Heikin Ashi Strategy Backtest: 9 Years of Bitcoin on 4-Hour Candles

00The verdict in three lines

• We tested the Heikin Ashi strategy on Bitcoin 4-hour candles from October 2017 to September 2026: always following the colour, the classic no-wick entry with an exit at the first opposite colour, and the same with a 200-period moving average filter. Real prices for every fill, 0.05% a fill.
• The filtered no-wick version is one of the strongest results on this site: +0.243% a trade after costs, x24.8 compound against x19.1 for buy and hold, in the market 39.3% of the time, better than random timing in all 500 runs, five gates of six.
• Still buried. Its profit factor from mid-2021 to 2023 was 1.14, under the 1.2 we require, and the average trade shrank in every period. Following the colour alone lost to costs.

01What is Heikin Ashi?

Heikin Ashi (Japanese for “average bar”) is a way of drawing candles that averages each bar with the one before it, so that a trend shows as a run of candles of one colour with few interruptions. Each Heikin Ashi candle is computed from the real candle:

• HA close = (open + high + low + close) / 4 of the real candle.
• HA open = (previous HA open + previous HA close) / 2.
• HA high = the highest of the real high, HA open and HA close; HA low = the lowest of the real low, HA open and HA close.
• Green when HA close is above HA open, red when below. A green candle with no lower wick (HA low equal to HA open) is read as a strong up candle; a red one with no upper wick as a strong down candle.

Because they are averages, Heikin Ashi prices are not prices you can trade at. Every fill in this test uses the real Bitcoin price at the open of the next 4-hour candle.

02The Heikin Ashi trading strategies we tested

• HA-FLIP: always in the market: long when the candles turn green, short when they turn red, reversing on every colour change.
• HA-NOWICK: buy on a green candle with no lower wick, sell short on a red candle with no upper wick; exit at the first candle of the opposite colour, then wait for the next no-wick candle.
• HA-NOWICK-200: HA-NOWICK, but longs only when the real close is above its 200-bar simple moving average (about 33 days of 4-hour candles) and shorts only below it.
• All three: the signal is read when a 4-hour candle closes (00:00, 04:00 … 20:00 UTC) and executed at the next candle's open; no stop or target other than the colour change.

These are the three versions we found most often in Heikin Ashi tutorials. We fixed them, the timeframe and the costs on 21 October 2026, before computing a single candle, and tried nothing else afterwards.

03How we tested it

Binance BTCUSDT 1-minute history from August 2017 to 30 September 2026, built into 4-hour candles on the UTC clock (19,702 candles from 1 October 2017, when trading starts; candles with fewer than 120 minutes of data are skipped). Costs: 0.05% of the price on every fill, the taker fee on Binance perpetual futures, so about 0.1% a round trip; a reversal pays two fills. Funding is reported separately. Periods: October 2017 to June 2021, July 2021 to 2023, and a holdout from 2024 already used by four earlier Bitcoin tests on this site. First-period bar t ≥ 2.0 (an indicator not tested before here). Control: each version's own positions, shifted as a whole to other days by a random number of days, 500 times: the same number of trades, holding times and long/short mix, with the timing broken. Engine check: hand-worked candles (the formulas, the wick tests, a short candle, a reversal, the moving average, the end of the data) and an independent re-implementation that matched every one of the 9,533 trades; it passed on its first run (ha_selftest_result.txt).

04Results: does the Heikin Ashi strategy work?

Growth of 1 unit fully invested in each trade, by month, log scale; dashed: buy and hold from 1 October 2017. * Holdout already used before this test.

Heikin Ashi win rate, holding time and risk

Heikin Ashi strategies win rarely and win big: about one trade in three makes money, and the winners carry the rest. On these 4-hour candles the colour changed every 3.9 candles on average, and 41.3% of candles had no wick on the entry side, so the no-wick signal is far from rare.

05Heikin Ashi vs buy and hold on Bitcoin

Buy and hold from 1 October 2017 to the end of September 2026 turned 1 unit into x19.1, with a fall of 84% from peak to trough on hourly closes along the way. The filtered Heikin Ashi strategy turned 1 unit into x24.8 while holding a position 39.3% of the time, and its largest fall on closed trades was 34%. The two falls are not measured the same way (the strategy's ignores losses while a trade is open, which last less than a day on average), so read them as an order of magnitude, not a precise comparison. Most of the gain came from long trades (+360.6% as a sum of trades, against +44.4% for shorts).

06Heikin Ashi or random timing?

The control keeps each version's own positions and moves them to other days. Placed at random times, the same number of trades lose about the cost of each round trip; the real Heikin Ashi timing did better in every version. For HA-FLIP that was not enough to cover its costs; for the two no-wick versions it was, with every one of the 500 shifted runs below the real result.

07By period and by year

The filtered version gained in every calendar year except 2025, but its average trade fell from +0.448% in the first period to +0.114% and then +0.074%. The unfiltered no-wick version was strong to 2021 and roughly flat afterwards.

08The gates

The data-gap check removed 30, 18 and 12 trades open over a gap in the data; no verdict changed. HA-FLIP is Costs (positive before costs, negative after). The two no-wick versions are Overfit under our protocol: they passed the first-period test and failed the validation profit factor, and the unfiltered one also lost in the holdout. HA-NOWICK-200 missed only that one gate. That is not a pass, and we do not move the bar after seeing a result; it is one of the strongest buried results on this site, and the right test for it is new data.

Forward test (added 22 Oct 2026): HA-NOWICK-200 is now being tested forward, with its rules, start date (26 October 2026) and evaluation frozen in advance: a paper series computed by this engine on new Binance data and reported every month. Protocol and journal.

TWhat this shows, and what to take into your own tests

What the results show

• The textbook Heikin Ashi entry, a candle with no wick on the entry side, filtered by a 200-bar moving average, made +0.243% a trade after costs on 1,670 Bitcoin trades from October 2017 to September 2026 and grew 1 unit to x24.8, against x19.1 for buy and hold, while in the market 39.3% of the time.
• It beat the same holdings shifted to random days in all 500 runs and passed five of our six gates. It failed one: the validation profit factor was 1.14 against the 1.2 we require.
• Simply following the colour (HA-FLIP) did not work: it changed side every 3.9 candles on average, and its edge before costs (+0.082% a trade) was smaller than the round-trip cost of about 0.1%.

What remains unknown

• Whether it keeps working. The average trade fell in each period: +0.448%, then +0.114%, then +0.074%, and the most recent period had already been used by four earlier Bitcoin tests, so it is not a clean confirmation.
• Other markets, other timeframes and other filters (an EMA instead of the simple average, a different length). We tested one market and one timeframe, with three versions fixed in advance.
• Live funding on perpetual futures. With 10% a year charged on every position the average trade was +0.221%; real funding varies, and was much higher in some months.

Why it was buried anyway

Our protocol asks for every gate, not most of them. The filtered version made +315.9% (sum of trades) in its first 705 trades to mid-2021, with a t-statistic of 2.93, but the next two and a half years earned a profit factor of only 1.14, under our 1.2. A result that weakens after the development period is the classic sign that part of it was luck, which is why the cause is recorded as Overfit even though the rules were fixed in advance and not tuned.

What you can apply to your own tests

• Use real prices for fills. Heikin Ashi candles are averages; an entry at the Heikin Ashi open is a price that never traded.
• Count the cost of every colour change. Always-in Heikin Ashi trading changes side every few candles, and 0.1% a round trip was enough to sink it.
• Judge a near miss on new data. The honest next step for the filtered version is a forward test, not a tweak that makes the validation period pass.

A note on the numbers: drawdown figures are measured on closed trades unless a table says it includes open losses. Open losses inside a trade are not counted in a closed-trade drawdown, so real account risk is higher, especially for systems with no stop. Backtests are not live trading.

FAQCommon questions

What is Heikin Ashi?

A way of drawing candles that averages each bar with the previous one: HA close is the average of the real open, high, low and close; HA open is the midpoint of the previous HA candle. Trends show as runs of one colour, but the prices are averages, so trades must be filled at real prices.

Does the Heikin Ashi strategy work?

The best version in our Bitcoin test made +0.243% a trade after 0.1% round-trip costs and grew 1 unit to x24.8 from October 2017 to September 2026 (buy and hold: x19.1), better than random timing in all 500 control runs. Its average trade shrank in each period and it failed the validation profit factor gate, so we could not call it proven.

What is the best Heikin Ashi strategy?

Of the three we tested, the no-wick entry (a green candle with no lower wick to buy, a red one with no upper wick to sell short) filtered by a 200-period moving average, with the exit at the first candle of the opposite colour. Always-in trading on every colour change was the worst, because it traded too often for its costs.

What is the win rate of a Heikin Ashi strategy?

About one trade in three: 34.1% for the filtered no-wick version, 32.4% without the filter and 31.0% for always following the colour. The profitable versions made their money from a minority of large winners.

Is Heikin Ashi good for Bitcoin?

On 4-hour candles its timing was better than chance in our test, but costs matter: the always-in version, which reversed more than 500 times a year, lost money, and the filtered version earned most of its gains on long trades.

How was the strategy tested?

On Binance BTCUSDT 1-minute data built into 4-hour candles, October 2017 to September 2026, with real prices for every fill, 0.05% a fill and rules written and dated before the data was run, costs included on every trade, results split into development, validation and holdout periods, a random-entry control as the benchmark, and every variant reported. Read the protocol.

09Check our arithmetic

Code provenance: our own code, written from the specification and the standard Heikin Ashi formulas.

10Data notes

Source. Binance BTCUSDT spot 1-minute history (data.binance.vision), the file of our earlier Bitcoin tests; the costs are those of the perpetual futures, where shorts are possible, whose prices stay close to spot. A 4-hour candle with fewer than 120 minutes (exchange outages) is skipped, and the Heikin Ashi chain continues from the last full candle. Fills: the real open of the first minute after the signal candle closes.

11Corrections

22 Oct 2026, addition: a paragraph in section 08 linking the forward test of HA-NOWICK-200. No result changed. 22 Oct 2026, navigation: added a link to the new guide comparing trading costs across all our tests. No result changed. 22 Oct 2026, navigation: added a link to the new guide comparing trend following and mean reversion across all our tests. No result changed.

+More tests and guides

• Turtle trading bitcoin backtest
• Bitcoin pullback strategy backtest
• Noise area breakout bitcoin
• Supertrend indicator backtest
• Trend following vs mean reversion: this test next to the others of its style
• Free prop firm challenge calculator
• What all the backtests found so far
• The backtesting protocol
• How to backtest a trading strategy without fooling yourself
• Backtesting glossary