The model in one sentence
Choose a clear external liquidity target, wait for price to raid liquidity against that direction, require displacement through meaningful structure, then enter a qualified fair value gap retracement with predefined risk.
The sequence matters. The model does not begin because an FVG appears. Direction, destination, time, liquidity and displacement must already form a coherent narrative before the gap can become an entry location.
- One directional target
- One New York morning session
- One sweep-displacement-FVG pattern
- One predefined risk and exit plan
- No obligation to trade every day
Build the market map before looking for an entry
Start with destination. A bullish map means the clearest unfinished objective is above price, such as an old swing high, equal highs, external buy-side liquidity or an unfinished higher-timeframe imbalance. A bearish map uses the equivalent objective below price.
If no objective is clear enough to justify risk, the bias is neutral. Neutral is not indecision; it is a valid trading position. Do not manufacture direction simply because the execution timeframe is moving.
A practical hierarchy is: external swing liquidity first, a higher-timeframe imbalance second, and the nearest logical waypoint only for partial profit or progress tracking. The final target should represent the larger narrative rather than the nearest minor high or low.
- Bullish: unfinished objective above price
- Bearish: unfinished objective below price
- Neutral: no objective is clear enough
- Keep the bias while its target and higher-timeframe narrative remain valid
- Flip only after opposite M15 structure, a respected parent FVG and a new external target
Use time as a filter, not a signal
The baseline model studies one New York morning. Use America/New_York time so daylight-saving changes are handled correctly rather than hard-coding a permanent broker-server offset.
Observe the 08:30 New York catalyst window first. If no complete setup appears, the 09:30 cash open is the second opportunity. A completed 08:30 trade disables another opening-window entry, and unfilled orders are cancelled by 11:30 New York.
Clock time only opens the observation window. It never replaces the price pattern. At 08:30 or 09:30, no target plus no raid plus no displacement still means no trade. The baseline also disables new entries on Nonfarm Payroll Friday so an exceptional release is not mixed into the ordinary sample.
- Primary observation: 08:30 New York
- Second opportunity: 09:30 New York cash open
- Cancel late pending orders by 11:30 New York
- Maximum one completed morning trade
- NFP Friday disabled in the baseline model
Define the liquidity raid correctly
The raid occurs against the intended trade direction. For a long, price first takes sell-side liquidity while the larger target remains above. For a short, price first takes buy-side liquidity while the larger target remains below.
A sweep is not every wick beyond a tiny lower-timeframe fractal. It should remove a pool that mattered to the current map: a session low or high, equal highs or lows, or a visible swing that traders could reasonably use as protection.
The raid creates the possibility of reversal or repricing; it does not confirm it. Entry remains forbidden until price shows force in the intended direction and breaks the selected reference structure.
- Long setup: sell-side raid against a bullish target
- Short setup: buy-side raid against a bearish target
- Mark the liquidity pool before it is taken
- Reject random wicks that have no connection to the market map
Demand real displacement and a meaningful MSS
Before the session, select one meaningful M15 bellwether swing tied to the external objective. This prevents the execution chart from promoting every M1 or M2 fluctuation into a market structure shift.
After the raid, require an energetic body close through that M15 reference. The breaking leg should reprice decisively and normally leave imbalance. A wick through the level, a slow drift or a break of an insignificant micro-swing does not qualify.
Traders use labels such as market structure shift, MSS or CHoCH differently. In this model the operational definition matters more than the label: opposite-side liquidity was taken, then a meaningful M15 swing broke with displacement in the target direction.
- Reference swing is selected before the break
- The swing must matter to the current narrative
- A body close is stronger evidence than a wick
- The displacement leg must be the leg that creates the entry FVG
Qualify the fair value gap
A three-candle bullish FVG exists when Candle 1 high is below Candle 3 low. A bearish FVG exists when Candle 1 low is above Candle 3 high. Candle 3 must close before the gap is treated as confirmed.
Geometry alone is not enough. The FVG must sit inside the structure-breaking displacement leg and in the correct half of that leg's dealing range: discount for a long and premium for a short.
Measure the actual impulse that broke structure, not the entire trading day or an arbitrary distant swing. Invalidate the setup when no gap forms, the gap belongs to another leg, it is in the wrong half of the range, or price decisively closes through the far side of the gap.
- Bullish FVG: Candle 1 high < Candle 3 low
- Bearish FVG: Candle 1 low > Candle 3 high
- Wait for Candle 3 to close
- Long entries require discount; shorts require premium
- Most visible gaps should be ignored
Execute the long and short patterns
Long sequence: target above price, sell-side liquidity raid, bullish displacement through the M15 reference, confirmed bullish FVG in discount, retracement entry, stop below the protected low and exit at the chosen buy-side objective.
Short sequence: target below price, buy-side liquidity raid, bearish displacement through the M15 reference, confirmed bearish FVG in premium, retracement entry, stop above the protected high and exit at the chosen sell-side objective.
Place the entry at the tested edge or level of the qualified FVG according to the rule being researched. If price never retraces, the model records a missed trade and moves on. Chasing the displacement candle changes both the entry logic and the tested reward-to-risk profile.
- Entry is a retracement, not a chase
- Stop belongs beyond protected structure
- Target is selected before the order
- A missed trade is preferable to an untested entry
Define entry, stop and target as one decision
Wait for Candle 3 to confirm the FVG, then define the pending entry, hard stop and liquidity target together. The stop comes from structure; the position size comes from the maximum money risk allowed for that stop distance.
Reject the trade if the target has already been reached, the remaining reward is insufficient, the stop is too wide for the broker's minimum position size, or the order would fill after the session cutoff.
A minimum reward-to-risk requirement such as 2R is a research parameter, not a universal truth. Test the threshold against missed trades, fill rate, spread, commission and slippage rather than choosing it because it sounds disciplined.
- Entry: qualified FVG retracement
- Stop: beyond the protected swing
- Target: predefined external liquidity
- Reject any setup that cannot fit the account's real executable risk ceiling
Protect capital mechanically
Risk budget equals account equity multiplied by the selected risk percentage. Position size is then calculated from the stop distance, tick or pip value and realistic trading costs. Round down to a broker-valid volume; if minimum volume exceeds the risk budget, skip the trade.
The supplied baseline begins at 0.50% risk, reduces to 0.25% after one loss and stops for the day after two losses. This is a research default, not a recommendation for every account.
The mechanical management hypothesis reduces risk when price reaches 50% of the path to target by moving the stop 25% of its original distance toward entry. At 75% target progress it moves to breakeven; at the target it closes the remainder. These rules need comparison against a fixed-stop version because early protection can also remove eventual winners.
- Never widen the hard stop
- Never increase size after entry
- Never add risk to recover a loss
- Stop trading when the daily rule is reached
- Stress-test spread, commission, slippage and missed limit fills
Respect the no-trade list
The model's edge is partly defined by what it refuses. A missing condition is not a lower-quality version of the same setup; it is a different trade and should not enter the baseline sample.
Keep disqualifications explicit so hindsight cannot convert a weak setup into a winner after the chart develops. In both manual journaling and an EA, record which gate failed and return the process to neutral or finish the session.
- No clear external target
- No meaningful liquidity sweep
- No break of the chosen M15 reference
- No energetic displacement
- No confirmed and correctly located FVG
- NFP Friday, rollover distortion or late-session fill
- Risk exceeds the budget or the daily limit is hit
- Target already taken or entry retracement missed
Translate discretion into an EA state machine
Automation requires the narrative to become a sequence of explicit states: PREP builds bias and levels; ARMED waits inside the enabled session; SWEPT confirms the opposing liquidity raid; DISPLACED confirms the M15 break; FVG READY validates Candle 3 and location; ORDER LIVE manages the pending order; MANAGING controls the open position; DONE prevents another entry that day.
Each state needs an invalidation path. A target can be taken before entry, the session can expire, an FVG can fail, spread can become unacceptable or minimum volume can exceed risk. The safest state transition is often back to neutral rather than forward to execution.
This state-machine discipline also improves discretionary trading. A journal can record the last completed state, showing whether a loss came from a valid model occurrence or from skipping a required gate.
- PREP → ARMED → SWEPT → DISPLACED
- FVG READY → ORDER LIVE → MANAGING → DONE
- One job and one failure path for every state
- No failed gate is silently treated as neutral evidence
Validate the model before risking capital
Use clean data with correct contract specifications, realistic spreads and America/New_York session conversion. Include commission, slippage and the possibility that a limit order is touched in historical data but not filled live.
Test multiple years and volatility regimes, then reserve unseen periods for out-of-sample evaluation. Report trade count, expectancy, profit factor, maximum drawdown, losing streak, time-of-day and the concentration of results—not only net profit.
Ablate optional filters one at a time so complexity has to earn its place. Forward-test on paper or demo before live use. The purpose of validation is to discover where the model fails, not to prove that it cannot fail.
- Data: contracts, rolls, time zones and session history
- Costs: spread, commission, slippage and missed fills
- Sample: multiple regimes plus out-of-sample periods
- Metrics: expectancy, drawdown, streaks and trade count
- Ablation: test every optional filter separately
- Forward test before committing live capital