4 / 7 · 7 min
Imbalance: a place where almost no trading happened
A fast move leaves a band on the chart that price skipped without two-sided trading. Price often returns there — but "often" is not "always", and that matters more than the pattern itself.
When price leaves very fast, part of the way is covered with almost no trades: there were not enough orders on the other side and the market jumped over a whole band of prices. That band is an imbalance. The mechanism is simple and honest: where no trading happened, unfilled intentions remain — buyers who wanted in cheaper and sellers who wanted out higher. Until that band has been travelled at a calm pace, the business counts as unfinished.
How to find it on a chart
Look at three consecutive candles. If the high of the FIRST is below the low of the THIRD, a band is left between them that the middle candle flew straight through. That is an upward gap. Downward is the mirror: the low of the first above the high of the third. Nothing else needs looking for — the construction is entirely mechanical, and that is its virtue: you cannot argue with it, it is either there or it is not.
The gap in numbers
Three hourly candles. The first has a high of 3,218, the third a low of 3,246. Between them lies a band of 3,218 … 3,246, twenty-eight points wide — price skipped it in one hour. The instrument's ATR is 34, so the gap takes up 0.8 of an ordinary hourly range: that is a noticeable hole, not a pixel. A day later price returns to 3,240 — the upper edge of the gap. From there continuation upward is common, because those who wanted to buy cheaper finally get their chance. The stop for such an entry goes below the LOWER edge of the gap, 3,218: if the band has been travelled in full, the reason for the entry is gone.
Without a threshold a chart has hundreds of gaps: on the minute scale they appear in almost every burst. Half an ordinary candle range is a sensible floor: anything smaller the market closes within the hour and without your help. The threshold is taken from the ATR of the scale you work on, not from absolute points — otherwise a rule tuned on bitcoin will not survive being moved to a three-cent coin.
Assuming a gap must close
The most dangerous belief in this subject, because it pushes you to trade AGAINST the move. A gap is the trace of a fast move, and a fast move usually means a new reason appeared: news, liquidations, a large entry. Price returns to a gap often, but sometimes it does not return for months and sometimes it never returns. The claim "price always closes gaps" cannot be refuted — you can always say "not yet" — and by that sign you can tell it is useless for decisions.
Trading a gap against the higher scale
A downward gap inside a daily uptrend is not an invitation to short but a place price will most likely return to and continue upward from. The construction on its own gives no direction: it points at a PLACE, and direction comes from the higher scale. The mistake looks convincing precisely because gaps genuinely do get worked — only they get worked as an entry in the direction of the trend, not as a reversal.
What to do with a partly filled gap
Most often price enters the band halfway and turns, never reaching the far edge. That is normal and even expected: the intentions were spread across the whole band and part of it was enough. Practical consequence: plan the entry from the near edge or the middle, not the far edge, and put the stop beyond the far one. Waiting for a full fill means regularly missing a move that started without you.
Gaps are where bar replay helps most. On history you see the gaps that closed and do not notice the ones that stayed open: the eye picks out confirmation. In replay the chart moves candle by candle, the future is not fed even into the indicators, and an honest count is possible only that way.
The chart and bar replayThirty gaps and an honest count
Find thirty gaps at least half an ATR wide on the hourly scale. For each, record three things: did it close within a week, did price enter at least halfway, and did the gap's direction agree with the daily trend. Then count the share that closed, separately for gaps with the trend and against it. If there is no difference, direction does not matter and the rule can be simplified. If there is a difference, you have found a filter worth keeping.
A gap has been open for three weeks and price has moved far away. Is it worth waiting for a return?
As an event, perhaps; as grounds for a trade, no. The longer a gap stays open the less of it is left: the intentions that were there have gone stale, and the participants who wanted in cheaper have settled the question another way over three weeks. A gap is the trace of a recent event, and its force decays with time. The mark can stay on the chart, but building an entry on it three weeks later means trading a memory.
Why are there far more gaps on a minute chart than on a daily one?
Because on a small scale any burst leaves a hole between candles: within a minute there are physically fewer orders on the other side, and jumping a band is easier. On the daily scale a gap of the same significance means the market skipped a price it could not close for a whole day — an event of an entirely different weight. Hence the threshold in fractions of ATR: it adapts to the scale by itself and stops you counting minute noise as a discovery.