Session Range Breakout: Morning Box, Asian Range and ORB Tested on Gold

A session range breakout marks a box from the high and low of a time window, such as the early morning or the Asian session, then waits for price to break out of it: a break up means Buy, a break down means Sell. It’s popular because the rules are clear and repeat every day. This article explains how to set the box, the entry, the stop loss and the target, the common variants, a step-by-step lot size calculation, and what we found when we tested it on almost five years of gold data. We did not find a clear edge.

Types of session range breakout

They all share one idea: a quiet stretch of the market builds a box, and the hope is that when more traders arrive, price picks a direction and keeps going.

• Asian range → London breakout uses the high and low of the Asian session and waits for a break when London opens, around 14:00 (during UK summer time) or 15:00 Thai time (UTC+7). Session times are in Forex market hours in Thai time
• Morning box uses a short early window in Thai time, such as the first 30 minutes or two hours of the trading day
• Opening range breakout (ORB) comes from stocks and futures and uses the first 5, 15, 30 or 60 minutes after the open. Toby Crabel wrote about it in a 1990 book. Forex and gold have no single open, so traders choose a clock time

The basic rules: box, entry, stop and target

1. Box Pick a fixed window, say 06:30–07:00, and record the highest high and lowest low of that window.
2. Entry Once the box closes, place a Buy stop above the high and a Sell stop below the low (orders that trigger when price reaches them, see order types explained). Many traders add a small buffer, around one to two spreads, so a mere touch doesn’t trigger them.
3. Stop loss The two common choices are the far side of the box or the middle of the box.
4. Target Usually a multiple of the box height, such as 1 or 1.5 times, or no target and a time exit.
5. Expiry If there’s no break by a set time, cancel both orders. If a trade is open, close it at a set time, for example before the day ends.

Common variants

• OCO (One Cancels the Other) Place the Buy stop and Sell stop together. Whichever fills first cancels the other, so a choppy day can’t hit you both ways. MT5 has no standard OCO button, so you delete the other order yourself or use an EA
• Box size filter with ATR Compare the box height with the daily ATR. If the box is too narrow, spread eats a big share of the stop. If it’s too wide, the stop is so far that the target is rarely reached. There is no standard threshold, so set your own before testing
• Time exit Don’t hold overnight, or close at a set time even if neither target nor stop has been hit
• Wait for a close outside the box Enter when a candle closes outside the box instead of using pending orders. Fewer triggers on a mere touch, but a worse entry price

A session box is a close relative of the Donchian Channel. The difference is that Donchian uses the last N bars and keeps sliding, while a session box is pinned to the clock.

Worked example: a morning box on gold and the lot size

Say the account is 5,000 dollars and you risk 1% per trade = 50 dollars. On gold, 1 lot = 100 ounces, so a 1-dollar price move = 100 dollars per lot (check the contract size in your broker’s specification).

1. Box 06:30–07:00 Thai time (UTC+7): high 3,012.40, low 3,006.80. Height = 3,012.40 − 3,006.80 = 5.60 dollars
2. Add a 0.20-dollar buffer on each side: Buy stop = 3,012.60, Sell stop = 3,006.60
3. Target 1.5 times the box height = 1.5 × 5.60 = 8.40 dollars. TP for the Buy = 3,012.60 + 8.40 = 3,021.00

	Stop at the far side	Stop at the middle
Stop loss level	3,006.60	3,009.60
Stop distance	6.00 dollars	3.00 dollars
Lot = 50 ÷ (distance × 100)	0.0833 → 0.08	0.1667 → 0.16
Loss if stopped	0.08 × 100 × 6.00 = 48	0.16 × 100 × 3.00 = 48
Gain at TP (8.40)	0.08 × 100 × 8.40 = 67.20	0.16 × 100 × 8.40 = 134.40
RR	8.40 ÷ 6.00 = 1.4	8.40 ÷ 3.00 = 2.8
Break-even win rate 1 ÷ (1 + RR)	41.7%	26.3%
Spread cost (assume 0.30 dollars)	0.08 × 100 × 0.30 = 2.40	0.16 × 100 × 0.30 = 4.80

Always round the lot down, so the real risk is 48 dollars, not 50. If the Sell stop fills instead, use the same numbers in reverse: stop at 3,012.60 (or the middle), TP at 3,006.60 − 8.40 = 2,998.20.

The middle stop doubles the RR on paper, but it isn’t free. The lot doubles, so spread per trade becomes 4.80, or 10% of the risk instead of 5% with the far-side stop, and the shorter distance gets hit by normal noise more easily. On days when the box is wider, the lot has to shrink. Recalculate every day with the lot size calculator or see position size from stop loss.

Our tests: does the morning box tell direction?

In our backtests on gold (XAUUSD) with M1 price bars from 2022 to 2026, about 1,000 days, we drew a morning box from 06:30 to 08:30 Thai time and asked whether what happened inside it helped predict the direction afterwards.

The result that looks sharp: on days when price closed the box at its top edge, price broke the top before the bottom 79% of the time. That sounds like a signal, until we drew an afternoon box from 13:30 to 15:30 with the same rules as a control and got 76%, almost the same.

That number is geometry, not foresight. If price moves randomly with no direction, the chance of touching the top first = distance to the bottom ÷ box height. Price sitting 20% of the box height below the top touches the top first about 1 − 0.20 = 80% of the time, with no edge at all.

The result you could actually trade: measured symmetrically, from the 09:00 price, did price travel 5 dollars in the direction the morning-box condition pointed before travelling 5 dollars the other way? Overall 51.3%, with individual conditions at roughly 47–56%. A coin flip.

A second study used M1 bars from January 2022 to August 2026, 1,184 days, and asked whether the price structure at 09:00 could predict direction to 14:00, 20:00 and 24:00.

• The best of several dozen conditions called the direction to 20:00 correctly 59% of the time over 83 days. That is only about 1.6 standard deviations from 50%, and it flipped by year: 70% in 2026, 53% in 2022
• “Price at the top of the 07:00–09:00 range → keeps rising” was right to 20:00 on 53.6% of 289 days, below the rate at which price rose on all days in the same period (54.6%), which sits above 50% because gold was in an uptrend during the test. The mirror, “at the bottom → keeps falling”, scored 46.8% over 235 days

Our tests: one OCO trade a day

Setup: gold, one trade a day, box 06:30–07:00 Thai time, OCO stop orders around the box waiting until 07:30, no break means no trade that day. Fixed stop loss 5 dollars, TP 30 dollars (RR 1:6, break-even win rate before costs 14.3%). Any open trade closes at 06:00 the next day. Lot 0.10, starting balance 10,000 dollars. 2022–2025 use one-minute price data; January to September 2026 uses real tick data.

Five years total +4,698 dollars, positive in 4 of 5 years, with the largest drop from an equity peak about 2,170 dollars in 2026. It looks good until you put it next to the controls.

• SE (standard error) for one year of about 250 trades at RR 1:6 is about ±2,000 dollars, and about ±4,500 dollars for the 5-year total. A system with no edge at all can easily total +4,698
• The simplest control is Buy every day with the same stop and target. Entering at 06:30 made +2,228. Entering at 07:00 made +5,408, more than the OCO. Sell every day at 06:30 made −4,925. Most of the gap between Buy and Sell comes from gold’s uptrend over the period
• The OCO beat both the Buy and the Sell control in the same year in only 2 of 5 years (2022 and 2023)
• Changing the target to 25 dollars (RR 1:5) drops the total to +2,684. The result moves a lot with a small parameter change

A good-looking profit has to beat the dumbest control first. If buying every day earns about the same, the box adds nothing.

Limitation: 2022–2025 use one-minute data, which can’t see the order of prices inside each minute or short spread spikes. With a 5-dollar stop, those years probably look better than reality. 2026, on real ticks, is the most trustworthy year. More in tick data quality in MT5. These are all backtest results. Past results do not predict future results, and they are not returns you should expect.

The daily rollover eats tight stops

In the test account, 07:00 Thai time was the server’s daily rollover (brokers set this differently). On real ticks in 2026, entering at 07:00 was worse than 06:30 in every direction: Buy every day −962 versus −150, Sell every day −2,364 versus +751 dollars. If this were about price direction, one side should improve when the other gets worse, but both got worse. The likely cause is spread widening right after the rollover and touching a tight stop.

Our 06:30–07:00 box ends exactly at that moment, and the pending orders wait from 07:00 to 07:30. If you use a morning box in Thai time, check your broker’s rollover time first. Details are in Forex market hours in Thai time and spread cost per trade.

Why controls matter

The morning box is an idea you see on charts again and again. The days that broke out and ran are easy to remember; the false breaks are easy to forget. That is hindsight bias. Controls separate an effect of the box itself from things that came along with it.

• A box at another time of day with the same rules separates the effect of time from the effect of geometry (79% vs 76%)
• Buy every day and Sell every day separate the effect of the entry from the market’s trend
• The ±2 SE band tells you whether a result is far enough from zero not to be luck
• Split by year, and test on data you didn’t use to choose settings, as in walk-forward testing, or reshuffle the trade order with Monte Carlo

Limits of our work: one market, a few box definitions and fixed rules. Other windows, markets or filters may behave differently, but they have to pass the same controls.

When the box misleads

• Big news days Price can break both sides within minutes. OCO stops you getting hit twice, but the first trade can still be stopped, possibly with slippage
• Thin early-morning markets A very narrow box breaks on a tiny push, then price slides back in
• An unusually wide box after news or a gap: the stop is far, the lot small, and the target may not be reached within the day
• Rollover and daylight-saving changes shift what a fixed Thai-time box lines up with in the market

Combining it with risk management

• Calculate the lot from that day’s stop distance every time. Don’t keep the same lot when box heights differ
• High RR means long losing streaks. At RR 1:6, winning about 1 trade in 7 breaks even, but when we simulated 250 trades at a 1-in-7 win chance 5,000 times, the median longest losing streak was 25 trades. See losing streaks are normal
• Always compare the win rate with the break-even win rate after spread. See win rate vs break-even

Test your box against a box at another time and against buying every day before you believe that time is special.

Common mistakes

• No control Seeing a number like 79% and believing it without testing a box at another time with the same rules
• Choosing the box window after seeing results Trying many windows and keeping the best one, which is usually the luckiest
• The same lot every day so wide-box days carry several times the risk of narrow-box days
• Forgetting to cancel the other order when there’s no built-in OCO, so a second trade fills when price swings back
• Ignoring the rollover Placing the box or pending orders across the wide-spread period and getting stopped on a tight stop
• Trusting one-minute data with a tight stop. That kind of data usually makes results look better than reality