What Is an Immediate Rebalance? And Which Rule You Have
What is an immediate rebalance? The opposite outcome to a fair value gap. The four published rules, and how often two of them fire on our EUR/USD bars.
An immediate rebalance is what happens when a fast move does not leave a gap behind: the very next candle trades straight back through the boundary the move just made, filling it on the spot. It is the opposite outcome to a fair value gap, which is the same moment leaving something unfilled.
The catch is that there is no single published rule for it. LuxAlgo states it one way in its reference pages and a different, incompatible way in the source its own script uses, and one of them is satisfied on almost every bar on a daily chart.
What does an immediate rebalance look like?
A strong candle pushes out of a range, and instead of leaving an untraded pocket behind it, the following candle reaches back into the area just vacated and then carries on in the original direction.
The reasoning attached to it is subtractive, and it is worth separating from the geometry. Traders who mark unfilled imbalances as places price might return to can cross an immediately rebalanced level off that list, because the auction there was already repaired. The move reads as efficient rather than overextended.
That is the claim, and LuxAlgo, the vendor making it, is careful to label it: continuation "remains a taught tendency rather than a measured law, and deserves verification on the instrument you actually trade". We publish no measurement of it either. A continuation rate is a performance claim.
Immediate rebalance vs fair value gap
Same moment, opposite outcome. A fair value gap is a fast move that leaves a span of prices barely traded, which some traders then watch as a level price may come back to. An immediate rebalance is that same displacement with the retrace already done, in the very next candle.
So if you are keeping a list of open imbalances, a fair value gap goes on the list and an immediate rebalance comes off it, or never gets added, and that bookkeeping is the practical difference between them.
LuxAlgo attributes the concept to Michael J. Huddleston, and it shows up in ICT material alongside the gap rather than instead of it.
What is the exact rule for an immediate rebalance?
There are four published versions, and only one of them is specified tightly enough to code.
The only fully specified version comes from LuxAlgo's published toolkit source, which reads three consecutive candles. Label them A, B and C:
| Bullish | Bearish | |
|---|---|---|
| C's wick | C dips under A's high but stops above A's close | C pokes above A's low but stops below A's close |
| B's close | B has to close above A's high | B has to close below A's low |
| C's close | C has to close above A's high and above B's close | C has to close below A's low and below B's close |
| Boundaries | Every test is a strict inequality, so a tie fails | Every test is a strict inequality, so a tie fails |
Its own worked example: A's high is 100 with a close of 98, B closes at 102, and C dips to 99 before closing at 103. The dip at 99 sits between 98 and 100, both later closes clear 100, and 103 beats 102, so it is detected. And the edge case it spells out: "If C's low were exactly 100, this implementation would not label the same pattern."
The second version is a midpoint convention. LuxAlgo's reference pages call this a two-candle pattern instead: "One candle pair, one wick measurement, one map edit." Their working rule is that "the new candle's wick should retrace at least to the midpoint of the prior candle's wick before the span counts as rebalanced". LuxAlgo's own pages call that midpoint the wick's consequent encroachment. It has no three-candle structure and no closes, just a halfway retrace.
The third asks for a full re-trade, and the same reference page half-concedes the spread in its own FAQ: "Some implementations demand a full re-trade of the boundary area rather than the halfway point."
The fourth is not a definition at all. TradingFinder's indicator page offers an "ImmR Filter" setting and never publishes what level it filters on.
That TradingFinder page describes immediate rebalance zones as "essentially a type of Liquidity Void" and says they help traders "predict potential trend reversal points". Both are backwards. A liquidity void is the unfilled imbalance; an immediate rebalance is the absence of one. And the same page contradicts its own reversal framing a few paragraphs earlier, where it says the zones "do not require a price retracement; instead, they strengthen the prevailing trend". If you have been working from that description, that is probably where the confusion started.
How often does each rule actually fire?
We ran both against our cleaned EUR/USD history, just under a million one-minute bars from the start of 2024 to the end of August 2026.
| Timeframe | Bars | Three-candle rule | Per 1,000 bars | Midpoint reading, share of bars |
|---|---|---|---|---|
| 1 minute | 991,637 | 15,008 | 15.1 | 86.9% |
| 5 minute | 198,328 | 6,570 | 33.1 | 95.6% |
| 15 minute | 66,110 | 2,621 | 39.7 | 97.9% |
| 1 hour | 16,528 | 831 | 50.3 | 99.2% |
| 1 day | 835 | 34 | 40.7 | 98.8% |
The three-candle rule is selective: it marks between one and five bars in a hundred depending on the timeframe. A halfway-retrace reading, taken as a rule on its own, is satisfied on 87% of one-minute bars and on virtually every bar of a daily chart.
A condition that is true 99% of the time is barely a condition: it almost never fails, so it almost never tells you anything. If you read the prose and your friend runs the indicator, you are looking at objects that differ by a factor of twenty to fifty-seven in how often they appear.
Our figures are the geometry only, a count of how often each written condition is true on real bars, with no entries, no outcomes and no rate of anything working.
Is an immediate rebalance bullish or bearish?
Neither on its own. It inherits the direction of the move it sits inside, and the bearish conditions are the exact mirror of the bullish ones, as the table above shows.
A bullish immediate rebalance is an upward displacement whose next candle reaches back down into the prior range and then closes higher still. A bearish one is a downward displacement whose next candle reaches back up and then closes lower. Calling the pattern itself bullish gets it backwards: the displacement carries the direction, and the rebalance only tells you the move left no unfinished business behind it.
Does it work on every timeframe?
The rule fires on all of them, and the shape of that runs against intuition.
Per bar, the three-candle rule is more common on higher timeframes, not lower: about fifteen bars in a thousand on one-minute data against fifty in a thousand on hourly. In absolute volume the picture inverts, because there are so many more minutes. Fifteen thousand events on one-minute bars against eight hundred and thirty-one on hourly, a factor of eighteen.
So minute data gives you a large sample of a rarer event per bar, and hourly data gives you a denser event on far fewer bars. If you want many observations, go down. If you want each observation to matter, go up. Neither is the correct answer, and the published material does not pick one.
The confirmation setting has a hole in it
LuxAlgo's toolkit marks a detected pattern as failed if a later bar breaches A's level. Its Confirmation (Bars) setting "defaults to two and accepts one to five", and it checks that one selected bar after it closes. Its own documentation flags the design: "it is a check on the selected later bar, not a continuous test of every intervening candle."
The hole is measurable. For each confirmation setting, the share of surviving patterns that had already breached A's level on a bar in between, and so would have failed a bar-by-bar reading:
| Confirmation setting | 1 minute | 5 minute | 15 minute | 1 hour |
|---|---|---|---|---|
| 1 bar | 0.0% | 0.0% | 0.0% | 0.0% |
| 2 bars (the default) | 11.1% | 12.9% | 15.2% | 12.8% |
| 3 bars | 21.7% | 24.4% | 27.2% | 23.6% |
| 5 bars | 34.5% | 38.3% | 40.9% | 38.8% |
At one bar the two checks are the same test, which is why that row reads zero. At the shipped default, roughly one surviving marker in nine had already been invalidated. At the top of its own range it is two in five.
If you are counting markers from that indicator, that is the correction to apply, and the fix is to check every bar rather than one.
How do you test an immediate rebalance by hand?
Write down which rule you are testing before you open a chart, because four descriptions are in circulation and they are not interchangeable. Name the candle count, every inequality, whether equality at a boundary passes or fails, and which later-bar check you are using and over how many bars.
Then step forward one bar at a time with the next bar hidden. This rule in particular needs that, because the pattern is not a pattern until C has closed, and knowing what came next makes it impossible to judge whether you would have called it at the time.
Two things not to write into your notes. The indicator also draws lines at 25%, 50% and 75% across the first candle's wick, and LuxAlgo says plainly those percentages "are not success probabilities". And a marker surviving its confirmation bar does not prove a profitable entry or that a stop held; it proves one bar did not breach one level.
ChartLabs does not detect this pattern for you, and for a three-candle rule that is less of a loss than it sounds. Bar-by-bar replay with the next bar hidden is what the test needs, and the drawing tools have keyboard shortcuts for marking A's high and close before B and C exist. Every trade captures a snapshot of your drawings and chart state at entry. The Orders panel shows up to five tags per trade and carries a notes column, and your notes come through in the CSV export, which is enough to record which of the four rules and which confirmation setting each trade used. You can also write either rule as an indicator in the built-in editor, which uses Pine Script v5/v6 syntax. The Pine Scripts tab has an "Example: Mark bars by a rule" that opens a copy in the editor, a reasonable place to start for a candle rule. There is no Pine backtesting, though: the testing stays manual.
Where to go next
Backtesting a Trading Strategy has the general workflow, and Manual vs Automated Backtesting is the honest version of what hand-testing a candle rule does and does not tell you.
Then pick one rule, one timeframe and one confirmation setting, and log every instance for a month of data. ChartLabs is $20 a month or $200 a year, one plan, and the trial is 7 days with no credit card: start the free trial.
Related Topics
Related
Backtesting a Trading Strategy: A Beginner's Guide
What backtesting is, why it matters, and how to run one step by step: defining your rules, choosing clean historical data, stepping through the chart trade by trade, logging results, and reading the metrics that tell you whether your strategy has a real edge.
Manual vs Automated Backtesting: Which One Should You Use?
Automated backtesting tells you whether a set of rules made money. Manual backtesting tells you whether you can trade them. A practical comparison of what each one tests, what it costs you in time and skill, and which to reach for first.