ICT Silver Bullet Strategy: Rules, Times, Example

The Silver Bullet is one of the most searched ICT models, and for a good reason: it's one of the few that comes with a clock. Instead of watching charts all day, you look for one specific sequence inside one specific hour. If it doesn't appear, you're done.

This guide covers the windows, the rules in plain language, a fully worked NQ example with sizing, and how to test the model before you trust it with money.

The three windows

All times are New York time. The model was popularised in ICT's 2023 mentorship videos, and the windows are summarised the same way in most references, including LuxAlgo's concept library:

In winter (EST) add one hour to the UTC times. The New York AM window sits inside the wider New York killzone; see ICT killzones explained for how the sessions fit together and how daylight saving shifts them.

The model in four steps

1. Map the draw on liquidity before the window

Before 10:00 am, mark where resting orders are likely to sit:

• the high and low of the 9:30 to 10:00 am range,
• the overnight session high and low,
• the previous day's high and low,
• any obvious equal highs or equal lows.

Decide which side is the likely target (the draw) for the next hour. If you can't name it in one sentence, the window probably isn't for you today.

2. Wait for a sweep inside the window

Price runs one of the nearby levels, trading a few points through a high or low, then fails to continue. That is the manipulation in the accumulation, manipulation and distribution sequence the model is built on.

3. Look for displacement and a fair value gap

After the sweep, price should move away sharply. That fast move leaves a three-candle fair value gap: a space between the first candle's high and the third candle's low (for a bullish gap) where price traded only one way. If you need a refresher, read fair value gap and inverse FVG explained.

4. Enter in the gap, stop beyond the sweep, target the opposite pool

• Entry: a limit order inside the gap, often at the edge or at the midpoint (consequent encroachment).
• Stop: beyond the swing that made the sweep.
• Target: the liquidity pool on the other side, the level you mapped in step 1.

If no gap forms before 11:00 am, there is no trade.

A worked NQ example

This is a hypothetical example to show the arithmetic, not a trade idea.

Before the window

• 9:30 to 10:00 am range low: 24,060.00.
• Overnight high: 24,142.00. Bias is bullish, so the draw is the overnight high.

Inside the window

• 10:07 am: price trades down to 24,054.50, taking the 24,060.00 low by 5.5 points, and closes back above it.
• 10:09 to 10:11 am: three strong one-minute candles up. Candle one's high is 24,071.00; candle three's low is 24,079.50. That leaves a bullish fair value gap from 24,071.00 to 24,079.50, with its midpoint at 24,075.25.

The plan

Sizing with $250 of risk

• One NQ: 22.75 × $20 = $455. Too large.
• One MNQ: 22.75 × $2 = $45.50. $250 ÷ $45.50 = 5.49, so 5 MNQ, risking $227.50.
• If the target is reached: 66.75 × $2 × 5 = +$667.50, which is +2.93R.

If the retrace never reaches 24,075.25 before 11:00 am, the order is cancelled and the day is a "no fill". That counts as a valid outcome in your log.

The model as a checklist

Writing the model down as a checklist makes it far easier to follow and to test. One way to score it:

In TJ, this is exactly what a strategy in Rate It! looks like: checks with points, vetoes that force "no trade", and a grade from A+ to D that sets the size. The community strategy library includes ICT templates, among them iFVG models, that you can copy and edit rather than starting from a blank page.

Which window should you trade?

Pick one, at least to begin with. The three windows suit different traders:

• 3:00 to 4:00 am: inside the London session. It suits traders in Europe and Asia, and markets that move in London, such as gold, EURUSD and GBPUSD. For US-based traders it means a very early start.
• 10:00 to 11:00 am: the most popular window for NQ and ES. The 8:30 data and the 9:30 open have already set the morning's range, so there are clear levels to sweep. It is also the busiest hour, which means more setups and more fake-outs.
• 2:00 to 3:00 pm: quieter on most days, with the exception of Fed decision days, when the 2:00 pm statement can make this the wildest hour of the week. Many traders skip it on those days.

Trading all three windows triples your screen time and makes your statistics harder to read, because each window behaves differently. Master one, log at least 50 windows of it, then decide whether a second adds anything.

Managing the trade

The model defines the entry, stop and target, but you still have to decide what happens in between. Write the rules before you trade:

• Breakeven: many traders move the stop to entry only after price has gone +1R, not earlier. Moving too soon turns normal retracements into scratches.
• Partials: taking half off at +1R and holding the rest for the target lowers variance but also lowers expectancy if the model's winners tend to reach the target. Your journal can tell you which is true for you.
• Time stop: if the trade hasn't moved meaningfully by the end of the window, some traders close it, on the logic that the window's move has failed.

Whatever you choose, keep it fixed for a full sample so the results mean something.

Common mistakes

Trading after the window closes. A gap that forms at 11:04 isn't part of the model. If you take it anyway, tag it separately so you can see what those trades are worth.

Calling every wick a sweep. A sweep should take a level you marked before the window. Random wicks through intraday noise don't count.

No displacement. A slow drift away from the low leaves small, overlapping candles and weak gaps. The model depends on a decisive move.

Second and third attempts. If the first trade stops out, the model hasn't promised another one. Decide your maximum trades per window in advance; many traders use one.

Stops that are too tight. Placing the stop at the gap instead of beyond the sweep makes the R:R look great and the win rate collapse.

Ignoring news. Some 10:00 am releases land at the very start of the window. Check the calendar before you sit down.

How to backtest the Silver Bullet

Because the model is tied to a clock, it's well suited to replay testing.

1. Pick one market and one window. NQ at 10:00 to 11:00 am is a common choice.
2. Replay day by day. Start each day before 9:30 am, mark your levels, then step through the window one bar at a time without looking ahead.
3. Log every window. Outcome options: win, loss, breakeven, no setup, no fill. A model that triggers on 40% of days behaves very differently from one that triggers every day.
4. Record R, not dollars. And note the grade you would have given the setup at the time of entry.
5. Aim for at least 100 windows before drawing conclusions, then test a different period that you didn't use while learning the model.

Our guide to backtesting a strategy properly covers sample size, look-ahead bias and how to compare backtest results with live ones.

What to review after 50 live trades

• Expectancy in R per window and per market.
• Results by entry type: gap edge vs midpoint.
• Win rate when the higher-timeframe bias agreed vs when it didn't.
• How many trades hit the target vs were closed early by hand.
• Results on news days vs quiet days.

If one of these splits shows a clear difference across a decent sample, turn it into a rule. If it doesn't, leave the model alone.

Not financial advice. The Silver Bullet is a way of reading price inside a time window; it doesn't predict outcomes, and trading futures involves substantial risk of loss.