Smart Money Concept Expert Advisor - Market Strucutre, FVG, Pullback - expert for MetaTrader 5

This Expert Advisor trades fair value gaps in the direction of market structure, the way Smart Money Concepts describes it. It is the third step of a series that has been built one piece at a time: In previous shared codes, the SMC Pullback Indicator decided which highs and lows matter, the SMC Market Structure indicator turned those swings into a trend with BoS, CHoCH, IDM and liquidity sweeps, and this EA trades the result.

The structure engine is not reimplemented here. Both classes are included from SmcMarketStructure.mqh, the same code the indicator draws from, so what the EA acts on and what you see on the chart cannot disagree.

The setup

Four things have to line up before a long trade. The short side is the exact mirror and is never described separately below.

Everything in that figure is produced by the EA logic itself, not drawn by hand. The pale line is the pullback swings, the gold line is the structure, and A and B are the two points every later calculation refers to: A is the structure low the leg started from and B is the high the IDM confirmed.

What counts as a fair value gap

A gap is a three-candle pattern. The middle candle travels far enough that the third candle never trades back to the first, leaving a band of price that was skipped:

Finding gaps is easy; the useful part is deciding which ones matter. Three filters apply, and a gap has to pass all of them:

• Position. It must sit between the structure low and the bar that broke the level. Gaps older than the leg belong to a move that is already finished.
• Side. It must lie entirely below the level the BoS broke. A gap above that level is not a discount, it is chasing.
• Size. It must be at least the minimum width you set, in points or pips. This is the only noise filter in the strategy, and on a low timeframe it is the one that matters.

Each surviving gap is drawn as a box labelled FVG, and the box extends to the right with every new candle until price touches it. Once touched it stops growing, which leaves a visible record of how long the gap stood untested.

How deep price has to come back

A gap right under the high is a weak place to buy. Price has barely corrected, the move has not been tested, and the stop has to sit a long way off. The depth filter exists to refuse exactly that.

The leg is A to B. Depth is measured from B back towards A, so 0% is the confirmed high and 100% is the structure low itself. Set the minimum to 50% and price must return at least to the midpoint of the leg before any gap in it may be traded.

In the figure the touch arrives at 73% of the leg, past the 50% threshold, so it is allowed. Had the gap sat in the shaded band near B it would have been refused no matter how clean it looked. Setting the input to 0 turns the filter off.

Entry, stop and target

The touch is the entry, at market. The rest follows from the structure rather than from a fixed distance:

Because the stop is structural, its distance varies from trade to trade — and that is the point. The position size is worked out from the percent of balance you are willing to lose, so a tight stop buys a larger position and a wide one a smaller position, and every trade risks the same money. A fixed lot is available but it is not the default, because it would make a tight stop and a wide stop mean completely different things.

Two guards sit in front of every order: the EA refuses the trade if the stop would land on the wrong side of price, and if the distance is inside the broker's stop level. If the smallest tradable volume would risk more than your percentage allows, it says so in the log rather than quietly risking more.

Managing the position

Once a trade is open, two optional rules can move its stop. Both live in a separate class, CPositionManager, which knows nothing about market structure or gaps — only a symbol and a magic number — so it can be reused in any EA.

They share one rule that is not optional:

Both triggers and distances are read in the same unit as the rest of the EA — points or pips, whichever you selected. The two rules work together: break even typically fires first and parks the stop at entry, then trailing takes over as the move extends. A level the broker would reject for being too close to price is skipped rather than retried.

Standing down after a CHoCH

A CHoCH is the structure admitting the trend has turned. The EA does not treat it as a signal to trade the other way immediately — it clears every gap it was watching and does nothing at all until the first BoS in the new direction confirms the reversal has follow-through. This is the single biggest difference between it and a naive FVG bot, and it is what keeps it out of a range that is chopping back and forth.

A new BoS in the same direction has a similar effect on a smaller scale: the previous gaps are discarded and the scan runs again. Only the current leg is ever traded.

Inputs

Structure engine

These four match the indicator exactly. If you run both, keep them the same or the chart and the EA will be looking at different structures.

Fair value gaps

Trading

Break even and trailing

The display group controls the gap boxes and the structure drawing. Turn the structure off for long optimisation runs — it redraws several hundred objects a bar, which is fine visually and wasteful otherwise.

What to know before running it

• Closed bars only. The structure is evaluated on closed candles; the forming candle is never used. Entries are checked on every tick, because a gap can be touched mid-candle.
• The structure repaints, by design. A pending extreme moves until its IDM is taken, and a liquidity sweep relocates a point backwards. Trades are only taken from confirmed legs, so this does not move a stop after the fact — but the picture before confirmation is provisional and is meant to be.
• It is a continuation strategy. It will sit out reversals entirely, and in a long range it will stand down after the first CHoCH and stay down.
• Trade management is deliberately plain. One entry, one target, and an optional break even and trailing stop. There are no partial closes, no scaling in, no grid and no recovery logic, because the point is to see whether the structure plus gap premise does anything on its own rather than to bury it under management.
• Test it before you believe it. There are no tuned parameters here and no presented backtest. The defaults are reasonable starting points, not recommendations.

Testing

Files

Put all three in the same folder and compile the .mq5; the includes are relative, so there is nothing to install anywhere else.