, the summary cards show month to date, quarter to date and year to date returns for the selected coin. These are the in-progress cells pulled out for quick reading. Another card shows the historical median for a given month along with the number of years behind it. That year count is as important as the median itself. A median built on sixteen observations is a thin statistic, and one built on six is thinner still.

The coin selector switches between BTC and several other large assets (ETH, SOL, XRP, DOGE and BNB). The Export CSV button downloads the table as raw data for your own calculations.

Average versus median: which summary to trust

The two summary rows answer different questions, and with bitcoin they often disagree.

• The average (arithmetic mean) adds every value in the column and divides by the count. One extreme month can pull it far from the typical result.
• The median is the middle value when the column is sorted. Half the years did better and half did worse. A single outlier barely moves it.

The November column shows the effect. It includes November 2013, when BTC rose about 450% in one month during its first major run to above $1,000. That single cell lifts November’s average to several times its median. Read the average alone and November looks like an extraordinary month in a typical year. The median tells you that a typical November was a respectable gain, not a multiple. The quarterly table has a more extreme case: Q2 2011 shows a gain of more than 2,000%, from a time when bitcoin traded for a few dollars on very thin markets. That one quarter dominates the Q2 average.

For a sense of what a typical year in a given month looked like, the median is usually the better guide. The average answers a narrower question: what would the arithmetic mean of these past returns be, giving the outliers full weight.

The geometric mean

Neither row tells you the compound growth rate a holder actually earned. That is the geometric mean:

Geometric mean = [(1 + r₁) × (1 + r₂) × … × (1 + rₙ)]^(1/n) − 1

For a volatile asset, the geometric mean is always below the arithmetic average, and the gap widens as volatility rises. Using the earlier hypothetical, +50% and −50% average to 0%, but the geometric mean is about −13.4% per period, which matches the real loss. If you export the data to estimate a long-run growth rate for bitcoin per year, compute it from the Year column with this formula, not by averaging the yearly percentages.

Win rate

Win rate is the share of years in which a month closed higher. The table doesn’t print it as a row, but you can count it from any column. Take October, the month behind the “Uptober” nickname. Counting the completed Octobers in the BTC table from 2010 through 2026 gives 11 up and 6 down. That’s a better record than a coin flip, but six Octobers still closed lower, including 2011, 2014, 2018 and 2025. September, often called bitcoin’s weakest month, has a negative median in the table, yet it closed higher in each of the four years from 2023 through 2026.

A win rate also says nothing about size. A month that rises slightly in most years and crashes in a few can show a high win rate and a poor average. Check the win rate, the median and the worst cell in the column together.

Seasonality: what the month columns can and cannot show

Seasonality is the idea that certain calendar months tend to perform better or worse. In BTC’s table, some columns lean positive and others lean negative. That is a description of the past sample, and a few properties of the data limit how far it can be pushed.

The samples are small. Each column has roughly sixteen data points for BTC and fewer for the other coins. With so few observations, two or three unusual years can flip a month’s median from positive to negative.

Twelve columns invite false patterns. If you compare twelve months, a few will look strong and a few weak even if returns were random. A month that ranks best or worst in hindsight may not keep that rank.

There is rarely a mechanism. Equity seasonality can sometimes be tied to things like tax deadlines or fund reporting dates. For bitcoin, monthly results in the table often line up with specific events such as exchange failures, regulatory actions and macro shocks, and those don’t repeat on a calendar.

Early years carry outsized weight. Monthly data in the BTC table begins in late 2010. In 2011 to 2013, trading took place on a few small venues, with Mt. Gox dominant until it collapsed in early 2014. Moves of 100% or more in a month were possible on very little capital. Those years remain in every column’s statistics.

A practical way to use seasonality is to treat a column as a description of the range of outcomes for that month, not a direction. If a month’s history includes both +40% and −35% cells, that’s the useful finding: position sizing should assume either is possible.

Yearly and quarterly returns and the halving cycle

The Year column shows the boom and bust pattern that has defined bitcoin’s history. Large gains in 2013 and 2017 were followed by losses of roughly 58% in 2014 and 73% in 2018. A strong 2020 and 2021 gave way to a loss of about 64% in 2022, before gains of over 100% in both 2023 and 2024.

Many market participants link this rhythm to the halving, the scheduled cut in the block reward paid to miners roughly every 210,000 blocks (about every four years). The halvings so far:

The argument is that a lower rate of new supply, against steady or rising demand, has preceded price peaks about a year or so later: late 2013, late 2017 and late 2021. In the table, the calendar year after each of those peak years was negative. That sequence has played out three times, which is a very small sample for a cycle theory. The halving is also known years in advance, so its timing is no surprise to anyone holding bitcoin. Read the cycle as history that frames the Year column, not as a schedule.

The data also shows shrinking peak-year gains: thousands of percent in 2013, over a thousand in 2017, and far less in the cycles since. That fits a larger, more liquid market, where moving the price by a given percentage takes more capital. Monthly extremes have narrowed too. The largest single-month moves in the table sit in 2011 to 2013, and later years show smaller maximums. Large monthly drawdowns, though, persist through the most recent cycles.

What moves bitcoin’s monthly returns

Several of the deepest red cells in the table coincide with well-documented events. Matching a cell to a headline doesn’t prove the headline caused the whole move, but it shows the kinds of shocks that have produced bitcoin’s worst months:

• February 2014 (about −27%): Mt. Gox, then the largest bitcoin exchange, halted withdrawals and filed for bankruptcy.
• March 2020 (about −25%): the COVID-19 market crash, when risk assets sold off worldwide. April 2020 then rose more than 30%.
• May 2021 (about −35%): Chinese authorities moved against crypto mining and trading, and mining capacity began leaving the country.
• May and June 2022 (about −15% and −37%): the collapse of the Terra/LUNA stablecoin system, followed by withdrawal freezes at crypto lender Celsius and the failure of the fund Three Arrows Capital.
• November 2022 (about −16%): the collapse of the FTX exchange.

Positive catalysts show up less neatly. US spot bitcoin ETFs began trading in January 2024, yet January 2024 itself was almost flat. The large gains came in February and March. Anticipation, timing within the month and how much of the news was already priced in all affect which cell an event lands in.

Beyond specific events, recurring drivers include:

• Macro conditions: interest rate expectations, dollar strength and broad risk appetite. Bitcoin has often moved with equities during sharp sell-offs.
• Leverage: derivatives markets let traders hold large positions on small margin. Forced liquidations can extend a move well past where spot selling alone would take it, which helps explain why bitcoin months are so often either big up or big down.
• Flows into and out of investment products and exchanges: since 2024, spot ETF creations and redemptions have been a visible source of demand.
• Sentiment: euphoric and fearful phases can feed on themselves within a month.

Comparing BTC with other coins

The coin selector applies the same layout to ETH, SOL, XRP, DOGE and BNB. Two differences matter when you compare.

First, the histories are shorter. The median card shows the number of years behind each figure, and for the other coins that count is well below bitcoin’s. A month’s median for a coin with six or seven years of data can hinge on one or two results.

Second, the extremes are larger. Smaller assets have posted single-month and full-year moves far beyond BTC’s in the same periods, in both directions. A strong altcoin month next to a modest bitcoin month in the same row does not mean the altcoin was the better holding across the year. Compare the Year columns, and check the down years, before drawing that conclusion.

Comparing rows across coins for the same year shows when assets moved together and when they split. In broad sell-offs, the down cells tend to line up across coins. Periods when smaller coins outperform bitcoin are tracked more directly by the Altcoin Season Index and Bitcoin Dominance.

Working with the exported data

The CSV export gives you the raw monthly or quarterly percentages, which is enough to calculate:

• Win rate per month: count the positive cells in a column and divide by the completed years. Leave out the in-progress cell.
• Best and worst per month: the maximum and minimum of each column, which show the range a holder has had to tolerate.
• Standard deviation per month: a measure of how widely results scattered. Two months with similar medians can differ greatly in risk.
• Compound return over any holding period: chain the monthly figures with the multiplication formula above.

A hypothetical example makes the last step concrete. Say a holding spans three months with returns of +10%, −20% and +15%. The period return is 1.10 × 0.80 × 1.15 − 1 = +1.2%, not the +5% you’d get by adding the three figures. A $5,000 position would end at about $5,060, before fees and taxes. For a specific entry and exit price, the Crypto Profit Calculator does this arithmetic directly.

If you exclude the earliest years when computing your own statistics, say so in your notes. Results that start in 2015 or 2017 describe a different market from results that include 2011 to 2013, and both are legitimate as long as the window is stated.

Limits of historical monthly returns

These are price returns only. The figures don’t include trading fees, spreads, custody costs or taxes, and a real investor’s result for the same period will be lower. They also ignore anything earned or paid while holding, such as lending yield or funding costs on leveraged positions.

The month boundary is arbitrary. A crash in the last days of one month and a rebound in the first days of the next splits one event across two cells. One month looks terrible, the next looks strong, and neither describes what happened. A month’s high and low can also differ far more than its close to close return suggests.

Past outcomes are not probabilities. A column’s win rate describes what happened in a small set of years with very different market structures. Bitcoin’s market today, with spot ETFs, large derivatives venues and institutional holders, is not the market of 2013, and the data treats those years as equivalent observations.

Exchange and data source differences. Early prices in particular come from a few venues, and different providers handle those years differently. Treat the oldest cells as approximate.

Volatility clusters. Large moves tend to follow large moves. A calm stretch of cells doesn’t mean the next month will be calm, and a run of big months often marks a regime that ends abruptly.