Crypto position sizing by volatility: the math, real numbers
Position size is your risk divided by your stop distance. The math, and why a stop set by each coin's own hourly range was hit about as often on 10 coins.
Position sizing by volatility means holding less of a coin that moves more. The math is one line: position size equals the money you accept losing divided by the stop distance, so $1 of risk with a 2% stop is a $50 position and with a 5% stop a $20 position. A stop set at 3 times each coin's own recent average hourly range was reached within 24 hours in 33.95% to 35.78% of windows on all ten Binance coins we checked, so a range-based stop meant about the same thing on a calm coin and a wild one. These are past cases, not a forecast.
This guide explains the arithmetic with measured numbers. It does not tell you how much to hold or where to put a stop.
For how often each fixed stop distance was touched, see How wide should a crypto stop loss be?.
What is the basic math?
Start with the most you accept losing on one idea. Call it your risk. Then pick a stop distance: how far the price can move against you before you leave the trade, as a percent of the entry price.
The position size is the risk divided by the stop distance. With $1 of risk:
| Stop distance | Position per $1 of risk |
|---|---|
| 1% | $100 |
| 2% | $50 |
| 5% | $20 |
| 10% | $10 |
A wider stop means a smaller position for the same risk. That is all the idea needs. The hard part is picking a stop distance that means the same thing on a calm coin and a wild one.
How big is a normal day for different coins?
We measured each coin's daily move, ignoring direction, on Binance USDT perpetual contracts. A perp is a futures contract that follows a coin's price and never expires. The median is the middle value: half of days were smaller and half were bigger.
The last column is the arithmetic from above: if your stop were one median day, how much would you hold per $1 of risk? It is an illustration of the math, not a suggestion.
| Binance market | Median daily move | Days over 5% | Position per $1 of risk |
|---|---|---|---|
| BTCUSDT | 1.38% | 9.24% | $73 |
| ETHUSDT | 1.95% | 16.72% | $51 |
| SOLUSDT | 2.91% | 27.04% | $34 |
| XRPUSDT | 1.96% | 18.08% | $51 |
| DOGEUSDT | 2.33% | 21.20% | $43 |
| ADAUSDT | 2.41% | 21.73% | $41 |
| AVAXUSDT | 2.88% | 27.47% | $35 |
| LINKUSDT | 2.79% | 25.74% | $36 |
| BNBUSDT | 1.57% | 13.75% | $64 |
| TRXUSDT | 1.28% | 11.10% | $78 |
Histories start between September 2019 and September 2020 and run to October 19, 2026. Bitcoin has 2,598 daily moves in this history, and the coin histories differ in length.
Solana's typical day was more than twice Bitcoin's. A Solana position sized the same way as Bitcoin's would carry about twice the daily swing.
Does the same stop mean the same thing for every coin?
No. We took every hourly close and looked at the next 24 hours. We counted how often the lowest price in that window reached a set distance below the starting close. Each window starts at an hourly close, so neighboring windows overlap and are not independent.
For a fixed 2% distance, the answer depends on the coin. For a distance in ATR units, it barely does. ATR, or average true range, is the average size of a bar's high-to-low range, including gaps from the last close. Here it is the average of the last 14 hourly ranges, known at the starting close.
| Binance market | Median ATR (% of price) | 2% below, within 24h | 1 ATR below | 3 ATR below |
|---|---|---|---|---|
| BTCUSDT | 0.73% | 39.29% | 73.57% | 34.62% |
| ETHUSDT | 0.98% | 49.99% | 73.10% | 34.97% |
| SOLUSDT | 1.40% | 64.12% | 74.65% | 34.89% |
| DOGEUSDT | 1.22% | 59.94% | 73.98% | 34.93% |
| TRXUSDT | 0.64% | 35.82% | 71.70% | 33.95% |
The fixed 2% column ranges from 35.82% to 64.12%. The ATR columns stay within about three points of each other. A stop set in the coin's own units was reached about equally often on every coin. That is the reason to size by volatility. For more on how far stops sat from price, see our stop-loss distance study.
It is also a warning. A one-ATR distance was reached in roughly 7 of 10 windows in 24 hours. A stop that tight would be hit often.
Does a coin's calm level stay the same?
No. Here is the median daily move by year for Bitcoin and Dogecoin, with the same illustration: position per $1 of risk if the stop equals one median day.
| Year | BTC median day | BTC position per $1 | DOGE median day | DOGE position per $1 |
|---|---|---|---|---|
| 2021 | 2.35% | $43 | 3.59% | $28 |
| 2022 | 1.47% | $68 | 2.60% | $39 |
| 2023 | 1.00% | $100 | 1.64% | $61 |
| 2024 | 1.37% | $73 | 2.68% | $37 |
| 2025 | 1.15% | $87 | 2.81% | $36 |
| 2026 | 1.17% | $86 | 1.59% | $63 |
2026 runs through October 19. Bitcoin's median day in 2021 was more than double 2023's. A size worked out from the calm of 2023 would have been more than double the size implied by 2021. So a sizing rule needs fresh volatility numbers, not one number for all time. Calm and busy stretches tend to come in runs, which we look at in volatility clustering.
What happens on the wild days?
A typical day tells you little about the worst one. Bitcoin's biggest down day in this history was March 12, 2020, at -39.98%. That is about 29 times its median day. Bitcoin's largest 1% of days moved at least 10.58%, and 1.31% of days moved more than 10%.
Dogecoin's wild days were wilder. Its largest 1% of days moved at least 22.44%, and its biggest up day, January 28, 2021, was +389.81%. More on that in Dogecoin volatility in numbers.
Sizing from the median day describes ordinary conditions. It does not describe days like these. For how deep the worst falls went over a whole year, see what is max drawdown.
What do these numbers not show?
- Direction. All daily moves here ignore direction. They say how much a coin moves, not where it goes.
- Real exits. The 24-hour check uses hourly highs and lows. It cannot tell the order of the high and the low inside an hour, and it ignores fees, funding and exchange rules.
- Independent cases. Overlapping windows are strongly dependent, so the number of truly independent cases is far smaller than the count of windows.
- The future. Volatility changes. These are past cases.
The coins are large ones that still trade, so they are a selected group.
How we measured this
We used completed Binance USDT perpetual bars. The daily numbers use close-to-close moves, with each day labeled by its UTC opening date. The 24-hour numbers start at each hourly close from September 8, 2019 and use the next 24 hourly bars. The ATR is the simple average of the last 14 hourly true ranges known at the start. We did not subtract any trading costs.
Try this in Comparand
Ask your connected AI agent: "Get daily bars for Binance BTCUSDT and DOGEUSDT perps for the last two years. Show each coin's median absolute daily move and the share of days over 5%."
- venue_bars returns open, high, low, close and volume for completed bars. Use the 1d bar size. It is free on every key.
- Your agent does the math. The tool returns prices, not volatility statistics.
For Hyperliquid, ask: "Show realized volatility and the volatility regime for BTC, ETH and SOL on 2026-09-01, and how they move together."
- hl_stats returns realized volatility, a volatility regime and a correlation matrix for up to 20 coins on one day. It costs $0.01 per call on pay-as-you-go.
Every answer says which data it came from and what the call cost. See the docs to connect your agent, and try 3 coins free. To see how closely two coins' daily moves line up, read our Bitcoin and Ethereum correlation study.
The short version
- Position size equals the money you accept losing, divided by the stop distance as a fraction.
- Dogecoin's median day was 2.33% and Bitcoin's was 1.38%, so the same risk gives a smaller Dogecoin position.
- A fixed 2% drop was reached within 24 hours in 39.29% of Bitcoin windows and 59.94% of Dogecoin windows. A stop in ATR units was reached about equally often on all coins.
- Bitcoin's median day was 2.35% in 2021 and 1.00% in 2023, so volatility numbers go stale.
- Typical-day numbers do not cover the worst day, which was -39.98% for Bitcoin in March 2020.
Questions people ask
What is position sizing by volatility?
It means holding less of a coin that moves more, so one bad move costs about the same amount. The arithmetic is position size equals the money you accept losing, divided by the stop distance as a fraction. This is a math idea, not advice.
How big is a normal daily move for Bitcoin and Dogecoin?
On Binance perpetual contracts from 2019 and 2020 to October 2026, the median daily move, ignoring direction, was 1.38% for Bitcoin and 2.33% for Dogecoin. Half of all days were smaller than these numbers and half were bigger.
Why is a fixed percent stop unfair across coins?
The same distance means different things for different coins. A 2% drop from the starting price was reached within 24 hours in 39.29% of hourly Bitcoin windows and 59.94% of Dogecoin windows.
What is ATR?
ATR, or average true range, is the average size of recent price bars, including gaps between them. In our test it was the average of the last 14 hourly ranges. Stops set in ATR units were reached about equally often across coins.
Does volatility stay the same over time?
No. Bitcoin's median daily move was 2.35% in 2021 and 1.00% in 2023. A position size based on one year can be far off in another.