Heikin Ashi Candles

The formula

For each bar you start with the real open, high, low and close. The Heikin Ashi candle is then built so that it leans on the candle before it:

The first bar has no earlier candle, so the HA open is usually set to the average of its real open and close. Heikin Ashi is Japanese for "average bar".

A worked example

Take bar 4 of the made-up path: open 101.1, high 102.6, low 101, close 102.4. The HA close is (101.1 + 102.6 + 101 + 102.4) / 4 = 101.78. The HA open is the average of the previous HA open 100.78 and previous HA close 101.45, which is 101.12. The HA high is the highest of 102.6, 101.12 and 101.78, which is 102.60. The HA low is the lowest of 101, 101.12 and 101.78, which is 101.00. All prices here are made up.

Notice that bar 3 closed lower than it opened, a down candle in real terms. In Heikin Ashi it is still an up candle, because its average close sits above its smoothed open. That is the smoothing at work.

Down candles show up 5 times in the real path and 1 time in the Heikin Ashi path, in a market that mostly went up. That is why the chart looks cleaner. The only Heikin Ashi down candle is the first bar, which has no earlier candle to lean on. A rule that exits on a colour change would have stayed in this whole path, but it would also have stayed in a path that had really turned, for the same reason.

What the smoothing costs

• Delay. Each Heikin Ashi bar leans on the one before, so it needs a real move that outweighs the old average before it changes colour. In a turn the signal arrives after a larger part of the move has gone.
• Not real prices. A signal can say a candle closed at 101.78 when the market closed at 102.4. Orders, stops and the stop loss must be set from real prices, or the trade is measured against a level that never existed.
• It hides gaps. A real gap between bars can be smoothed away, and a stop placed on the smoothed chart may be inside a real gap.

How a bot misuses it

A common rule is to hold while the candles are one colour and leave when the colour changes. In a clean trend that keeps you in. In a range it flips often, and each flip pays the spread and sometimes slippage. A backtest on Heikin Ashi prices can also fill trades at prices that did not trade, which makes the result look better than it was. That is the same trap as the perfect fills in a backtest, and it is easy to do by accident when a platform draws the chart one way and the test runs on the other.

How to check a rule that uses it

Make sure every fill in the test uses real prices, even if the signal comes from Heikin Ashi. Test it in a trend and in a range, count the trades and use full costs. Keep data back and test once, to guard against overfitting. Look at the max drawdown as well as the total.

What this page does not tell you

It does not say that Heikin Ashi improves results. It shows the arithmetic and the costs of the smoothing. The prices are made up and it is education, not a signal.