What is the London breakout strategy?
The London breakout is a rule that measures the high and low of the Asian session, waits for the London session to open, and enters long on a break above the Asian high or short on a break below the Asian low, with a stop on the far side of the range or at its midpoint.
The premise is that the Asian hours are quiet on the European pairs, so the range they form is narrow, and the arrival of European participation at the open resolves that range in one direction with enough momentum to carry a trade. The range provides the levels, the open provides the timing, and the break provides the direction.
Everything about the rule is observable on a clock and a chart, which is its appeal. There is no pattern to recognise and no judgement about structure. That also means there is nowhere for a poor result to hide: if the rule does not work under a fixed definition, it does not work, and no amount of discretion is available to rescue it.
Variants abound, including entering on the retest of the broken level rather than the break itself, trading only the first break, and fading the break when it fails. Each is a different rule and should be graded as one.
How is the Asian range defined?
The Asian range is defined by a start time, an end time, a timezone in which those times are fixed, and a choice between the wick extremes and the closing extremes of the bars inside the window.
The window is the first parameter and it is not standardised. Some definitions run from the Tokyo open to the London open; others use a narrower block of a few hours; others end an hour before London to leave a buffer. Each produces a different range on the same day, and a test has to state which was used and hold it fixed.
The timezone is the parameter that breaks tests silently. Tokyo does not observe daylight saving and London does, so the length of the window in hours changes twice a year if it is anchored to the London open, and the alignment drifts if it is anchored to a fixed server time. The window has to be defined in a stated local time and converted per date.
Wicks versus closes changes the range's width and therefore both the break level and the stop distance. Wick extremes give a wider range with fewer breaks and wider stops; closing extremes give a narrower range with more breaks and tighter stops. Neither is more correct, and the choice belongs in the stated definition.
How is a break confirmed?
A break is confirmed by a stated condition, a touch of the level, a close beyond it, or a close beyond it by a buffer scaled to volatility, together with a time limit after the open beyond which no entry is taken.
A touch is the most aggressive and admits the most false breaks, since the spread alone can carry price through a level by a pip or two on a quiet morning. A close beyond the level on the entry timeframe filters some of those and delays the entry by a bar. A volatility-scaled buffer filters more and delays further, and on narrow ranges it can consume much of the expected move before entry.
The time limit matters because a break that occurs late in the session is a different event from one that occurs in the first hour. The participation argument applies to the open, and a rule that takes breaks all day is not a London breakout rule but a range breakout rule with a session-defined range. Stating the window inside which entries are valid keeps the rule what it claims to be.
A related choice is whether to take the second break if the first fails. A rule that enters long, is stopped, and then enters short on the opposite break has two trades per day at most and a different profile from a rule that takes only the first.
Why do false breaks decide the result?
False breaks decide the result because a narrow Asian range is broken on a large share of mornings by a move that does not follow through, and the stop for the breakout trade sits at a distance that those failed breaks reach easily.
The mechanism is the same participation the rule relies on. The London open brings orders in both directions, and the first move after the open is frequently a probe of one side of the range before the day's real direction emerges. A rule that enters on the first break is entering on the probe some of the time, and the probe often reverses to the far side of the range, which is where the stop is.
The stop placement choice interacts with this. A stop at the far side of the range is wide relative to the range and is hit by full reversals; a stop at the midpoint is tight and is hit by ordinary noise. Both lose on false breaks, in different amounts and at different frequencies, and the trade-off between them is a parameter to grade rather than to argue.
The reversal variant exists precisely because of this. Fading a break that fails within a stated number of bars is a coherent rule in its own right, and comparing it with the breakout rule on the same history shows which side of the false-break problem has the edge, if either does.
How does QuantParadox test the London breakout?
QuantParadox supplies the Tokyo session's high and low as of the bar and the London session's open as a clock event in London time, so the range, the break, the stop and the entry window can each be stated as a condition and the rule graded across a decade of minute history.
The session times are converted per date in the local time of each market centre, which handles the daylight-saving drift that a fixed-offset definition gets wrong for weeks at a time. The Tokyo session used by the platform is its fixed session window; a trader who defines the Asian range differently states the window as clock bounds instead, which the rule accepts.
The entry-and-stop candle is resolved with finer data where the archive has it, which matters for the midpoint-stop variant, whose stop can be a handful of pips from the entry on a narrow morning. Where the finer data does not settle the order of events, the loss is booked, which is the conservative reading.
The per-instrument breakdown is where this rule tends to be decided. A rule that works on one European pair and not on the others is a finding about that pair, and the out-of-sample split on the pooled result is what says whether the finding is more than one good year.