ICT kill zones backtest: inside vs outside the zones
Trades entered inside the London and New York kill zones lost −0.10 R per trade after costs, against −0.16 R for trades entered outside them, over 7,450 backtested trades of five strategies on 20 assets. A gap of +0.05 R or more appeared in 21% of random relabellings of the same trades, so the kill zones did not separate good trades from bad ones.
What the kill zones are
In ICT terminology, kill zones are windows of the New York clock when the largest moves of the day are expected to start. The common definitions are the London kill zone, 02:00–05:00 New York time, the New York kill zone, 07:00–10:00, and the Asian kill zone, 20:00–00:00. They are tied to New York time, including its daylight saving time, so in UTC each of them moves by an hour twice a year, and in the weeks when Europe and the US switch clocks on different dates the London window also shifts against the London open.
The claim behind them is simple: a setup taken inside a kill zone is more likely to work than the same setup taken at another time. That claim can be tested directly, without building a new strategy.
How we tested them
We took the trades of five strategy templates that already exist in our grid (fair value gap, order block, level sweep, range deviation and Asian range), each run separately on 20 assets, and sorted every trade by the New York time of its entry. The rules, the stops, the targets and the costs are identical; only the entry time differs. If kill zones matter, trades inside them should do better.
The result
| Entry time, New York | Trades | Share of trades | Win rate | Per trade after costs |
|---|---|---|---|---|
| London kill zone, 02:00–05:00 | 1,297 | 17% | 27.0% | −0.09 R |
| New York kill zone, 07:00–10:00 | 1,278 | 17% | 23.8% | −0.11 R |
| Asian kill zone, 20:00–00:00 | 909 | 12% | 23.2% | −0.11 R |
| Outside all kill zones | 3,966 | 53% | 22.7% | −0.17 R |
| Inside London or New York | 2,575 | 35% | 25.4% | −0.10 R |
| Everything else | 4,875 | — | 22.8% | −0.16 R |
Trades inside the London and New York kill zones did a little better than the rest: 25.4% winners against 22.8%, and −0.10 R per trade against −0.16 R. The London window looked best of the three. The kill zones were better for 3 of the 5 templates, and worse for the others.
Is the difference real?
Some difference between any two groups of trades appears by chance. To measure how much, we shuffled the labels inside and outside between the trades of each template 2,000 times, keeping the number of trades in each group, and measured the gap every time. A gap at least as large as the real one came up in 21% of the shuffles. That is far too often to call the kill zones an effect: the same gap is easily produced by random sorting of the same trades.
More importantly, even inside the kill zones the result stayed negative after costs. A time filter can remove some of the worst trades, but it cannot turn a set of trades without an edge into one with an edge. If a setup does not work, trading it only at certain hours usually makes it lose less, not win.
What this means for your own testing
- Test a kill zone as a filter on a setup that already has rules: same trades, sorted by entry time.
- Check the gap against random relabelling of the same trades before trusting it.
- A filter chosen on the same history needs its own test on a later period.
- Use the New York clock with daylight saving time; fixed UTC hours put a part of the trades into the wrong window.
How we measured
- Engine backtests on the production server, five strategy templates, each of 20 assets separately, period 2022-01-01 to 2026-01-01, fixed risk of 1% per trade.
- Commission and swap measured for one commission-charging broker; entries and exits at bid and ask prices.
- Entry times converted from UTC to the America/New_York time zone with daylight saving time. Win rate = take-profits ÷ (take-profits + stop-losses). Placebo: inside/outside labels shuffled within each template 2,000 times; the share shown is how often the absolute gap was at least the real one.
Backtests run on October 30, 2026.
Questions
What time is the New York kill zone?
07:00–10:00 New York time in the most common definition; some traders use 08:30–11:00. It follows daylight saving time, so its UTC time changes in March and November.
Do kill zones work for forex?
In this test, on currency pairs, indices, metals, oil and bitcoin together, trades inside the kill zones did slightly better, but the gap was within chance and the result was still negative after costs.
Should I trade only inside kill zones?
Only if your own setup, tested inside and outside them, shows a difference larger than random relabelling produces, and the difference holds on a later period.
Why is the Asian kill zone often skipped?
Spreads are wider in the Asian hours, so the cost of a trade in R is higher; in this test the Asian window did no better than the London one.
For education, not investment advice. Past results, including backtests, do not guarantee future results.
← All articles