An order block is the last candle of opposite direction just before an impulsive move. The idea is that a large position was built there before price left, and that the level therefore keeps some capacity to react if price comes back.
The last opposing candle before the move
Contents
The order block comes from ICT vocabulary and rests on an assumption: before an important move leaves, somebody must have accumulated a position, and did so in the direction opposite to that move. The last candle going that opposite way is therefore assumed to contain that accumulation, and it is the one you mark.
It is worth saying plainly what that assumption is worth: it is not verifiable. A candlestick chart does not show who bought, and nothing in the marked candle proves a position was built there. What can be stated without error is more modest: price often reacts when it comes back to these levels, because a lot of people watch them and place orders there.
That nuance changes how you use it. An order block is not a trace of an institutional player, it is a shared reference level, and that is already reason enough to mark it. Our lesson on order blocks and breaker blocks covers the full method.
Bullish order block = the last bearish candle before an upward move that breaks structure
The bearish version is symmetrical: the last bullish candle before a downward move that breaks structure.
The range used varies between methods: the whole candle including wicks, or only its body. The exact half has a name, the mean threshold, and serves as an intermediate marker.
The 'breaks structure' condition is not a stylistic detail, it is everything that separates an order block from an ordinary candle. Without it, every chart contains hundreds.
| What you observe | Valid or invalid order block |
|---|---|
| The move breaks structure | Valid. This is the fundamental condition, and the only one that is not negotiable. |
| The move is just a big candle | Invalid. With no break, nothing separates that candle from its neighbours. |
| Several opposing candles in a row | Keep the last one. The earlier ones are not order blocks. |
| Price returns and holds above the mean threshold | The reaction is clean. This is the most readable case. |
| Price goes all the way through the order block | The level failed. It can then be read the other way round, under the name breaker block. |
The question of what makes an order block valid or invalid has only one solid answer, and it is the first row of the table: the move that follows has to break structure. Everything else, bodies or wicks, one candle or two, is preference and changes little. That condition, on the other hand, separates a level worth marking from a candle picked at random.
The second point is the timeframe, and it is underrated. A daily order block describes a level the market may take weeks to revisit; a 5-minute order block is wiped out within the session. Both carry the same name and have nothing in common: mixing them in one list of levels produces an unreadable chart.
The order block is the 1.0940-1.0960 band. A return that holds above 1.0950 does not read like a return that goes all the way through: that is what the mean threshold settles, and it is the only intermediate level there is any consensus on.
This is the main use. The order block is spotted at the moment of the displacement, so well before price comes back, and it gives two precise edges. That leaves time to prepare a decision calmly rather than improvising it when price arrives.
It is also what makes it compatible with a resting order: a level known in advance can be worked without sitting at the screen, which no reading decided in real time allows.
A chart often offers several plausible return areas. The order block's timeframe lets you rank them: a daily level outranks an intraday one, and the two are read together rather than against each other.
That hierarchy is the part of the method that transmits best, because it requires no aesthetic judgment. It can be checked, and it can be written into a plan.
The first flaw is the story told around it. Presenting the order block as the trace of an institutional order gives the level an authority it does not have, and that authority costs: you hold a losing position longer when you believe a bank is defending the level. The same level, presented as a shared marker, can be abandoned without drama when it does not hold.
The second is the missing validity condition in most presentations. Without the requirement of a structure break, every chart contains hundreds of order blocks, and there is always one just below current price. An object available everywhere cannot serve as a decision criterion.
The third is mixing timeframes, already mentioned, and it is made worse by indicators that mark order blocks automatically: they display everything they find on the displayed timeframe, with no hierarchy, and turn a method into decoration.
The classic mistake. Believing a bank is defending the level. It is a convenient and unverifiable story, and it is expensive: you hold a losing position far longer when you think an institutional player is on your side. The same level, read as a shared marker, can be dropped without drama as soon as it fails.
| Indicator | Formula | What it measures | When to use it |
|---|---|---|---|
| Order block | The last opposing candle before the displacement | A shared reference level | Preparing a return in advance |
| Fair value gap | 3 candles, the 1st and the 3rd do not overlap | A price range that barely traded | Locating a return, or aiming at a fill |
| Breaker block | An order block traded through, then retested | The level's failure, and its role swap | Confirming a reversal |
| Mitigation block | An order block revisited with no new extreme | A position being corrected rather than opened | Telling two kinds of return apart |
It is the only condition there is consensus on and the only one that changes how many levels sit on your chart. Applied strictly, it takes a list of dozens of order blocks down to two or three, and those two or three then deserve watching.
A daily level and a 5-minute level do not compare. Decide which timeframe commands your plan, mark your order blocks on it, and use lower timeframes only for execution, never to add levels.
That is information which costs nothing to log and reveals itself after thirty or so cases: if your winners hold the half and your losers go through it, you have just found a numeric invalidation criterion, on your own data.
It is the last candle of opposite direction just before an impulsive move. A bullish order block is the last bearish candle before a rise; the bearish one is its mirror. Its price range is marked as a level to watch if price comes back to it.
Only one condition has consensus: the move following the candle has to break structure. If it is just a big candle with no break, the level is nothing special. The other criteria, body or wicks, one candle or two, are preference and change little.
The higher timeframe always wins. A weekly or daily order block describes a level the market may take weeks to revisit, whereas a 5-minute one is often wiped out within the session. Both carry the same name without the same reach: they do not belong in the same list.
The order block is a candle where the market traded a lot; the fair value gap is a hole where it barely traded, and the two often come from the same move. The mitigation block is an order block revisited with no new extreme, so a position being corrected rather than opened. All three often overlap, and the level is only more widely watched for it.
Both exist and neither imposes itself. The whole range, wicks included, gives a wider area, so a further stop and fewer premature exits. The body alone gives a tighter level and more invalidated trades. The only real advice is to choose once, write it down, and measure the result over thirty or so trades rather than switching after every loss.
The order block is a reading, and Tradoshi measures what it produces for you. Tag the trades taken on an order block, and the Performance by tag block in the Journal screen isolates their count, their win rate, their average R and their net P&L from the rest of your activity. Keep two tags, one per timeframe, and the hierarchy described above stops being a conviction and becomes a number.
The Reports screen adds MAE, the maximum unrealised loss endured before closing. On trades taken at a precise level, that is the measure that says whether your stops sit in the right place relative to the order block, or whether you are being taken out just before the reaction you were waiting for.
To go deeper on this: read the full guide. Every term is gathered in the glossary.