Do breakouts work in crypto? 20-day highs tested
On 10 Binance perps, 39.83% of daily 20-day-high breakouts fell back in 3 days. Ten days on, the average gain was 6.26%, but the middle case was 1.53%.
On 10 Binance perps, price 10 days after a daily close above its 20-day high was up 6.26% on average, against 2.47% for all days. But the middle case gained only 1.53%, price closed back below the old high within 3 days after 39.83% of breakouts, and the average was 3.06% in 2023 to 2026 against 10.67% in 2019 to 2022.
These are past cases, not a forecast.
For whether high volume made breakouts better, see Does volume confirm a crypto breakout?.
What is a breakout, in plain words?
A bar is the open, high, low and close price over a set time. A daily bar covers one day. A 4-hour bar covers four hours.
An up breakout is a bar that closes above the highest high of the previous 20 bars. A down breakout is a close below the lowest low of the previous 20 bars. Some traders expect price to keep going once it passes a level it had not passed in 20 bars.
A false break is when price closes back across the old level within the next 3 bars. We counted how often that happened.
How often did breakouts fall back?
We tested the ten largest Binance USDT perps. A perp is a futures contract that never expires. The data run from September 8, 2019 to October 19, 2026.
| Breakout type | Cases | Closed back across the old level within 3 bars |
|---|---|---|
| Daily, up | 1,411 | 39.83% |
| Daily, down | 784 | 48.60% |
| 4-hour, up | 7,462 | 43.81% |
| 4-hour, down | 5,391 | 47.88% |
Daily up breakouts failed about four times in ten, and 4-hour ones a little more often. Down breakouts failed about as often as they held.
What happened 10 bars after a breakout?
We measured the price move from the breakout close to the close 10 bars later. We compared it with the same move from every bar of the same coin and timeframe.
Breakouts come in clusters, so counting every one overstates the evidence. We skip new breakouts while an earlier one's 10-bar window is still open. These are the separate cases.
| Breakout type | Separate cases | Average move 10 bars later | Average, all bars | Ended higher |
|---|---|---|---|---|
| Daily, up | 594 | +6.26% | +2.47% | 54.71% |
| Daily, down | 442 | +1.03% | +2.47% | 57.24% |
| 4-hour, up | 3,407 | +0.77% | +0.38% | 47.67% |
| 4-hour, down | 2,853 | +0.35% | +0.38% | 54.57% |
Read the daily up row carefully. The average (6.26%) is far above the middle case (1.53%), because a few huge runs pull it up. Without DOGE, the average was 5.43%. The 594 cases fell on only 383 different days, because coins often break out together. The standard error, a measure of how much an average can wobble by chance, was 1.28 points.
The 4-hour row is weaker. The average was 0.77%, but the middle case was -0.19% and fewer than half ended higher.
After daily down breakouts, price on average went up, the opposite of "the break keeps going".
Did it change over the years?
We split the daily up breakouts at January 1, 2023.
| Period | Separate cases | Average move 10 bars later | Average, all days | Ended higher |
|---|---|---|---|---|
| 2019 to 2022 | 250 | +10.67% | +3.93% | 58.80% |
| 2023 to 2026 | 344 | +3.06% | +1.40% | 51.74% |
Breakouts still beat the all-days average in 2023 to 2026 (3.06% against 1.40%). But the gap shrank, and only 51.74% ended higher. The earlier period includes strong trend phases, so it is hard to say how much was the breakout and how much was the market.
Did it work for every coin?
Not for all ten. This table shows daily up breakouts. The last two columns compare the middle case 10 days later with the middle of all days.
| Coin | Daily breakouts | Fell back within 3 days | Separate cases | Middle case 10 days later | Middle day, all days |
|---|---|---|---|---|---|
| BTC | 163 | 39.26% | 74 | 1.67% | 0.69% |
| ETH | 175 | 37.14% | 73 | 1.45% | 1.01% |
| SOL | 148 | 31.76% | 56 | 4.43% | 0.36% |
| XRP | 107 | 39.25% | 48 | -1.64% | -1.00% |
| DOGE | 86 | 40.70% | 39 | 3.94% | -1.00% |
| ADA | 129 | 41.86% | 54 | -0.30% | -0.63% |
| AVAX | 120 | 31.67% | 48 | 1.26% | -1.38% |
| LINK | 155 | 49.03% | 71 | -1.05% | 0.44% |
| BNB | 160 | 38.75% | 64 | 1.94% | 0.91% |
| TRX | 168 | 47.02% | 67 | 1.75% | 1.17% |
Eight of the ten coins had a better middle case after a breakout than on an ordinary day. XRP and LINK did not. Each coin has only 39 to 74 separate cases, so one row can swing a lot. Counting every breakout instead of only separate cases moves XRP's middle case from -1.64% to +2.45%.
Does high volume help?
We called a breakout "high volume" when its volume was more than 1.5 times the median volume of the previous 20 bars. For daily up breakouts, the high-volume group averaged +8.31% after 10 days (337 separate cases). The others averaged +3.58% (257 cases).
Counting every breakout, the middle cases were 2.47% and 1.06% (844 and 567 breakouts). It is a hint, not a proven rule: the groups are small and both averages lean on a few big runs. We look closer in does volume confirm breakouts?.
What if you had followed every breakout?
We also ran a plain rule: go long (gain if price rises) at the breakout close, exit 10 bars later. For down breakouts, go short (gain if price falls). Each side pays a 0.05% taker fee, or 0.10% round trip.
| Rule | Trades | Average result after fees |
|---|---|---|
| Long after daily up breakout | 594 | +6.16% |
| Long after 4-hour up breakout | 3,407 | +0.67% |
| Short after daily down breakout | 442 | -1.13% |
| Short after 4-hour down breakout | 2,853 | -0.45% |
The short side lost in both timeframes. The long side was positive on average, as the averages above suggest. That does not make the rule safe. Funding payments and slippage are not included.
What do these numbers not show?
- They do not say what happens next. They describe past cases.
- The coins are ten large survivors on one exchange. Coins that faded away are not here.
- We tested one rule, a 20-bar level. Other lengths, stops and exits were not tested. Testing many versions on the same history makes a lucky one easy to find, as we showed with 1,000 random rules.
- Good runs in 2020 to 2021 and 2023 to 2024 favour breakout-following in this history.
How we measured this
We used Binance USDT perp bars for BTC, ETH, SOL, XRP, DOGE, ADA, AVAX, LINK, BNB and TRX, from September 8, 2019 to October 19, 2026. Daily bars use the UTC day. 4-hour bars were built from hourly bars, only when all four hours existed. The 20 bars before a breakout had to be complete. Moves start at the breakout close. The all-bars comparison uses every eligible bar of the same coin and timeframe. Fees are 0.05% per side. For ways to test a pattern without fooling yourself, see how to backtest a crypto chart pattern.
Try this in Comparand
Ask your connected AI agent: "Get daily bars for Binance BTCUSDT since 2024. Mark each day that closed above the highest high of the 20 days before it, and show the close 10 days later."
- venue_bars returns open, high, low, close and volume for completed bars. You can ask for 1d or 4h. It is free on every key.
- Your agent finds the breakouts. The tool returns bars, not a report.
Then ask: "Find the 10 closest past charts to BTC's latest 20 daily bars on Binance."
- shape_analogs with venue binance, symbol BTCUSDT, timeframe 1d and window 20 matches BTC only with its own past on Binance, one match per episode.
- With window 20, each match shows its move 5, 10 and 20 bars later, in bp. 100 bp is 1%. These are the horizons we tested here.
- It ranks charts by shape; it does not detect breakouts. It costs $0.01 per search from credits and gives no average or hit rate.
You can try 3 coins free. See the tool docs to connect an agent.
The short version
- Daily breakouts fell back within 3 days after 39.83% of 1,411 cases.
- Ten days later the average gain was 6.26%, but the middle case was 1.53%.
- The average fell from 10.67% (2019 to 2022) to 3.06% (2023 to 2026).
- 4-hour breakouts were weak: middle case -0.19%, and 47.67% ended higher.
- Shorting breakdowns lost 1.13% per trade after fees on daily bars.
Questions people ask
Do breakouts work in crypto?
Partly, and not evenly. After daily closes above a 20-day high on 10 Binance perps, the average price move 10 days later was 6.26% (594 separate cases), against 2.47% for all days. The middle case was 1.53%, and the average was 3.06% in 2023 to 2026 against 10.67% in 2019 to 2022.
How often is a crypto breakout a false breakout?
We called it false when the price closed back below the old high within 3 bars. That happened after 39.83% of 1,411 daily breakouts and 43.81% of 7,462 four-hour breakouts, on 10 Binance perps from September 2019 to October 2026. For breakdowns below a 20-bar low, the figures were 48.60% daily and 47.88% four-hour.
Does high volume make a crypto breakout more reliable?
In our daily test, breakouts on volume above 1.5 times the recent median averaged +8.31% ten days later (337 separate cases), against +3.58% (257 cases) for the others. Those groups are small and the averages are pulled up by a few big runs, so we treat the gap as a lead to check, not a rule.
Do 4-hour breakouts work as well as daily ones?
Not in our test. After 3,407 separate four-hour breakouts, the average move 10 bars later was 0.77% against 0.38% for all bars, but the middle case was -0.19% and only 47.67% ended higher. We do not read that as breakouts usually continuing.
Do breakdowns below a 20-day low keep falling?
Not on average. After 442 separate daily closes below a 20-day low, the average price move 10 days later was +1.03%, and 57.24% ended higher. A test that sold short after each breakdown lost 1.13% per trade after fees.