Key takeaways
- On GBP/USD the textbook London breakout made +94.7R before costs over ten years, and −37.8R after a 1.4-pip spread. The edge was real, but smaller than the spread.
- It worked from 2016 to 2020 (+35R) and stopped working after (−72.8R from 2021). A backtest that ends in 2020 would have sold you a strategy that then lost.
- EUR/USD lost in nine of the eleven years tested (−135.8R). USD/JPY was flat.
- One change helped GBP/USD and USD/JPY: a 2R target instead of 1R. But picking the best variant after the fact is exactly how backtests fool people.
The London breakout is one of the oldest intraday strategies in forex, and one of the most taught. The idea is simple: the Asian session is quiet, so price spends the night in a range. When London opens, liquidity floods in and price breaks out of that range, often with momentum. Buy the break of the high, sell the break of the low, and ride the move.
It sounds logical. The question is whether it makes money once you trade it the way a retail account actually would: with a spread, with stop orders, and without hindsight. So we tested it on ten years of prices.
The rules we tested
We used the most common version of the strategy, with nothing tuned:
- The range. The highest and lowest bid price between 00:00 and 07:00 London time.
- The orders. From 07:00 to 11:00 London time, a buy stop 1 pip above the range high and a sell stop 1 pip below the range low. The first to trigger is the trade; the other is cancelled.
- The stop. The other side of the range, plus the same pip. So the risk on each trade is the range’s height, which we call 1R.
- The target. 1R beyond the entry: an equal reward for the risk.
- The time exit. Anything still open at 16:00 London time is closed at the market.
- Costs. A spread charged on every trade at typical standard-account levels: 1.4 pips on GBP/USD, 1.0 on EUR/USD, 1.2 on USD/JPY.
The result on GBP/USD
GBP/USD is the pair most often recommended for this strategy, because sterling reacts sharply to the London open. Over 2,691 trading days the strategy traded almost every day, with a typical risk of 30 pips.
Show the data
| 2016-01-04 | +1.0R |
| 2016-05-10 | +3.3R |
| 2016-09-26 | +1.6R |
| 2017-02-03 | +6.6R |
| 2017-06-15 | +18.0R |
| 2017-10-25 | +15.0R |
| 2018-03-08 | +19.7R |
| 2018-07-13 | +20.5R |
| 2018-11-20 | +21.7R |
| 2019-03-29 | +19.8R |
| 2019-08-13 | +27.2R |
| 2019-12-23 | +28.5R |
| 2020-05-06 | +38.7R |
| 2020-09-16 | +30.6R |
| 2021-01-22 | +39.2R |
| 2021-06-01 | +24.6R |
| 2021-10-05 | +23.5R |
| 2022-02-09 | +11.8R |
| 2022-06-20 | +9.0R |
| 2022-11-01 | +7.0R |
| 2023-03-10 | -2.0R |
| 2023-07-18 | -9.8R |
| 2023-11-22 | -21.5R |
| 2024-04-03 | -29.3R |
| 2024-08-12 | -33.2R |
| 2024-12-18 | -16.2R |
| 2025-05-12 | -20.5R |
| 2025-09-22 | -28.4R |
| 2026-02-04 | -32.5R |
| 2026-06-19 | -32.0R |
| 2026-10-16 | -37.8R |
After costs the strategy finished at −37.8R, with a win rate of 51.4% and a profit factor of 0.97. That is close to break-even, and close to break-even after a decade is a loss: you would have spent ten years taking around 250 trades a year, through a worst drawdown of 87.8R, to end up behind.
What makes this interesting is the result before costs. Without the spread, the same trades made +94.7R. The breakout did have a small edge on GBP/USD, of about 0.035R a trade. The 1.4-pip spread costs about 0.047R a trade on a 30-pip risk. The edge was real; it was just smaller than the cost of taking it.
It worked, until it didn’t
The equity curve has a shape that should worry anyone who backtests: it rises, then falls.
Show the data
| 2016 | +2.8R |
| 2017 | +21.2R |
| 2018 | +3.7R |
| 2019 | -0.5R |
| 2020 | +7.8R |
| 2021 | -16.6R |
| 2022 | -17.7R |
| 2023 | -23.6R |
| 2024 | +8.1R |
| 2025 | -13.1R |
| 2026 | -9.9R |
Split the test into two halves and you get two different strategies:
| Period | Trades | Win rate | Result | Profit factor |
|---|---|---|---|---|
| 2016–2020 | 1,246 | 53.4% | +35.0R | 1.06 |
| 2021–2026 | 1,445 | 49.6% | −72.8R | 0.89 |
Anyone who tested the London breakout in 2020, with data to that year, would have found a modestly profitable system with a 24R drawdown. Then it lost money in five of the next six years. This is the most important lesson in the whole test, and the reason our guide to backtesting insists on testing a strategy on data it has never seen.
EUR/USD and USD/JPY
The same rules on two more majors:
Show the data
| 2016-01-04 | -1.0R |
| 2016-05-13 | -0.5R |
| 2016-09-29 | -14.2R |
| 2017-02-17 | -14.4R |
| 2017-06-28 | -28.6R |
| 2017-11-14 | -24.3R |
| 2018-03-28 | -26.9R |
| 2018-08-02 | -30.6R |
| 2018-12-24 | -42.2R |
| 2019-05-02 | -57.7R |
| 2019-09-16 | -70.9R |
| 2020-01-27 | -81.1R |
| 2020-06-08 | -81.4R |
| 2020-10-23 | -81.4R |
| 2021-03-09 | -72.4R |
| 2021-07-15 | -76.7R |
| 2021-11-24 | -93.7R |
| 2022-04-05 | -91.7R |
| 2022-08-12 | -88.5R |
| 2022-12-28 | -106.8R |
| 2023-05-10 | -107.6R |
| 2023-09-15 | -119.7R |
| 2024-01-30 | -126.8R |
| 2024-06-10 | -140.8R |
| 2024-10-18 | -146.4R |
| 2025-03-07 | -129.4R |
| 2025-08-08 | -138.0R |
| 2025-12-31 | -139.6R |
| 2026-05-21 | -141.1R |
| 2026-10-12 | -137.4R |
| 2026-10-16 | -135.8R |
EUR/USD was worse: −135.8R over 2,614 trades, losing in nine of eleven calendar years. It lost even with no spread at all (−17.5R). EUR/USD’s Asian range is narrower (a typical risk of 24 pips), so the spread is a bigger share of each trade, and the breakouts it produces follow through less often.
Show the data
| 2016-01-04 | -0.1R |
| 2016-05-20 | +4.1R |
| 2016-10-19 | +8.6R |
| 2017-02-28 | +13.0R |
| 2017-06-29 | +14.5R |
| 2017-10-31 | +15.9R |
| 2018-03-20 | +5.3R |
| 2018-08-09 | +3.9R |
| 2018-12-31 | +5.7R |
| 2019-06-06 | -6.2R |
| 2019-11-07 | -9.7R |
| 2020-03-20 | -14.4R |
| 2020-08-14 | -2.2R |
| 2020-12-15 | -3.7R |
| 2021-04-27 | +3.8R |
| 2021-08-17 | -3.9R |
| 2021-12-27 | -13.8R |
| 2022-05-12 | -4.7R |
| 2022-09-21 | -2.0R |
| 2023-01-25 | +5.3R |
| 2023-05-29 | +1.2R |
| 2023-09-25 | -0.7R |
| 2024-01-29 | -3.6R |
| 2024-06-06 | -6.6R |
| 2024-10-16 | -14.7R |
| 2025-03-07 | -10.6R |
| 2025-07-10 | -3.6R |
| 2025-11-06 | -9.3R |
| 2026-03-20 | -5.8R |
| 2026-07-30 | -8.1R |
| 2026-10-13 | +4.6R |
USD/JPY finished at +4.6R over 2,076 trades: effectively zero. The yen trades actively in Tokyo, so its Asian range is wider (40 pips) and its breakout at London is less of an event.
What about the variations?
Every trading forum has a fix for the London breakout. We tested the two most common:
| Variation | GBP/USD | EUR/USD | USD/JPY |
|---|---|---|---|
| Classic, 1R target | −37.8R | −135.8R | +4.6R |
| 2R target | +44.1R | −87.7R | +59.9R |
| Only narrow ranges (below the 20-day median) | −30.9R | −95.8R | −0.6R |
A 2R target turned GBP/USD and USD/JPY positive. Filtering for narrow Asian ranges, a popular tip, didn’t help.
Before you rush to trade the 2R version, consider how we found it. We tested three variations on three pairs and picked out the two winners. Test enough variations and some will look good by chance. A 2R target on GBP/USD returned 0.016R per trade over 2,691 trades, through a 34R drawdown. That is thin enough that a slightly wider spread, a bit of slippage on stop orders, or the regime change we saw after 2020 could erase it.
Our verdict
The London breakout isn’t a myth: on GBP/USD the open does produce follow-through more often than chance. But in its classic form, the edge is smaller than a standard account’s spread, and it has weakened since 2020. If you want to trade it:
- Cost is everything. A raw-spread account with low commission turns a losing version into a break-even one. Measure your real all-in cost per trade.
- Use GBP/USD or USD/JPY, not EUR/USD, and test a 2R target forward on a demo or tiny size before trusting it.
- Judge it over hundreds of trades. Drawdowns of between 27R and 154R happened in every version we tested. At 1% risk per trade, even the smallest is a 27% fall.
- Expect it to change. On GBP/USD its best years were 2017 (+21.2R) and 2020 (+7.8R); from 2021 it lost in five years out of six. Re-test it every year.
Volatility at the London open is real, and you can see it in every GBP/USD hourly profile. Turning that volatility into profit takes more than a range and two stop orders.
Method: Dukascopy 15-minute bid candles, 4 January 2016 to 16 October 2026. Weekdays only; days with less than six hours of Asian-session data or a shortened London session were skipped. Spread charged once per trade; no commission, swap or slippage on stop orders, so live results would be somewhat worse.
Frequently asked questions
Does the London breakout strategy work?
In our ten-year test with spreads charged, the classic version (Asian-range breakout between 07:00 and 11:00 London, stop at the other side of the range, 1R target) lost money on GBP/USD and EUR/USD and broke even on USD/JPY. GBP/USD showed a small edge before costs that faded after 2020.
What is the Asian range in the London breakout?
The high and low of price between midnight and 07:00 London time, before European traders arrive. The strategy places a buy stop just above that range and a sell stop just below it.
What is the best pair for the London breakout?
In our test GBP/USD and USD/JPY did better than EUR/USD, and both were profitable with a 2R target. Treat that as a hypothesis to test forward, not a proven edge: we tested several variants and reporting only the best one overstates it.
This article is for education and information, not investment advice. Trading forex and CFDs on margin is high risk and most retail accounts lose money. Read our risk warning.