ICT intraday model · sweep, gap, limit · 1 or 5-minute
Silver Bullet
Three one-hour windows, New York time. Inside one, price runs through a level nobody has hit yet, then a
strong candle snaps back and leaves a gap. Buy or sell the return to that gap, stop past the sweep, target the next
untouched level. That's the model. Tested over 16 years, it lost money in every hour.
Three windows, New York time: 3 to 4 am (London), 10 to 11 am and 2 to 3 pm. Nothing outside them counts.
Liquidity is any level price hasn't traded through yet: swing highs and lows, the Asia and London session ranges,
the previous day's high and low, and the range of the hour before the window.
ICT's idea: stops pile up just past these levels, and the market runs them before it moves the other way.
2
The sweep
A high is swept when a candle trades at least one tick above it. A low, at least one tick below. A wick is enough; an exact touch isn't.
Each level can be swept once, and only the window's first sweep on each side counts.
A swept high starts the search for a short. A swept low, for a long.
The sweep is the stop run. On its own it isn't a signal; it only tells the indicator which way to look.
3
Displacement leaves a gap
A fair value gap the other way: three candles where the first and third don't overlap. The middle candle's body is at least 1 ATR(14).
The gap is at least 0.1 ATR (and 2 ticks). All three candles are inside the window, and the gap can't start before the sweep.
A swept high needs a gap down; a swept low needs a gap up. Only the first one per side in a window counts.
A candle that big right after a stop run is the snap-back the model is waiting for.
4
A limit at the gap, stop past the sweep
Entry: a limit order at the edge of the gap nearest to price.
Stop: one tick beyond the sweep's extreme, the highest high for a short or the lowest low for a long, from the sweep candle to the end of the gap.
Target: the nearest untouched level on the other side. Closer than 1.5 times the risk, the setup is skipped. No level at all, the target is 2R.
The order waits for price to come back into the gap rather than chasing the displacement candle.
5
The fill, or no trade
The order fills only if price trades at least one tick through it before the window ends.
It's cancelled if price reaches the target first, or the window ends.
Each window takes one trade at most.
This is where the study found the problem: the orders that fill are mostly the trades going wrong.
6
Target, stop or the clock
The trade closes at the target, the stop, or one hour after the window ends, whichever comes first.
If the stop and the target are both hit in the same candle, it counts as a stop.
Results are counted in R after slippage and fees. R is the distance from entry to stop.
Counting in R makes a trade on a quiet morning comparable with one on a busy afternoon.
The video
Part 1 walks through the indicator and its settings. Part 2 is the 16-year backtest.
Every decision is made on a closed bar, never on a bar that's still forming.
No fill in the window
The limit is cancelled when the window ends, or as soon as price reaches the target without filling it.
The time stop
An open trade is closed one hour after its window ends: 5 am, noon or 4 pm New York.
One trade per window
Once one side fills, any order waiting on the other side of the same window is dropped.
The liquidity it hunts
Five kinds of level. Each one is in play until price trades a tick through it.
Swing highs and lows
A high above the 5 candles before it and the 5 after. Marked once those 5 have closed, dropped after 300 candles.
Asia range
The high and low from 8 pm to midnight New York.
London range
The high and low from 2 to 5 am New York.
Previous day
Yesterday's high and low, with the day starting at 6 pm.
The hour before
The high and low of the clock hour just before the window.
Does it work? The 16-year test
The backtest in part 2 of the video asks one question: are these three hours special, or would the same rules work just as
well, or just as badly, at any other time of day?
92 / 92
After costs, the strategy lost money in all 23 trading hours on all four markets.
≈ 0
Before costs it roughly broke even or lost a little. There's no hidden edge for costs to eat.
73%
of trades hit the stop. It wins about one trade in four.
3 in 4
of the setups whose order never filled would have won.
Exact rules. Every vague step of ICT's description became a rule a computer follows with no judgement calls.
Every hour. The same rules ran in all 23 trading hours, so the three windows could be compared with all the others.
16 years of data. One-minute bars, 2011 to 2026: ES and NQ to look for an effect, the Dow and gold futures to check it.
Real costs. Two ticks of slippage and a $4.50 round-trip commission on every trade, shown with and without.
Measured in R. +1R means you won what you risked. The average R per trade says whether it makes money over time.
Why the Silver Bullet hours look better
At first the three windows looked slightly better than the average hour. They still lost money, and the reason isn't the clock.
ES slippage — 2 ticks × 0.250.50 pt
Commission — $4.50 ÷ $50 a point0.09 pt
Cost per trade0.59 pt
Quiet hour, 3-point stop0.59 ÷ 3 = 0.20R
Busy hour, 8-point stop0.59 ÷ 8 = 0.07R
Busy hours, wider stops, cheaper costs
The windows sit on some of the busiest hours of the day. Bigger moves mean wider stops, so the same fixed cost is a smaller
share of each trade. That alone makes them look better than a quiet hour.
Matched against other hours that move just as much, the advantage disappeared on all four futures.
The cost figures above are arithmetic for ES, not results from the study.
The trades you miss are the winners
The same gap, two endings. A limit order only fills when price comes back, and a move that comes all the way back is often a
move that's failing.
In the study about three in four of the unfilled setups would have reached the target. That points to a
different entry worth testing: a market order at the sweep instead of a limit at the gap. These two charts are illustrative.
Long and short, side by side
The whole rule set on one table, as the indicator codes it at its default settings.
Longafter a swept low
Shortafter a swept high
Window
3:00–4:00, 10:00–11:00 or 14:00–15:00 New York
Sweep
A low 1 tick under an untouched sell-side level
A high 1 tick over an untouched buy-side level
Gap
Up: third candle's low over the first's high
Down: third candle's high under the first's low
Displacement
Middle body ≥ 1 ATR(14), gap ≥ 0.1 ATR and 2 ticks, all three candles in the window, after the sweep
Entry
Buy limit at the gap's top edge
Sell limit at the gap's bottom edge
Fill
A low 1 tick under the limit, before the window ends
A high 1 tick over the limit, before the window ends
Stop
lowest low (sweep → gap) − 1 tick
highest high (sweep → gap) + 1 tick
Target
Nearest untouched high, at least 1.5R; 2R if there's none
Nearest untouched low, at least 1.5R; 2R if there's none
Exit
Target, stop (both in one candle counts as the stop) or one hour after the window ends
Cancelled by
The target trading before the fill, the window ending, or the other side filling first
Example: one winning long, 1-minute bars
An example session run through the indicator's own rules, with the pre-market included. Inside the 10–11 am window, the
10:14 candle trades a few cents under the 9:47 low, the low of the hour before (the sweep). The next candle snaps up and leaves a gap,
the buy limit at the gap's top edge (100.62) fills on the pullback at 10:17, the stop sits a tick under the sweep at 100.35, and at 10:38 price
reaches the 9:34 high at 101.20, the nearest untouched level above: about +2R after slippage. The bars are made up to show the setup clearly.
They aren't a real ticker, and one clean winner is not a result: in the 16-year test above, most of these trades lost.
Loading chart…
1-minute
Indicator for this setup
Add this to a TradingView chart on 1 to 15-minute bars. It marks the windows and the liquidity, the sweeps
and the gaps, places and tracks every trade, and keeps score by window or by hour. It does not place trades.
Silver Bullet [GBB]
alerts
The three windows framed and tinted, the session levels named, the window's first sweep crossed, the
gap boxed, and each trade's risk and reward zones with a result chip. A stats panel shows trades, win rate and average net R
for each window. Hour Scan mode runs the same rules in every clock hour. Alerts for the window, the sweep, the armed setup,
the fill and the exit.
On a stock chart, set Commission to 0 or to your per-share cost. The default $4.50 is per futures
contract and would be counted as $4.50 a share. The 3 am window needs a futures chart with overnight hours.
With rules built from ICT's teaching, the Silver Bullet hours only look better because they're the busiest hours, and the
strategy lost money in every hour on every market tested over 16 years. The parts can still be useful on their own.
Use the sweep detection as a trigger and combine it with your own setups.
Use it as a very short scalp right as a sweep happens, not as a limit order waiting at the gap.
Test a market entry at the sweep. The unfilled setups were mostly the winners.
Rules as coded in the Silver Bullet [GBB] indicator (Mozilla Public License 2.0) at its default settings. Study results
as reported in the video. The walkthrough and the two charts above are illustrative, not a real ticker.