Indicator library · Bar structure
Balance of Power, Where the Bar Closed
Where the close finished relative to the open, measured against how far the bar travelled. One of the few indicators here that needs the open, and the only one that puts a number on a candlestick reading.
The calculation
BOP = (close − open) ÷ (high − low), computed per bar.
The numerator is what the session actually achieved between its first print and its last. The denominator is how contested it was, the whole distance travelled. Dividing one by the other gives a figure between −1 and +1, where +1 is a bar that opened at its low and closed at its high, and −1 the reverse. A bar that opened and closed at the same price reads zero regardless of how wide it was.
Because the range normalises the result, the reading does not carry the instrument’s price level or volatility with it. That is unusual in this library and it means the thresholds transfer between instruments better than most, a +0.7 on a utility and a +0.7 on a small-cap describe the same kind of session.
What the number captures
Read a candlestick chart by eye and you are already using part of this measure: a long upper shadow with a small body near the bottom is a bar that reached higher and gave it back. Balance of power is that impression, arithmetically.
| Reading | The bar |
|---|---|
| Near +1 | Opened at the bottom of its range and closed at the top. Whatever was offered during the session was absorbed. |
| Near 0, wide range | Travelled a long way and finished where it started, a contested session with no resolution. |
| Near 0, narrow range | Nothing happened. Arithmetically identical to the row above and a completely different market. |
| Near −1 | Opened high, closed at the low. Everything gained during the session was given back. |
The middle two rows are the reason this measure should not be read alone. Dividing by the range deliberately discards the size of the range, so a violent round trip and a dead session produce the same figure. Whatever else is on the chart has to supply that, the range itself, from ATR, or the volume beneath the bar.
The relationship to reading the bar
This page and the volume spread analysis page describe the same observation with different degrees of rigour, and the comparison is instructive.
Volume spread analysis takes the spread, the position of the close and the volume together and weighs them with judgement. That is broader, it uses more of what the bar contains, and it is exactly why the method resists testing: two analysts applying the same rules identify different bars. Balance of power isolates one component, computes it identically every time, and can therefore be plotted, averaged and studied.
Neither is a substitute for the other. The number is what you use to check whether an impression holds up across a hundred sessions; the judgement is what you use when a single bar matters and the volume is part of the story. Using the number and then ignoring the volume is the failure mode unique to this measure, because the formula quietly discarded it.
Where it misleads
| Situation | What goes wrong |
|---|---|
| Narrow bars | A tiny range in the denominator turns a few ticks into a large reading. The measure says a dead session was decisive. |
| Range discarded | A violent round trip and a flat quiet day both read near zero. The figure cannot distinguish them. |
| Unreliable opens | On thin instruments the open can be one unrepresentative early trade, and the whole numerator depends on it. |
| Gaps | The measure works inside the bar only, so a session that gapped five per cent and then drifted reads as ordinary. |
| Volume ignored | A textbook reading on the thinnest session of the month describes an empty book, not a balance of power. |
| Smoothing conventions differ | Some platforms plot a short average of the raw series. Two charts of "balance of power" can be different lines. |
What volume adds
More here than almost anywhere else in this library, because the formula throws volume away explicitly. The name promises a statement about who was in control, and control without size behind it is not control: a close at the top of the range on the lightest session of the month means the sellers were absent, not that the buyers were strong.
In the figure above the two readings move together, the bars close progressively weaker inside their ranges while the volume beneath them thins, and that combination is worth considerably more than either half. Weakening closes on expanding volume would describe a market being actively fought over; the same closes on drying volume describe an advance that has simply run out of participants.
Frequently asked questions
How is balance of power calculated?
For each bar, (close − open) ÷ (high − low). The numerator is what the session achieved between its first and last print; the denominator is how far it travelled. The result runs from −1, meaning the bar opened at its high and closed at its low, to +1 for the reverse. Igor Livshin published the measure in 2001; several platforms smooth it with a short moving average before plotting, which is worth knowing when comparing two charts.
What does it measure that price does not?
Who finished the session in control, relative to how contested it was. A bar can rise a full per cent and still print a low reading if it opened near its high, spent the day well below it and only recovered at the close. Price says the day was up; this says the day was fought over and the buyers barely won it. That distinction is the whole content of the measure.
How is it different from volume spread analysis?
It puts a number on one of the readings that method takes by eye. Volume spread analysis looks at the spread, the position of the close and the volume together and weighs them with judgement; balance of power computes the close-within-range component alone, ignores volume entirely, and produces a value you can plot and test. One is more rigorous and narrower; the other is broader and depends on the analyst.
Does it need the open?
Yes, and that makes it unusual on this site, most measures here use only high, low and close. It also makes the reading sensitive to how your data provider records the open, which historically was the least reliable of the four values in American equity data. On instruments where the open is an auction print the figure is solid; on thinly traded ones it can be a single early trade at an unrepresentative price.
What counts as a strong reading?
Above roughly +0.5 the close landed in the top quarter of the range having started well below it; below −0.5 the reverse. Because the measure is already normalised by the range it does not carry the instrument’s price level or volatility, so those levels transfer between instruments better than most thresholds in this library. What they cannot tell you is whether the bar mattered, which needs the volume.
Should it be smoothed?
For reading, yes, the raw series jumps between extremes on ordinary data because a single bar with a narrow range can produce a large value. A three or five-period average makes the trend of the readings legible. For studying individual sessions, no: the raw value is the whole point, and smoothing removes exactly the single-bar reading a method like volume spread analysis is built on.
What happens on a bar with no range?
The denominator is zero and the value is undefined. It happens on halted or extremely illiquid instruments and on some intraday bars where nothing traded. Implementations differ — some return zero, some carry the previous value forward, some drop the bar — and any of those is defensible as long as it is stated. Code that returns NaN and plots a gap is arguably the most honest of the three.
Does a divergence from price mean anything?
It is the reading the measure is most often used for, and it means the same cautious thing every divergence on this site means. Price making higher highs while the readings weaken says each successive advance is finishing further from the top of its own range; buyers are getting less for the same effort. That is a description of a changing character, it can persist for a long time, and it is worth more when the volume is thinning at the same time.