Non-Time Charts: What Breaks When the X-Axis Is Not Time

Most charts put time on the horizontal axis. A one-hour bar covers an hour whether the market moved or not, so a flat afternoon produces as many bars as a violent one.

Another family discards that entirely. A new bar appears only when price has travelled far enough, so the axis measures movement instead of duration.

The construction is well documented. What is rarely followed through is what it does to everything applied on top.

What a Non-Time Chart Actually Changes

The three chart shapes most traders start with are the line, the bar and the candlestick, and our page on the three time-based chart shapes covers those. All of them share one property: the horizontal axis is time, and each bar occupies a fixed slice of it.

Renko, range bars, point and figure and Kagi charts all drop that property. The horizontal axis becomes a sequence of completed price movements, and the spacing between bars carries no duration information.

The usual argument for them is noise reduction, and it is a fair one. A quiet session that would fill a time chart with small indecisive bars may produce no new bars at all. What that argument leaves out is the cost: every tool built on the assumption of a regular time axis is now running on an irregular one, and most do not announce it.

How Each Construction Decides to Print a Bar

The family shares a principle and differs in the rule: each defines a distance, and a new bar appears when price covers it.

A Renko chart is built from bricks of a set size. StockCharts describes bricks as not being added unless price changes by a specified amount, with the brick size set either as a fixed value or from Average True Range. Our page on the Renko chart in detail covers that construction on its own terms.

A range chart works from a fixed high-to-low distance. The thinkorswim documentation describes a new bar being plotted after a specified price range has been accumulated, and states that each new bar opens at the previous bar’s close, which coincides with either the high or the low of that same bar depending on direction.

Point and figure and Kagi belong to the same family and add a reversal amount, deciding how far price must go the other way before the chart changes direction rather than extending.

Why an Indicator Period Stops Meaning a Period

Set a 14-period indicator on a daily chart and it looks back over fourteen days. The number in the settings box has a unit attached, and that unit is time.

Put the same indicator on a range chart and the unit changes underneath it. It now looks back over the last fourteen completed price excursions, which might have taken four minutes during a fast move or two days during a quiet stretch.

Nothing in the interface signals this. The setting still reads 14 and the line still plots, so a lookback quietly loses any consistent meaning: two readings a week apart can measure windows of wildly different duration, and comparing them is not the like-for-like comparison it appears to be.

It also breaks the assumption behind aligning several charts. Our page on aligning timeframes depends on a higher timeframe containing a known number of lower ones, and a non-time chart has no timeframe to place in that hierarchy.

The Range Bar Result That Turns ATR Into a Setting

The clearest demonstration is Average True Range, most often used to size a stop.

True Range, devised by J. Welles Wilder in his 1978 book New Concepts in Technical Trading Systems, is the greatest of three quantities: the current high less the current low, the absolute value of the current high less the previous close, and the absolute value of the current low less the previous close. The last two exist to capture gaps.

Apply that to a standard range chart. Every completed bar has the same high-to-low distance by construction, so the first term is the configured range on every bar.

The gap terms collapse too. Because each bar opens at the previous bar’s close, and that open coincides with either the high or the low, the previous close is either the current high or the current low. One gap term therefore equals the full range and the other equals zero.

The greatest of the three is the configured range, on every bar. True Range is constant, so Average True Range returns the chart setting rather than any measurement of volatility.

The qualification belongs with the claim. This follows from the standard construction in which bars open at the previous close and no gaps intervene, so implementations that permit gaps at a session open will produce exceptions.

The point is not that ATR is broken everywhere. It is that on this chart type the number it returns is largely a restatement of a setting the trader chose.

The Bar That Has Not Closed, and Why It Can Redraw

A time bar closes when the clock says so, and once closed it is fixed. A non-time bar closes only when price has travelled the required distance, so the bar being watched stays unresolved for as long as that takes. It can keep extending, and the completion being waited on may never arrive.

Some implementations also redraw the forming bar as price moves, so what is on screen now is not necessarily what will be there when it settles, and an indicator computed on that bar moves with it.

This is not the same phenomenon as an indicator revising its own history, which our page on indicators that repaint covers, but the hazard is identical. A reading taken before the bar resolves is provisional.

The claim usually made for these charts is therefore backwards in one respect. They are cleaner once settled and less stable while forming, because a time chart at least tells you when the current bar will stop changing.

Why a Backtest on These Charts Flatters the Strategy

A Renko brick or a range bar is printed once price has already covered the distance. The level recorded as that bar’s close is defined by the construction, and is not necessarily a price that was available to trade when the bar appeared.

A naive test fills at the bar close anyway. Every entry and exit is placed at a constructed level rather than a traded one, and small differences across many trades can turn a losing rule into a profitable-looking one. These constructions also sit outside bar replay, which the platform documents as unsupported on Renko and range charts.

This is the same artefact that affects other constructed series, and our page on why a simulated result overstates execution sets out the wider reasons a simulation reads better than reality. What is specific here is that the chart type introduces the artefact before any of those apply.

What Happens to Volume and Session Boundaries

Volume becomes the only remaining proxy for participation, because bar width no longer carries any. On a time chart two volume bars covered the same interval and are comparable; here two bars of equal volume may have taken very different lengths of time to fill.

Session boundaries also stop lining up, because a bar can begin in one session and complete in the next. Anything anchored to them is affected: a daily open, a session range, a release at a fixed hour, or a financing charge accruing by calendar date all reference a clock the chart has stopped tracking.

When These Charts Help, and Who Should Not Use Them

The genuine case for them is narrow and real. Where the question is whether price has travelled far enough to matter, and the time taken is irrelevant to it, a construction that measures distance answers it more directly than one measuring duration.

They are the wrong tool whenever anything in the method is time-anchored: trading around scheduled releases, session-based approaches, strategies holding positions across financing dates, and any method that aligns several timeframes.

They are equally a poor choice where indicator settings need to keep their usual meaning, since a stop sized from ATR or a lookback chosen as a number of days does not survive the change of axis.

This page names no setting, brick size or reversal amount as correct, because the right value depends on the instrument and the question, and any number offered here would be invented rather than sourced.

Frequently Asked Questions