What Is ICT? Liquidity Sweeps, FVG, OTE and Real Tests on Gold
Common ICT terms made simple, including liquidity sweep, MSS, FVG, OTE and Silver Bullet, plus backtest results for 3 setups on gold M1 over almost 5 years.
- ICT holds that price often runs to sweep the Stop losses resting above old highs or below old lows before it reverses
- An FVG is a price gap made by 3 candles. OTE is the 62–79% pullback zone of a swing
- We tested 3 setups on gold M1 from 2022 to September 2026. None of them performed well consistently every year
- Always read win rate together with Risk:Reward. A 25% win rate at RR 1:3 only breaks even
- Profit that comes from a handful of trades, or from a single year, is not yet proof that a method works
ICT stands for Inner Circle Trader, a chart-reading approach taught by Michael J. Huddleston. It has become one of the most talked-about topics among newer traders. Its core idea is that price often runs to take out most traders’ Stop losses first, and only then heads in its real direction. This article explains the main terms in plain language, then looks at what happened when we turned ICT rules into numbers and tested them on almost 5 years of gold data.
ICT terms you need to know
- Liquidity: in ICT language, this means orders waiting in the market, especially Stop losses above old highs and below old lows. Sellers tend to put their Stop loss above a High, and buyers tend to put theirs below a Low
- Liquidity sweep: price pushes just past an old High or Low, takes out the Stop losses resting there, then comes back into the old range. Many people call it a stop hunt
- MSS (Market Structure Shift): after a sweep, price breaks the most recent minor swing point in the opposite direction. For example, it sweeps a High and then closes below the latest minor Low. This is the first sign the direction may be changing
- FVG (Fair Value Gap): a gap made by 3 candles in a row. In an up move, the Low of candle 3 sits above the High of candle 1, and the space where they don’t overlap is the FVG. ICT holds that price often comes back to fill this gap, so traders place Limit orders there
- OTE (Optimal Trade Entry): the 62–79% pullback zone of a swing, measured with the Fibonacci tool. The popular midpoint is 70.5%
- PDH / PDL: the Previous Day High and Low. ICT sees these as places where a lot of liquidity rests
- Silver Bullet: the 10:00–11:00 New York time window (21:00–22:00 Thai time, UTC+7, during US daylight saving time, or 22:00–23:00 during US winter time). Traders use it to look for an FVG after price sweeps a High or Low formed earlier
What we tested
We turned 3 setups into fully numeric rules and ran them in the MT5 Strategy Tester.
- PDH/PDL sweep followed by MSS: enter in the direction of the reversal
- Silver Bullet: price sweeps the High/Low of 08:30–09:59 New York time (19:30–20:59 or 20:30–21:59 Thai time), then a Limit order is placed at an FVG within the 10:00–11:00 New York hour (21:00–22:00 or 22:00–23:00 Thai time)
- FVG inside the OTE zone: wait for price to pull back into the 61.8–78.6 zone, then enter at an FVG inside it. This is compared with a benchmark that simply places a Limit at 70.5 and ignores FVGs
Conditions: gold (XAUUSD) M1, 0.10 lot per trade, starting balance 10,000 dollars. 2022–2025 used 1-minute price data, and January to September 2026 used real tick data. Only the spread was counted. Commission was not included.
All results are backtest results. Past results do not predict future results. The figures below are not returns you should expect.
Setup 1, PDH/PDL sweep + MSS: lots of trades, no edge
This setup fired about 20 trades a month, the most of the three. But across almost 5 years, the Profit factor (total profit divided by total loss, above 1 means profit) was 0.94–1.01 for every Take profit variation. In other words, roughly break-even.
The clearest tell is the win rate compared with the RR. Let’s work out the break-even point. Say you risk 100 dollars per trade at RR 1:3 (a win makes 300, a loss costs 100) and take 100 trades.
- 25 wins make 25 × 300 = 7,500 dollars
- 75 losses cost 75 × 100 = 7,500 dollars
- Net = 0
The formula is break-even win rate = 1 ÷ (1 + RR). At RR 1:2 that is 1 ÷ 3 = 33.3%. At RR 1:3 it is 1 ÷ 4 = 25%.
| Take profit | Win rate needed to break even | Actual win rate | Profit factor |
|---|---|---|---|
| RR 1:2 | 33.3% | 31.2% | 0.96 |
| RR 1:3 | 25.0% | 24.6% | 1.01 |
Both actual win rates sit slightly below break-even. That is what a method with no edge looks like: the result is pure noise. In 2026, with real tick data, the RR 1:3 version lost 1,614 dollars, and along the way the balance fell by as much as 6,857 dollars, or 45.7%.
Setup 2, Silver Bullet: one great year
The best overall version was RR 1:3 with entry at the middle of the FVG. Over almost 5 years it made 4,031 dollars with a Profit factor of 1.23. It looks good, until you split it by year.
It lost every year from 2022 to 2025. All of the profit came from 2026 alone. And across every version tested in 2022–2024 (6 versions × 3 years = 18 cells), 17 cells lost money.
Even within 2026 the profit was concentrated. The two biggest winning trades made 1,969 and 1,585 dollars.
- Together = 1,969 + 1,585 = 3,554 dollars
- That is 3,554 ÷ 5,156 = 69% of the whole year’s profit
- Remove those two and you are left with 5,156 − 3,554 = 1,602 dollars from 109 trades
A multi-year total that looks great may come from one year or a few trades. Always split results by year and look at the biggest trades.
This matches what we have seen in other tests on gold M1: price behaviour in 2022–2024 was clearly different from 2025–2026. A method that works in the recent period may simply fit the current market rather than have a lasting edge.
Setup 3, FVG inside the OTE zone: FVG helps a bit, but too few trades
The best version put the Stop loss at the start of the swing (level 100) and took profit at 38.2. Over almost 5 years it made 2,566 dollars, with a Profit factor of 1.19 and a 54.4% win rate.
Compared with the benchmark that ignores FVGs, the FVG versions won all 4 pairs on the overall total. But split by year, FVG lost every pair in 2022 and 2023, then won every pair in 2024–2026. Once again, the result depends on the period.
The bigger problem is the number of trades. In 2022–2025 there were only about 1–4 trades a month, so each year had just 15–48 trades, too few to conclude anything with confidence. Holding for a target at the end of the original swing (level 0) lost money in almost every year and every version, because price rarely made it back there.
Summary of all three setups
| Setup | Trades per month | Almost-5-year total (best version) | Notes |
|---|---|---|---|
| PDH/PDL sweep + MSS | ~20 | +926, PF 1.01 | Win rate equals break-even |
| Silver Bullet | ~12 | +4,031, PF 1.23 | Profit came from 2026 alone |
| FVG inside OTE zone | ~3 | +2,566, PF 1.19 | Few trades, result depends on period |
The limits of this test: it is our own interpretation of ICT as fixed rules, tested on one market and one timeframe. Traders who use ICT with discretion, or define the terms differently, may get different results.
Terms that sound precise do not make a trading method precise. You have to measure it with rules written down in advance.
Common mistakes
- Looking only at win rate. A 30% win rate can be profitable or losing, depending on the RR. Always compare it with the break-even point, 1 ÷ (1 + RR)
- Marking sweeps and FVGs after seeing the price. On a past chart there are enough Highs, Lows and gaps to explain every reversal. If you didn’t note it beforehand, don’t count it as a signal
- Trusting the total without splitting by year. A method that only worked in one period will look good in the total, until the market changes
- Ignoring Drawdown. ICT puts the Stop loss beyond the sweep point, which on gold can be several dollars away. Work out your lot size from the real Stop loss distance, rather than using the same lot on every trade
Frequently asked questions
Does this mean ICT doesn't work?
Not necessarily. We tested only 3 setups, on gold M1, using rules we turned into numbers ourselves. Every ICT trader interprets it differently, and many add their own judgement. This result only says these fixed rules show no evidence of a consistent edge over the market.
What time is the Silver Bullet in Thai time?
The Silver Bullet is 10:00–11:00 New York time. During US daylight saving time (roughly mid-March to early November) that is 21:00–22:00 Thai time (UTC+7). For the rest of the year it is 22:00–23:00.
I want to try ICT. Where should I start?
First write clear rules for what counts as a sweep and what counts as an FVG. Then log every signal in advance, including the ones you miss, split the results by year, and check whether the profit is concentrated in a few trades, all before risking real money.
This is general education, not investment advice. Trading is risky and you may lose all of your capital.