What is a Heikin-Ashi chart, and are the prices on it real?

A Heikin-Ashi chart takes an ordinary bar's open, high, low and close and replaces all four with averages, then draws the result as a candle. Sierra Chart's documentation for the study, read on 4 November 2026, calls the output average price bars, and warns on the same page that a bar treated this way no longer shows real values. That pairing is the whole subject of this post: the smoother candles are what the chart is for, and the prices printed on them are computed numbers that may never have traded.

The mechanics below come from two vendors' documentation of the calculation and from one working paper that back-tested an indicator built on those bars; our own rules come from our help center. The vendors' pages do carry interpretive language — one of them presents the transform as noise filtering that makes trend and reversal points easier to read — and that is the vendor's reading, not a tested result. Nothing here is a reason to take a trade.

What is a Heikin-Ashi chart?

A Heikin-Ashi chart is a candlestick chart drawn from recalculated values instead of the four prices the market actually printed. Quantower's documentation describes it as a modified candlestick chart based on a recalculation of the OHLC data, and in its software a Heikin-Ashi chart is selected from the list of aggregations, with two build parameters its page names Value and Build from. Quantower also notes that the transform is available only with the candle chart style.

Sierra Chart treats it as a study rather than a chart type. By default the Heikin-Ashi study draws in a chart region underneath the main price graph, so you see the averaged bars and the real ones at the same time. A setting named Display As Main Price Graph moves it into the main region and replaces the chart's existing bars, so the real ones come off the screen and only the transform is left.

That distinction matters more than it sounds. The other way a chart can depart from plain time bars is in when a bar begins and ends. Sierra Chart's pivot point help topic says that bars of that sort — renko, range, volume, number of trades and the rest — do not have exact starting times, and its separate topic on their accuracy is about bar period and session settings. Heikin-Ashi leaves the bar boundaries alone and changes the four numbers inside them.

How is each Heikin-Ashi bar calculated?

Each bar's four values come from arithmetic on the current bar and the bar before it. Sierra Chart's study page gives the close as the sum of the bar's open, high, low and close divided by four. The open is the average of the previous Heikin-Ashi bar's own open and close. The high is the larger of the bar's true high and the Heikin-Ashi open, and the low is the smaller of the true low and the Heikin-Ashi open.

The open is where the recursion lives. Because each Heikin-Ashi open is built from the previous Heikin-Ashi bar, the chain runs back to the first bar on the chart, where Sierra Chart seeds it with that bar's actual open.

Two of the sources that print the formula state the high and the low with a third operand. Quantower's version takes the high as the largest of the true high, the Heikin-Ashi open and the Heikin-Ashi close; the working paper discussed below prints the same three-operand form and credits it to a charting website. None of them comments on anyone else's wording, so here is the arithmetic. The Heikin-Ashi close is the mean of four prices that all sit between the bar's low and high, so it can never exceed that high or undercut that low; a third operand that cannot be larger than the high cannot change which value is the largest, and the same holds at the low. The two statements therefore produce the same number from the same bar — worth knowing before you conclude two platforms are computing different things.

Are the prices on a Heikin-Ashi chart real traded prices?

No. Sierra Chart's study page states that every value of a bar is modified by the study, that the bars "no longer display real values", and that bars produced this way cannot be compared with a chart that does not have the study applied. The close you read off a Heikin-Ashi candle is an average of four prices, so in a bar that closed at its high the printed close sits below where the contract last traded.

Put numbers on it. The bar below is invented for the arithmetic rather than taken from the market: an open of 6800.00, a high of 6805.00, a low of 6799.00 and a close of 6804.00. The Heikin-Ashi close is the average of those four, 6802.00, two full index points under the 6804.00 that actually printed. On the E-mini S&P 500 the tick is 0.25 and each tick is $12.50, per our instruments page, so those two points are eight ticks, or $100 a contract. On the micro at $1.25 a tick it is $10. Nothing is wrong with the chart; the number on it is answering a different question from the one your order will be filled at.

There is one documented exception. Sierra Chart's Heikin-Ashi study has a single input, Set Close to Current Price for Last Bar, which defaults to Yes and makes the final bar's last value the symbol's live price. It applies to the final bar only. On the smoothed version of the study, the documentation adds that with the same input switched off the last bar's close stays a calculated figure and is not the true traded price.

How does a Heikin-Ashi chart differ from a standard candlestick chart?

A standard candlestick reports four prices that happened; a Heikin-Ashi candle reports four numbers derived from them. Three differences do real work. The close becomes an average of the whole bar instead of its final trade, so it lands inside the bar's range. The open stops being a traded price at all and becomes a function of the previous candle, which makes the series dependent on where the chart's history starts. And the comparability goes: Sierra Chart's documentation rules out comparing bars carrying the study against a chart without it. Its help topics already list several reasons two ordinary charts of one symbol can disagree — session times, the settlement price, missing data, the data service behind the chart — so the transform adds a difference on top of those.

Why do two Heikin-Ashi charts of the same contract disagree?

Heikin-Ashi values depend on the bars they are computed from, and those bars depend on settings that differ from one trader's chart to the next. Sierra Chart's study page says outright that the last bar's close will differ between charts of the same symbol using the study, depending on the bar timeframe. The recursion explains it: change the length of a bar and you change every input to the chain.

The second reason is that "Heikin-Ashi" names more than one calculation even on a single platform. Sierra Chart's studies contents page carries exactly two study names containing the word — the plain one and a smoothed one — and the smoothed version is a three-stage pipeline. It first smooths the four prices with a moving average over one period input, then applies the Heikin-Ashi transform to the smoothed series, then smooths the result again over a second period input. At each stage the averaging method can be swapped for any of six alternatives the page names, so one named study covers a large family of possible outputs. Its published spreadsheet was produced with the simple average chosen at both stages.

Even the name is unsettled. The study pages and the contents page spell it Heikin-Ashi; one of the same vendor's help topics spells it Heiken Ashi, and Quantower uses both spellings on one page. If you are searching documentation and finding nothing, search the other spelling.

What else on the chart stops working when Heikin-Ashi is the main price graph?

Sierra Chart documents two things breaking when the transform is the main price graph: the accuracy of non-time bars, and pivot point values. Its help topic on inaccurate or inconsistent non-time bars — covering eight bar types, among them renko, range, volume and point and figure bars — opens its checklist by telling you to look at the chart's study list and remove the Heikin-Ashi study if it is there, before any of the bar period or session settings are touched.

The same vendor's help topic on differing pivot point values gives the study a section of its own, under the spelling Heiken Ashi. With the study set as the main price graph, it says, the value of the pivot lines is affected, and changing the timeframe per bar makes those values differ again. The topic's other sections are about session times, the settlement price, missing or insufficient data and the formula type setting, which is a useful reminder that pivot levels are already sensitive to their inputs before any transform is layered on.

Both cases are the same mechanism: a study computed from bar values is computed from whatever the bars now hold, and the transform has rewritten the open, high, low and close. If you are comparing your levels with someone else's and they do not match, the price graph itself is worth checking before the indicator settings.

Has Heikin-Ashi been tested?

The one test cited here is a 2018 MPRA working paper by Smita Roy Trivedi, and it builds an indicator from Heikin-Ashi bars rather than testing the bars alone. No claim is made about what else has been published on the subject; this is the paper that was read for this post. The indicator, which the paper calls HASTOC, measures the gap between a Heikin-Ashi open and close and scales it from 0 to 100 against a ten-day window; a second variant does the same with the wick length. Entry rules fire when the scaled value drops below 5%, and two of the three strategies in its results table add an exit when the value rises above 50%.

The back test runs on one currency pair, USD/INR, sampled hourly across five months of 2018, May through September. Its results table reports 211 signals and an average profit of 0.0306% for the simplest variant, and 0.00305% across 1,318 signals for the version with the exit rule. In each case the standard deviation it reports alongside — 0.063% and 0.064% — is larger than the average, which tells you how much of the variation a single average is hiding. A word search of the paper's text on 4 November 2026 returns no occurrences of transaction, cost, spread, commission or slippage, so nothing in it adjusts those averages for what trading a signal costs.

That is one working paper, on one currency pair, over five months, on an indicator derived from the bars. It is evidence about that object and that period. If you are weighing a rule that was fitted to a short sample, curve fitting is the problem to read about next.

What does this mean on a funded account?

A funded account here is a simulated account that pays out real profits, and it is scored on fills rather than on chart values. Our drawdown line sits a fixed distance under your highest end-of-day balance and is monitored live through the session, so what moves it is the balance produced by the prices you actually traded at. A chart whose printed close is an average of the whole bar rather than its last trade is a reasonable way to look at trend and a poor way to judge how close you are to a line.

Two practical consequences follow. First, read your risk in real prices: work out stop distance, tick value and position size from the ladder or a standard bar chart, which is where position sizing has to be done anyway. Second, expect your platform's trade log and your eyes to disagree. A fill at the bar's last trade against a candle printing the bar's average — the example above — is the transform doing its job, so the log is the record to reconcile against, and the chart is the picture.

If you want the smoothing without losing the real prices, the vendor's default placement is the honest arrangement: leave the study in its own region under the main price graph, so the averaged bars and the traded bars are both on screen.

FAQ

Does a Heikin-Ashi chart repaint?

Neither of Sierra Chart's two pages for this study uses the word repaint, but they document the part that moves. The study has one input, on by default, that sets the final bar's close to the symbol's current price, and it applies to the last bar only; the smoothed study's page says that with that input off, the final close stays a computed number and is not the symbol's last trade. So the bar you are watching carries either the live price or a running average of a bar that has not finished, and both change while it is forming.

Can I place a stop at a Heikin-Ashi price?

Nothing stops you typing the number, but it is an average rather than a level that traded. Sierra Chart's documentation says every value of a chart bar is modified by the study, that such bars "no longer display real values", and that you cannot compare them with a chart that does not have the study applied. Work out stop and target levels from a chart of the traded prices or from the order ladder, then look at the Heikin-Ashi chart for context.

What is the difference between Heikin-Ashi and Heikin-Ashi Smoothed?

Smoothed adds two more averaging passes. Sierra Chart documents the smoothed study as averaging the four prices first, applying the Heikin-Ashi transform to those averaged values, then averaging the output again — and the page names six alternatives to the default averaging method at each of those two stages. Where the plain study has one input, this one has five: two periods, two averaging methods and the last-bar setting. So two charts both labelled "Heikin-Ashi Smoothed" can hold different numbers.

Why did my pivot points change when I switched to Heikin-Ashi?

The pivot study reads the bars beneath it, and Heikin-Ashi rewrote them. Sierra Chart's help topic on pivot points that differ gives the study its own section: with it set to display as the main price graph, the value of the pivot lines is affected, and the line values shift again when the bar timeframe changes. Its topic on non-time bar accuracy tells you to remove the study from the chart's study list; leaving it in its own region below the price graph, where it goes by default, is the other way out.