The previous parts covered what liquidity is, where it sits, how it gets taken and which entry models apply. This part names and orders the execution mechanics — the decisions that turn a correct reading into an executed trade, and the ones that quietly ruin it.
When liquidity is most likely to be taken
Trading activity is not spread evenly through the day. Two things concentrate it: session opens and scheduled news.
Session opens
- The London open — typically when volume and volatility arrive on the major pairs. The Asian range has usually left an obvious level on both sides.
- The New York open — either continues London's direction or trades back through the levels London left behind.
- The London–New York overlap — the deepest participation of the day on forex and gold. Moves tend to travel further; so does the noise.
In thin conditions — mid-Asian hours on non-JPY pairs, the last hour of New York, holiday sessions — there are fewer participants, so breaks more often lack follow-through and the reading is harder to interpret.
Scheduled news
A release is a liquidity event: orders cluster ahead of it, and the print releases them at once.
- Stand aside through the release, then trade the aftermath once a level has been taken and rejected.
- Mark the levels beforehand and treat the print as the trigger — accepting that spreads widen, that fills slip, and that a position held through a release can lose more than the amount you intended to risk.
Entry confirmations
A liquidity take on its own is not a trade. Four things are used as confirmation, in order. Miss one of the first three and there is no setup. The fourth arrives later, and is what Model C waits for.
1. The close back inside
The level was cleared on the wick and the candle closed back inside, judged on the timeframe where the level was marked.
A candle that has not closed can still change shape — what looks like a rejection can close as a continuation. The close is the only version of the level test that cannot be revised afterwards. Waiting for it costs a few points of entry price in exchange for a reading that is reproducible.
2. The rejection candle
A long wick on the side of the take with a small body says price went there and was refused. A large body sitting beyond the level says the opposite.
3. Displacement
A short run of candles moving away, each with a body of roughly two-thirds or more of its high-low range, and the group visibly larger than the twenty or so candles before it. There is no fixed count — what matters is the contrast with what came before. Treat the two-thirds figure as a convention to start from and adjust.
Displacement is read as evidence that absorbed orders are fuelling something. A sweep followed by slow, overlapping candles set nothing in motion, however textbook the wick looked.
Often the run leaves a fair value gap — a three-candle imbalance. It is timeframe-dependent: an M5 imbalance may not exist on M15.
4. Break of structure
After a bullish sweep, a minor lower high is taken out; after a bearish sweep, a minor higher low breaks. It confirms that the reversal has changed the local structure, not just produced one strong candle.
Entry models
Three ways in. They differ in what they wait for, not in quality.
Model A — Return to the origin
Wait for displacement, identify the order block it came from, and place a limit order in that zone.
- Entering closer to the invalidation gives a larger ratio for the same target — that is arithmetic, not a prediction.
- It requires price to return, which it may not do.
Model B — Return to the imbalance
A limit order in the fair value gap left by the displacement, instead of at its origin.
- The gap sits closer to current price, so it is reached more often.
- The stop stays where Model A puts it, so the same target gives a smaller ratio.
Model C — On the break of structure
A market or stop order once structure breaks in the new direction.
- It removes the wait, at the cost of a worse price and a wider stop.
- Workable when the target liquidity is far enough away to leave a usable ratio.
How they map to the confirmations: Models A and B enter on confirmations 1–3, before structure has broken. Model C waits for confirmation 4 and pays for it in price.
Cancel the pending order when the premise expires — the session ends, the swept level is retaken, or the displacement is fully retraced. A limit order left sitting after its reason has gone is no longer part of the plan.
Which charts
- The chart levels are marked on.
- The chart the sweep and the close are judged on — the same one.
- A chart one or two steps down, where the entry is refined.
A common day-trade pairing is H1 for levels and judgement, M5 for refinement; a common swing pairing is D1 and H1. These are starting conventions, not the only valid ones.
Stop placement
The rule
The stop goes beyond the wick of the sweep, not just beyond the level. Placing it one tick under the obvious low puts it inside the very cluster this framework says attracts orders.
Size the buffer so it scales across instruments: the wick extreme, plus the current spread, plus a small fraction of the average candle range on the entry timeframe.
One honest caveat: this placement is widely taught, so the area past a sweep wick is itself an order cluster. A wider stop lowers the frequency of being clipped; it does not put you somewhere nobody is looking.
Where exactly, by model
- Model A: beyond the order block or beyond the sweep wick, whichever is further.
- Model B: beyond the order block or beyond the sweep wick, whichever is further. The gap is the entry, not the invalidation.
- Model C: beyond the swing point that produced the break.
Sizing: the arithmetic
The invalidation level is set by structure. Position size is the variable you control — so it needs a calculation, not a feel.
- Fix in advance the fraction of the account you accept losing on one trade. Any figure used below is an illustration for the arithmetic; choosing your own is your decision.
- Measure the stop distance in points, entry to stop.
- Read the value per point for that instrument and volume from your platform's contract specification.
- Volume = risk amount ÷ (stop distance × value per point).
Worked shape, with round numbers purely to show the mechanics: a 1,000-unit account and 1% accepted risk gives a 10-unit risk amount. A 25-point stop, on an instrument worth 0.10 per point at 0.01 lots, gives 10 ÷ (25 × 0.10) = 4 — so 0.04 lots.
If the structurally correct stop forces a volume below your broker's minimum, the setup is untradable on that account — skip it, or trade a smaller-contract instrument. Do not move the stop.
Do not widen it
A stop widened mid-trade is no longer a stop: it converts the loss the setup was built around into one you have not sized for.
Take profit
The primary target: opposite liquidity
If price took sell-side liquidity to go long, the plausible destination is the buy-side liquidity above: the equal highs, the previous day high, the range high. The reasoning mirrors the entry — that is where resting orders are inferred to concentrate, so it is where the move has a mechanical reason both to travel and to stall.
Intermediate targets
- An imbalance on the way — a fair value gap often produces a pause. A sensible place for a partial.
- Equilibrium — the midpoint of the range — when trading from one extreme.
- An opposing order block — where counterparty was absorbed in the other direction.
Leave a buffer inside the move
Price often turns just short of an obvious pocket rather than tagging it. And on a short, your take-profit fills on the ask — the visible bid candle has to travel about one spread further than your line.
Both argue for placing the exit a little inside the move. Size that buffer scale-free — a fraction of the recent average candle range, or a small multiple of the current spread — rather than by eye. The trade-off is explicit: a nearer target is reached more often but each win is smaller, and whether that helps your results is something to measure, not assume.
Ratio governs selection
Before entering, drag the crosshair from entry to stop, then entry to target, and read the point distances.
Use this article's own rules for both: entry after the close, stop past the wick plus buffer, target short of the pocket — then subtract spread and commission. That number, not the naive level-to-pocket figure, is the one to plan around; it is always the smaller of the two.
The minimum ratio you will accept is a parameter you fix in advance and record. It does not make any outcome likely; it decides which setups you take.
Managing the position
Partial exits
A workable default: take part of the position at the first intermediate target, leave the rest for the opposite liquidity. What matters more than the split is that it is decided before entry.
Early exits are discretionary; the stop is not
The stop is the hard invalidation and the risk you sized for. Everything below closes the trade for less than a full stop, by choice:
- Price trades back beyond the swept level. The premise — that the level held — is gone.
- Displacement fades. Define it observably rather than by feel: successive candles closing inside the prior candle's range with no new extreme over a stated number of bars, or a close back inside the fair value gap.
Discretionary exits change your realised distribution, so log them separately from stop-outs — otherwise you cannot tell which of the two is costing you.
On moving the stop to entry
Factually: a stop at your entry price is not breakeven. You entered on one side of the spread and exit on the other, so it books a small loss of roughly spread plus commission.
Neutrally: moving the stop up removes risk from the position; it also ends trades that would have retraced and then worked. There is no general answer to which matters more — it depends on your instrument, timeframe and risk tolerance, and an article cannot decide it for you.
Mechanically: if you do it, tie it to a defined structural trigger — a new structure point printed beyond the first intermediate target, for instance — rather than to elapsed time or to how the candle feels.
The record
A log makes your own decisions visible, which is worth more than another entry model. Minimum columns: date, instrument, model used, which confirmations were present, entry, stop, target, planned ratio, and result in multiples of the risk taken.
Be careful what you read into it. Over a small number of trades, outcomes are dominated by chance; a short record shows what you did, not what works.
Key takeaways
- Nothing here is observable. A pocket is an inference; a sweep and a break look identical until after the fact.
- Timing: volume arrives at session opens; thin hours and releases make the reading harder to interpret.
- Confirmation: close back inside, rejection, displacement — the first three are the minimum. Break of structure arrives later.
- Entry: origin (limit), imbalance (limit, reached more often, smaller ratio), break of structure (market, no wait, wider stop).
- Stop: beyond the sweep wick plus a scaled buffer — and it is an instruction, not a guarantee. Compute the volume; never shrink the stop.
- Target: the opposite liquidity, with a buffer inside the move, and the ratio measured net of costs.
- Management: the stop is the hard invalidation; everything earlier is discretionary and needs logging separately.