# Bracket preview — lesson

Gym heading: Sizing the order. Written before the code (Sep 7 2026); the choices revised the same day, after the first harness pass (a run into the entry on half the reps, and one acceptance condition for every case; the rule is unchanged). The teach screen shows this text; the answer key implements the rule in it and nothing else. Distances are in M everywhere. Harness, confidence, persistence, and reveal format follow stop-vs-noise; the bars come from point-at-the-bar's generator v2.

Choices made where the order was silent:

- The stop is Stop vs noise's survivor, placed by rule instead of judged: the last completed swing on the stop's side (one confirming bar each side, on the stop's side of the entry, within the last 20 bars; the extreme of those bars stands in when there is none) plus 0.10 M beyond it, the gym's margin (the zone half-width in Point at the bar, the closed-back margin in Acceptance); or 1.5 × the noise band, the median range of the last 12 bars; whichever is farther.
- The target is exactly half the room, the farthest target Runway allows. The room is the distance from the entry to the one level drawn on the trade's side. The level behind the entry is part of the map and enters nothing.
- Stop and target are read in M at two decimals and R:R is their ratio at two decimals, so the arithmetic on the screen is the arithmetic the rule did. 1.50 clears; 1.49 does not.
- Half the reps clear. Of the half that don't, half sit at 1.20–1.49 (near-misses) and half at 0.60–1.19 (no room). Some of the reps that clear sit at 1.50–1.64, so the line is approached from both sides. Shares in config.
- Half the reps (config) end with a run into the entry in the trade's direction: the last 5 to 9 bars rewritten as a monotone leg 1 to 4 M tall, drawn once per rep and independent of the case, so the last completed swing on the stop's side sits at the leg's origin and the swing, not the noise, sets the stop. Added after the first harness pass: with generator v2's bars as they came, two thirds of the stops were the noise stop at about 1.5 M, and a guess from the room alone was right 81% of the time.
- The regime, the side, the mirroring and the tail run are drawn once per rep and kept across rebuilds, and the bars are accepted or rejected on one condition every case shares (the top of no-room's R:R range must be reachable beyond the window's extreme, and 1.5 inside the largest room drawn), so none of them can carry the case: the stop has the same distribution in Yes and No. The room is set from the intended R:R, so it carries half the arithmetic and a guess from it alone can never be a coin flip; the harness measures how often such a guess would be right (the room alone 67–68%, the spacing between the two levels alone 62–64%, the stop alone 51–52%, on 6,000 reps per family) and caps it.
- The next 20 bars are a continuation drawn with no knowledge of the bracket: a drift with the trade, against it, or none, at equal odds; the drawn levels turn closes back. The reveal shows which of the stop and the target the bars reached, and on what bar. No rate is shown or kept for the screen: what one chart did next is not evidence about the rule.
- Soft clock 10 seconds: two distances to estimate, one to halve, one division.

Persistent header: **"Training your eye for whether the room pays for the risk."**

## Before you enter

- **The bracket** — the stop and the target, drawn before the entry exists. Both are distances from the entry, in M. Before any entry, the bracket is drawn and judged against the room; only then does the read get a vote.
- **The stop** — beyond the last completed swing on the stop's side: 0.10 M below the swing low for a long, 0.10 M above the swing high for a short. And no closer than 1.5 × the noise band, the median range of the last 12 bars. Whichever is farther. This is the stop that survives the noise, placed by rule.
- **The target** — half the room. The room is the distance from the entry to the nearest level on the trade's side; the target sits at half of it, the farthest target that still has runway. Past half, the crowd in front of the level gets paid before you do.
- **R:R** — the target's distance divided by the stop's distance, both in M. Reward over risk. It has no units and no opinion about direction.
- **The rule** — the bracket clears when **R:R is at least 1.5**. Below 1.5 the trade is not taken, no matter how good the read. The panel's verdict line, R:R inadequate, is this rule and nothing else.
- **Why it matters at the trade** — a right read with no room is not a trade. The read says which way. The bracket says whether being right pays enough to cover being wrong at the same size. At 1.5, two wins pay for three losses. At 1.2 the same two wins don't cover them; the room isn't there, and no read fixes that.
- **What it doesn't say** — clearing 1.5 is not a forecast and not a green light. It says the room pays for the risk if the read is right. Whether the read is right is every other module. Whether the level is a magnet or a wall is Magnet or wall. Whether the stop and the swing are what they seem is Stop vs noise.
- **M** — the median high-to-low range of the last 20 bars. Every distance here is in M. The number for the current chart is on the screen.
- **Confidence is a number** — Low means 40%, Medium 65%, High 90% that your call is right. Scored Brier-style: sure and wrong costs 0.81, unsure and wrong costs 0.16, sure and right costs 0.01. Being certain is as expensive as being wrong.

## The read

Two distances, then a division. The stop: the farther of the swing plus its margin and 1.5 × the noise band, in M. The target: half the distance to the level on the trade's side, in M. Then the question: is the second at least one and a half times the first?

## The rep

A chart of 40 bars with a level above and a level below, the last completed swing on the stop's side marked, and a proposed entry at the last close with its side stated. The stop and the target are not drawn: you place them by the rule and judge them. M is stated. The question, in full: **"Long (or short) from the last close. Stop by the rule, target at half the room. Does this bracket clear 1.5?"** Yes or No, with a confidence.

## The reveal

The chart again with the stop and the target drawn and their distances in M; the arithmetic (the swing's distance and 1.5 × the noise band, and which one set the stop; the room and its half; the division); the verdict; one takeaway line; then the next 20 bars, with the first bar to reach the stop and the first to reach the target marked. Templates:

- Clears: **"Stop {stop} M, target {target} M, R:R {rr}. The room pays for the risk."**
- Does not clear: **"Stop {stop} M, target {target} M, R:R {rr}. The read may be right; the room isn't. No trade."**

Every rep ends on one of these two lines. Each is a decision: the order goes in with this bracket, or there is no order.

## Answer key

`stop = max(swingDistance + 0.10 M, 1.5 × median(range of last 12 bars))`, the swing as in Stop vs noise: a low lower than the bar on either side and below the entry (long), or a high higher than the bar on either side and above the entry (short), the most recent one within the last 20 bars; swings on the wrong side of the entry are ignored; otherwise the extreme of those 20 bars. `target = 0.5 × room`, room = the distance from the entry to the nearest level on the trade's side. Both in M at two decimals; `rr = target / stop` at two decimals; `clears = rr >= 1.50`. Half the reps clear (config); of the reps that don't, half sit at 1.20–1.49 on purpose; the reveal shows the numbers either way. The next 20 bars are drawn without knowledge of the bracket and score nothing.