Markets › FX · GBPUSD spot (broker data), 1-minute

London Breakout Strategy Backtest: GBPUSD, 22 Years of 1-Minute Data

The classic Asian range breakout at the London open, on every trading day of GBPUSD from 2004 to 2026: win rate, three exits, costs, and the test most backtests skip, a comparison with the same trades on random days.

GBPUSD · 1-minute data · Asian range 00:00-07:00 London, breakout 07:00-10:00 · Protocol v1.0 plus the 8 Oct 2026 rules

Buried · No proven edge

It beat chance. It did not beat the spread.

BornFrozen 21 Oct 2026 (lb_spec.md), before any range was computed

Died21 Oct 2026

Variants tried: 3 · Passed every gate: 0 · Beat the random-day control: 2 of 3 · Holdout opened: yes (already used)

Short answer: almost, but not after costs. On GBPUSD from 2004 to 2026, buying or selling the first break of the Asian range (00:00 to 07:00 London) between 07:00 and 10:00, with the stop at the other side, beat the same trades placed on random days, but its edge was about the size of the spread: after 1.5 pips a round trip it made -0.018 R a trade with a 1R target and +0.016 R with no target, over 5,278 trades.How much costs took here, compared across all 54 of our tests →

00The verdict in three lines

We tested the London breakout strategy on GBPUSD from 2004 to 2026: mark the Asian range (00:00 to 07:00 London), buy a break of its high or sell a break of its low between 07:00 and 10:00, stop at the other side, out by 16:00 London. 1.5 pips a round trip.
It has a small real edge, smaller than the costs. Over 5,278 trades the breakout made +0.025 R to +0.059 R a trade before costs and beat the same trades placed on other days (1R and 2R targets), but after costs: -0.018 R (1R), +0.004 R (2R), +0.016 R (no target).
Win rate: 50.9% with a 1R target, 43.3% with 2R, 41.6% with no target. Not proven in the first nine years on their own, flat from 2013 to 2019: buried.

01What is the London breakout strategy?

Sterling is quiet during Asian hours and wakes up when London opens. The London breakout (also called the Asian range breakout or London open breakout) trades that wake-up: the high and low of the night form a box, and the first break of the box after the open is taken as the day's direction. Write-ups agree on the core and differ on the details; we fixed this common reading before running anything:

Asian range: the high and low from 00:00 to 07:00 London time (19:00 to 02:00 New York in winter). Days with fewer than 300 of the 420 minutes in the data are skipped.
London breakout time: from 07:00 London, a buy stop 1 pip above the range high and a sell stop 1 pip below the low; the first one hit is the trade, the other is cancelled. One trade a day, no entry after 10:00 London.
Stop: the other side of the range (1 pip beyond it). That distance is 1R; on GBPUSD it was a median 38 pips.
Three exits: LB-1R (target 1R), LB-2R (target 2R) and LB-CLOSE (no target). Every trade still open closes at 16:00 London, the end of the London session.

02How we tested it

GBPUSD, the 1-minute bid history of an IC Markets MetaTrader 4 demo server, January 2004 to 28 September 2026. London time is computed with each country's own clock changes and checked against the operating system's time-zone database for every trading hour (0 differences). A check of the file's own clock found no week to correct; the check we first wrote (the jump in activity at the London open) could not tell, because activity already rises at 06:00 with Frankfurt, and was replaced before any result by the week-open hour. Costs: 1.5 pips a round trip, raised from the 1.0 pip of our earlier FX study before any result, because stop orders at the London open slip and spreads were 2 to 3 pips in the early years; results at 1.0 and 2.5 pips are shown. No swap: every trade closes the same day. Periods: 2004 to 2012, 2013 to 2019, and a holdout from 2020 already used by our FX study. Control: every trade moved to another day at the same London time, same direction and stop distance, 500 times. Engine check: hand-worked days and an independent re-implementation (with the time-zone database instead of our clock rules) that matched all 5,278 trades of every exit (lb_selftest_result.txt).

03Does the London breakout strategy work?

Net R after 1.5 pips a round trip. R is the distance from the entry to the stop.

Variant	Exit	Trades	Win rate (95% range)	Gross per trade	Cost per trade	Net per trade	Net total	Net pips	Profit factor	Verdict
LB-1R	Target 1R (the other side of the range = 1R)	5,278	50.9% (50% to 52%)	+0.025 R	0.043 R	-0.018 R	-97 R	-2,344 pips	0.96	Buried
LB-2R	Target 2R	5,278	43.3% (42% to 45%)	+0.047 R	0.043 R	+0.004 R	+20 R	+2,062 pips	1.01	Buried
LB-CLOSE	No target: out at 16:00 London	5,278	41.6% (40% to 43%)	+0.059 R	0.043 R	+0.016 R	+83 R	+3,936 pips	1.03	Buried

Cumulative net R by month. * Holdout already used before this test.

By period

Variant	2004 to 2012 (trades)	2013 to 2019 (trades)	Holdout* 2020 to 2026 (trades)	t, first period	PF, second period
LB-1R	-1.3 R (1,968)	-40.4 R (1,683)	-54.9 R (1,627)	-0.03	0.95
LB-2R	+34.2 R (1,968)	-28.3 R (1,683)	+14.4 R (1,627)	0.72	0.97
LB-CLOSE	+48.6 R (1,968)	+14.5 R (1,683)	+20.2 R (1,627)	0.92	1.01

London breakout success rate: how the trades ended

Variant	Hit the target	Hit the stop	Closed at 16:00 London
LB-1R	41.5%	39.1%	19.4%
LB-2R	17.2%	44.2%	38.6%
LB-CLOSE	–	45.4%	54.6%

The win rate of a breakout is mostly set by its exit: a 1R target wins about half the time by construction, a 2R target less often, and holding to the close lets many trades end small either way. The question is whether the average beats the costs, and here it barely did not.

04Breakout or chance?

The moved-entry control places every trade on another day, at the same London time, in the same direction, with the same stop distance and exits, 500 times. If the breakout signal adds nothing, the real trades should look like those.

Variant	Trades	Real net per trade	Control median	Control 5th to 95th	Percentile (95 needed)
LB-1R	5,278	-0.018 R	-0.043 R	-0.067 R to -0.020 R	95.4
LB-2R	5,278	+0.004 R	-0.029 R	-0.059 R to +0.001 R	96.8
LB-CLOSE	5,278	+0.016 R	-0.008 R	-0.046 R to +0.035 R	80.8

With a 1R or a 2R target the breakout beat at least 95% of the control runs: the trades placed at random lost about the cost (-0.043 R at 1R), the breakout trades did several hundredths of an R better. Very few strategies we have tested pass this gate. Without a target it did not (percentile 80.8): holding to the close lets the day's noise swamp the signal.

05The spread decides it

Variant	Before costs	At 1.0 pip	At 1.5 pips (verdict)	At 2.5 pips
LB-1R	+0.025 R	-0.004 R	-0.018 R	-0.047 R
LB-2R	+0.047 R	+0.018 R	+0.004 R	-0.025 R
LB-CLOSE	+0.059 R	+0.030 R	+0.016 R	-0.013 R

Net per trade at three costs. At 1.0 pip (a tight raw-spread account with good fills) the no-target version made +0.030 R a trade; at 2.5 pips (a standard account at the open) every version lost. A difference of half a pip a trade is the whole result.

06The Asian range is shrinking

Year	Median Asian range (pips)	Trades	LB-1R net	LB-2R net	LB-CLOSE net
2004	61	237	+10.2 R	+29.2 R	+36.5 R
2005	55	219	+6.3 R	+7.5 R	+6.2 R
2006	45	238	+1.7 R	+12.7 R	+29.3 R
2007	59	134	+8.6 R	+13.5 R	+19.2 R
2008	82	189	+6.3 R	+16.9 R	+16.4 R
2009	88	231	-0.5 R	-0.2 R	-4.6 R
2010	63	240	-16.8 R	-12.6 R	-13.7 R
2011	50	237	-19.6 R	-18.9 R	-28.9 R
2012	31	243	+2.6 R	-13.8 R	-11.8 R
2013	33	236	+10.9 R	+32.7 R	+45.7 R
2014	28	235	-25.3 R	-25.8 R	-26.2 R
2015	28	247	-44.8 R	-52.5 R	-39.7 R
2016	36	237	-2.7 R	+16.9 R	+27.9 R
2017	27	236	+14.6 R	-4.1 R	-8.6 R
2018	27	251	+0.5 R	+1.9 R	-9.9 R
2019	22	241	+6.5 R	+2.6 R	+25.3 R
2020	35	240	+8.7 R	+12.9 R	+12.0 R
2021	28	249	-1.9 R	+0.4 R	-11.1 R
2022	39	234	-23.4 R	-6.1 R	+16.1 R
2023	28	247	-22.4 R	-7.1 R	-1.5 R
2024	20	250	+13.5 R	+25.2 R	+14.0 R
2025	28	231	-16.6 R	-3.1 R	+1.6 R
2026 (to 28 Sep)	24	176	-12.7 R	-7.8 R	-10.9 R

GBPUSD's overnight range was about 61 pips in 2004 and 82 in 2008, and about 20 to 28 in 2024 and 2025, so a fixed cost takes a bigger bite now. 2007 has only 134 trades because the file is missing many night minutes that year, and days with fewer than 300 minutes in the range were skipped as the rules say.

By Asian range size (descriptive, not a variant)

Range	Trades	LB-1R net per trade	LB-2R net per trade	LB-CLOSE net per trade
0 to 20 pips	686	-0.071 R	-0.020 R	0.000 R
20 to 30 pips	1,327	-0.002 R	0.000 R	+0.015 R
30 to 40 pips	1,009	-0.050 R	-0.027 R	-0.012 R
40 to 60 pips	1,183	+0.010 R	+0.041 R	+0.055 R
60 to more pips	1,073	-0.006 R	+0.013 R	+0.011 R

Ranges of 40 to 60 pips did best for all three exits; the weakest were the smallest ranges (under 20 pips, where the cost is largest in R) and, oddly, 30 to 40 pips. The pattern is not orderly, and this is the best of five buckets seen after the fact. A filter chosen from this table would be curve fitting; one fixed in advance could be tested as a new strategy.

Long and short

Variant	Longs: trades / gross / net (R)	Shorts: trades / gross / net (R)
LB-1R	2,585 / +10.8 R / -101.2 R	2,693 / +120.3 R / +4.6 R
LB-2R	2,585 / +68.4 R / -43.6 R	2,693 / +179.6 R / +63.9 R
LB-CLOSE	2,585 / +90.6 R / -21.3 R	2,693 / +220.4 R / +104.7 R

Short breakouts did better than long ones; GBPUSD fell from about 1.80 to 1.33 over these years, so part of that is the trend.

07The gates

Variant	Trades	1. PF ≥ 1.2 in validation	2. ≥ 50 trades in each period	3. t ≥ 2.0 on development	4. Above the 95th percentile of the moved-entry control	5. Positive without the 10 best trades	6. Holdout net positive (already used)
LB-1R	5,278	fail	pass	fail	pass	fail	fail
LB-2R	5,278	fail	pass	fail	pass	fail	pass
LB-CLOSE	5,278	fail	pass	fail	fail	pass	pass

The data-gap check

Variant	Trades	Open across a suspect gap	Net per trade	Verdict
LB-1R	5,278	10	-0.018 R → -0.018 R	Buried → Buried
LB-2R	5,278	12	+0.004 R → +0.004 R	Buried → Buried
LB-CLOSE	5,278	14	+0.016 R → +0.016 R	Buried → Buried

08Verdict by variant

Variant	Cause	Gates failed	Verdict
LB-1R	Costs	1, 3, 5, 6	Buried
LB-2R	Fragile	1, 3, 5	Buried
LB-CLOSE	No proven edge	1, 3, 4	Buried

Costs (LB-1R): positive before costs, negative after. Fragile (LB-2R): positive overall, not without its ten best trades. No proven edge (LB-CLOSE): positive, but not better than random days and not proven in the first period. None of this says the London breakout cannot work for anyone; it says this common version, at realistic costs, did not clear our bar.

TWhat this shows, and what to take into your own tests

What the results show

The London breakout on GBPUSD does pick better moments than chance: with a 1R or 2R target its trades beat at least 95% of runs where the same trades were placed on other days at the same London time. That is rare on this site.
But the edge is small: +0.025 R to +0.059 R a trade before costs, against about 0.04 R of costs on a typical 38-pip stop. After 1.5 pips a round trip the three exits made -0.018 R, +0.004 R and +0.016 R a trade.
The win rate depends on the exit, not on the edge: 50.9% at 1R, 43.3% at 2R, 41.6% with no target. None of the three was proven in 2004 to 2012 on its own.

What remains unknown

Other pairs (EURUSD, gold, the DAX) and other range hours. We tested one common reading on GBPUSD only.
Filters (a minimum or maximum Asian range, a news calendar, a trend filter) chosen in advance. Our range-size table is descriptive; picking the best bucket after the fact would be curve fitting.
Real fills at the London open. We assumed 1.5 pips a round trip; a raw-spread account with fast execution might pay less, a standard account more.

Why it was buried

The breakout's timing edge is real enough to beat our random-day control at 1R and 2R, but it is about the size of the spread and commission: GBPUSD's Asian range shrank from about 61 pips in 2004 to about 20 in 2024, so the same 1.5-pip cost is a bigger share of each trade every year. The first period, 2004 to 2012, did not show the edge on its own (t below 1 for all three exits), and the validation years 2013 to 2019 were flat to negative.

What you can apply to your own tests

Compare your breakout with the same trades on other days. Ours did better than that, which a profit total alone would not have shown, and still lost after costs.
Put your cost in R. On a 38-pip stop, 1.5 pips is 0.04 R; on a 15-pip Asian range it is 0.1 R.
Do not judge it by the win rate. A 1R target wins half the time by construction; what matters is the average after costs.

A note on the numbers: drawdown figures are measured on closed trades unless a table says it includes open losses. Open losses inside a trade are not counted in a closed-trade drawdown, so real account risk is higher, especially for systems with no stop. Backtests are not live trading.

FAQCommon questions

Does the London breakout strategy work?

It has a small real edge on GBPUSD but not enough to beat realistic costs. Before costs it made +0.025 R to +0.059 R a trade; after 1.5 pips a round trip, -0.018 R (1R target), +0.004 R (2R) and +0.016 R (no target). None passed our gates.

What is the London breakout strategy success rate?

In our test the win rate was 50.9% with a 1R target, 43.3% with a 2R target and 41.6% holding to 16:00 London. The win rate mostly reflects the exit; the average result after costs is what decides.

What time is the London breakout?

The Asian range is usually taken from 00:00 to 07:00 London time and the breakout traded from 07:00, the London open (02:00 New York in winter, 09:00 on most MT4 server clocks). We allowed entries until 10:00 London and closed every trade by 16:00 London.

What is the Asian range breakout?

The same idea: the high and low of the quiet Asian session form a box, and a stop order just beyond each side catches the first strong move after London opens. The stop goes at the other side of the box.

How was the London breakout tested?

On 22 years of 1-minute GBPUSD with rules written and dated before the data was run, costs included on every trade, results split into development, validation and holdout periods, a random-entry control as the benchmark, and every variant reported. Read the protocol.

09Check our arithmetic

Frozen specificationThe range, the entry, the exits, costs, periods, gates and control, dated before any range was computed, with its two dated corrections (the cost, the clock check).
lb_spec.md

CodeThe engine (its readings of the rules at the top), the independent re-implementation, the self-test, the run (node lb_run.js), and the clock check and gap audit (node lb_clock.js). Needs your own MT4 GBPUSD history.
lb_lib.js · lb_ref.js · lb_selftest.js · lb_run.js · lb_clock.js

Shared code it usesThe MT4 loader, statistics and control helpers.
control_audit_lib.js · fx_data.js · gh_lib.js · gx_lib.js · hourly_lib.js · ma_lib.js · rsi2_protocol.js

ResultsSelf-test result, the console output of the run, all numbers, the clock check and the gap list.
lb_selftest_result.txt · lb_results.txt · lb_summary.json · lb_clock_result.json · data_holes_gbpusd.json

Trade logs (CSV)Every trade: entry and exit (New York time), the London day, direction, entry, stop, exit, R and the Asian range in pips, the exit reason, gross and net pips and R.
LB-1R · LB-2R · LB-CLOSE

Code provenance: our own code, written from the rules in the specification.

10Data notes

Source. A broker's demo history of spot GBPUSD, bid prices; minutes with no trades are missing, most often at night and in the early years (see 2007). Spreads are not in the file, so costs are an assumption, shown at three levels. Gaps. 173 suspect gaps outside weekends and the daily rollover; without the trades that cross one, no verdict changes.

11Corrections

22 Oct 2026, navigation: added a link to the new guide comparing trading costs across all our tests. No result changed. 22 Oct 2026, navigation: added a link to the new guide comparing trend following and mean reversion across all our tests. No result changed. 22 Oct 2026, navigation: added a link to the new guide comparing versions with and without a stop loss across our tests. No result changed.