ORB Risk Management: Stops, Sizing, and Daily Loss Limits

The box is the risk model

The opening range is the high and low of the first 15 one-minute bars after the fixed session open. The clock draws it, not you: Sydney 22:00, Asia 00:00, London 07:00, New York 13:30 — all GMT, identical every day. Once minute 15 closes, the box is fixed and never redrawn. Consequence: your stop exists before your entry does, and both are numbers, not opinions.

Entry is the close of the confirming M1 bar: a bar that closes beyond the range — wicks do not count — with tick volume at or above 1.5x the average of the 20 bars immediately before it. No volume confirmation, no trade. Stop is the opposite side of the box: OR low for longs, OR high for shorts. The alert embed prints entry, stop, 1R in points, Target 1 and Target 2 before you touch anything. The only variable left is size. Size is arithmetic, not judgment.

Define 1R before entry

1R = |entry − stop|, where entry is the confirming bar's close and stop is the far side of the range. Worked example, New York session on US100:

• 13:30–13:45 GMT opening range: high 21,540, low 21,496. Range = 44 points.
• 13:52 GMT: an M1 bar closes at 21,548 — above the OR high — on volume 1.8x the prior 20-bar average. Long confirmed. Both conditions on the same closed bar; a wick through 21,540 at 13:47 confirmed nothing.
• Entry 21,548. Stop 21,496 (the OR low). 1R = 52 points.
• Target 1 = 21,600 (entry + 1R). Target 2 = 21,652 (entry + 2R).

1R is wider than the range itself because the confirming close sits above the box. Normal. It is also why every target gets expressed in R before entry: a wall 104 points away is 2R here — a trade. The same wall on a day where 1R is 200 points is 0.5R — a skip. No R math, no trade.

Fixed fractional sizing

Risk the same fraction of the account on every signal, win or lose, session after session. Default 0.5%, ceiling 1%. The formula:

Lots = (account × risk%) ÷ (stop distance in points × point value per lot)

Always round down. Rounding up means risking more than you decided to risk, which deletes the point of deciding. On a $10,000 account at 0.5%, the risk budget is $50 per signal — every signal, every session:

Note the US100 row: 1.0 lot would risk $52, which is over budget, so the size is 0.9 — not 1.0. Point values vary by broker; check your contract specs and run the numbers through the position size calculator before the session opens, not while the bar is closing. Two rules that never bend:

• Wider stop → smaller size. A Sydney signal on a thin night can carry a wide 1R relative to the follow-through on offer. Size down. The dollar risk stays $50 whether 1R is 20 points or 200.
• Never tighten the stop to justify more size. The stop sits at the opposite side of the range because that is where the breakout is factually wrong. Pulling it inside the box converts a structural stop into a wish with a price tag.

Stop placement: the opposite side, or a buffered version of it

Default: stop exactly at the opposite side of the opening range. A confirmed close back through the entire box kills the breakout premise — there is nothing left to defend past that line. Two adjustments are acceptable, both in one direction only: wider.

• Spread and noise buffer. A few points beyond the OR line, so a wick-tag of the exact level plus spread does not eject you from a trade that never actually invalidated. The buffer widens 1R, so it also shrinks size. Recompute the lots; do not eyeball them.
• Nearest mapped level. Premium context from the ORB Flow Engine sometimes shows a structural level — a put wall under a long, a call wall over a short — sitting just beyond the opposite OR side. Parking the stop beyond that level is legitimate. It is also a wider stop: smaller size, same $50. The mapped level moves the stop; it never generates the signal.

Never acceptable: a stop inside the box, a mental stop, or adding to a position after price has closed back inside the range. Each is the same mistake — arguing with the level that defined the trade.

Per-session and daily loss limits

The engine already caps frequency. Each session fires at most one long and one short per day: once a side breaks, it latches and cannot fire again. Four sessions, two sides each — eight signals is the theoretical daily maximum, and real days produce far fewer. There is no "one more setup" to hunt after a loser. The trade budget is built into the code.

Layer personal limits on top of the latch, denominated in R so they scale with your sizing automatically:

The arithmetic of that cap: at 0.5% per trade, a maximum −2R day costs 1% of the account. Twenty consecutive maximum-loss days — a sequence the one-long-one-short latch makes nearly impossible to assemble — draws down about 18%. Painful, recoverable, survivable. That is the design goal: no single day, and no realistic bad week, gets to decide the account.

Skips are part of the same budget. A breakout aimed straight into a nearby gamma wall gets skipped. A setup against the morning flow bias is graded weak and can be passed without apology. A pass costs exactly 0R, and 0R is frequently the best result available in a session.

Why the box enforces restraint

Most risk plans fail at the moment they depend on the trader choosing, mid-session, to stop. This one removes most of those moments. The range is fixed by the GMT clock, so there is no redrawing a level to manufacture a setup. Confirmation requires a closed M1 bar beyond the range plus the 1.5x volume filter, so there is no jumping a wick. Each side latches after it fires, so there is no re-entering the same breakout four times. The stop is the far side of the box, so there is no mid-trade debate about where invalidation lives. What the machine leaves to you is exactly three decisions: the fixed fraction, the rounding-down arithmetic, and standing up at −2R.

Then log it. Every trade: session, setup grade, planned R, realized R. Expectancy — (win% × average win) − (loss% × average loss) — is the only number that says whether the plan works, and it can only be computed from a complete journal. The lab holds itself to the same rule: every signal the engine fires, losers included, is posted openly in the results channel.