ICT trading strategy: one complete model, step by step

What ICT is, plainly

ICT stands for Inner Circle Trader, the name under which Michael Huddleston has taught a large set of ideas about market structure, liquidity and timing, mostly through free videos. Smart money concepts (SMC) is the wider community that grew out of those ideas and renamed several of them. The vocabulary is now everywhere: liquidity sweeps, fair value gaps, change of character, order blocks, kill zones, premium and discount.

The framing is that large institutions engineer price moves to trigger retail stop orders and then trade against them. It is worth saying plainly: there is no public evidence that this is what drives these patterns, and a retail chart cannot show who is buying or selling. What ICT does offer is a precise, shared language for things that do happen on charts: stops clustered above obvious highs get triggered, fast moves leave gaps, and certain hours are more active than others. That is enough to build a testable strategy on, without needing to believe the story.

The trouble most people have is volume. ICT has dozens of named models and hundreds of hours of material, and traders end up applying a different concept to every chart until something fits. The fix is to pick one model, define every step so that two people would mark the same chart the same way, and test it. The model below is the one most ICT traders learn first.

The model: bias, sweep, shift, gap

1. Daily bias and the draw on liquidity

Before the session, decide which way you expect the day to go and where price is likely heading. ICT calls the target the "draw on liquidity": usually the previous day's high or low, or an old swing high or low that has not been traded through. A simple mechanical bias: if the daily chart is making lower highs and the previous day closed in its lower half, look only for shorts towards the previous day's low.

2. The Asian range

Mark the high and low of the quiet Asian session, roughly midnight to 06:00 UK time. Stop orders from traders who bought and sold that range collect just beyond both edges, and that is the liquidity the model expects to be taken.

3. The sweep, during a kill zone

In a short setup, wait for price to push above the Asian high during the London kill zone (about 07:00 to 10:00 UK) or the New York one (about 12:00 to 15:00 UK). The push takes out the stops, then fails. On its own this is not a signal; plenty of pushes above a high are real breakouts.

4. The shift

The signal is what comes next: a strong move back down, with large candles, that closes below the last swing low formed before the sweep. That break is the change of character: the short-term structure has turned. The same move usually leaves a bearish fair value gap.

5. The entry in the gap

Sell on the retest of that gap, with the stop above the sweep high.

In the chart above, the Asian session ranges between 1.0838 and 1.0862. At the London open price spikes to 1.0871, above the range, and closes back inside it. The next candle falls through the 1.0846 swing low, leaving a gap between 1.0852 and 1.0858. Price retraces into the gap, a short fills at 1.0855, the stop is at 1.0874 above the sweep, and the target is the previous day's low at 1.0817, twice the risk.

Entry, stop, target and management

Entry. A limit order inside the fair value gap, at its near edge or its midpoint. If the gap is thin, the near edge; if it is wide, the midpoint, so that the stop does not balloon. If price never returns to the gap within the kill zone, the trade is off. Do not chase.

Stop. A few pips beyond the extreme of the sweep, which is the price at which the "false move" turned out to be real. On EUR/USD this is usually 12 to 25 pips on a 15-minute setup; on gold, several dollars. Size the position from that distance with the position size calculator so a loss costs your normal fixed risk.

Target. The draw on liquidity you named before the session, usually the previous day's high or low or the opposite side of the Asian range if it is far enough. Require at least twice the risk. A common habit is to take part of the position off at the Asian range's opposite edge and move the stop to breakeven, then hold the rest for the main target.

Time stop. ICT setups are intraday. If the trade has not reached its target by the end of the New York session, close it. Holding an intraday idea overnight turns it into a different trade with a different risk.

Why the kill zones matter

Kill zones are simply the hours when the most volume arrives: the London open and the New York open, plus the overlap between them. Outside those windows, moves above and below ranges are more often noise from thin markets than real sweeps followed by real shifts. Restricting the model to two or three hours a day is one of its better features, because it cuts the number of marginal trades and keeps you away from the screen the rest of the time.

The windows move with daylight saving, and the UK and US change their clocks on different weekends in March and in October and November, so check the times rather than memorising them. The guide to forex trading sessions lists the hours, and the academy lesson on session timing and kill zones goes into the variations. Avoid the minutes around high-impact data such as US jobs or inflation figures: a release can create a sweep and a shift in seconds, and neither means much.

How the model fails

The commonest loss comes from treating the sweep as the signal. Price pokes above the Asian high, the trader sells straight away because "liquidity has been taken", and the market keeps going because it was a genuine breakout. The rule that protects you is the shift: no break of the last swing low, no trade.

Here GBP/USD pushes above the 1.2712 Asian high to 1.2718 in the London session. A short taken on the sweep looks reasonable for a few candles, but the pullback stops at 1.2706, well above the 1.2698 swing low. There is no change of character, and price runs to 1.2760, through the stop at 1.2724. Waiting for the shift would have kept you flat.

The other weaknesses are less visible. ICT has enough concepts that one of them will explain any move after the fact, which makes the material feel more predictive than it is when you study charts in hindsight. Bias calls are partly discretionary, and a wrong bias makes every later step look right while pointing the wrong way. And the model produces few trades: some weeks there will be one or two good setups per market, and taking the marginal ones is where most of the damage is done. Watching two or three markets (EUR/USD, GBP/USD, gold) gives more chances without lowering the standard.

How to test an ICT strategy honestly

The test is only as good as the definitions. Before you start, write down: the exact Asian range hours, the kill zone hours, what counts as a swing low (for example, a candle with two higher lows either side), and what counts as displacement. Then:

1. Choose one market and the 15-minute chart, and go back at least three months.
2. Hide the chart to the right of the current candle and replay forward, one candle at a time.
3. Log every setup that meets the rules, with entry, stop and target written before the outcome is visible, and include the days when price never returned to the gap.
4. Record results in units of risk after the spread, and collect at least fifty trades.

Expect the number of trades to be small, which is why a test over a few weeks says almost nothing. Track the worst run of losses as well as the average result, and check that run against the drawdown calculator before choosing your risk per trade.