# Value migration — lesson

Gym heading: Situational awareness. Written before the code (Sep 8 2026), from the trader's own framework in the Aug 30 2026 transcript. 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; each session's profile is built the way the Value module builds it (rows of 0.1 M, the 70% value area around the POC), on bars that carry the session VWAP module's synthetic volume.

Choices made where the order was silent:

- Five sessions of 40 bars each (15-minute bars from the 18:00 New York anchor, so the volume's time-of-day profile is the VWAP module's, unchanged), oldest on the left, plus a sixth built for the reveal and shown only there. All five are drawn at one price scale, so a higher profile is a higher price. Nothing on the chart carries a label: no prices, no dates, no session names, no axis.
- Each session's profile is its own: rows of 0.1 M anchored at the middle of that session's range, each bar's volume spread evenly over its high-to-low range, the POC the heaviest row, the value area the 70% of that session's volume built outward from the POC one row at a time, the heavier neighbour first. VAH is the top of the highest value row, VAL the bottom of the lowest. The profile is drawn in full: the value rows in blue, the POC row in yellow, the rows outside value in grey. The grey is the session's extremes, and the extremes are not the read.
- M is the median high-to-low range of the last 20 bars of the fifth session, the one that has just completed. Every shift is in that M.
- "The value area moves higher" between two sessions means both edges move: the later session's VAL is above the earlier one's VAL and its VAH is above the earlier one's VAH, read at two decimals of M. Lower is the mirror. A session whose area widens over the previous one without lifting both edges has not moved value.
- Balance requires measured overlap across all five sessions: the intersection from the highest VAL to the lowest VAH must cover at least 50% of the widest value area's width, with that percentage read at two decimals. This is the drill's definition of substantial overlap, not a universal market threshold. A failed trend without that shared band is mixed; the generator rejects it and rebuilds. Mixed is not a fourth answer and is separate from a user's uncertain read.
- The three answers come up a third each (config). Half the charts are mirrored. The extremes are drawn without looking at the case: every session probes beyond its balance on both sides, to distances from one range for every case. In at least a quarter of the UP and DOWN reps (config) one session's high or low steps against the migration on purpose, so the answer cannot be read off the tops and bottoms of the profiles. In a share of the BALANCE reps the extremes drift one way while value stays put.
- The sixth session is built with an outcome rate that is a config constant and is never shown. The reveal says where its POC landed against the fifth's, in M, and whether its value area moved. No percentages anywhere.
- Soft clock 8 seconds: five value areas, one direction.

Persistent header: **"Training your eye for where value has been going, before you look at a candle."**

## Before you look

- **Value area** — The band where most of a session's volume traded: from VAL to VAH, with the POC as the heaviest price. The profile counts it, each bar's volume spread over its range in rows of 0.1 M; the value area is the 70% of the session's volume around the POC, built outward one row at a time, the heavier neighbour first.
- **Value migration** — Each session builds a value area — the band where most of its volume traded. Put four or five sessions side by side and the question is not where value is, but where it has been going. Each session's value area higher than the last is the market accepting higher prices: value up. Each one lower is the market repricing lower: value down. Value areas landing in the same band session after session is balance, and balance is the market saying it has found agreement for now.
- **Why it matters at the trade** — Decision: this sets the path of least resistance for the day, before any level or any candle. In a value-down market a rally into premium is where shorts have location. In a value-up market a pullback into discount is where longs do. In balance, neither edge is a trend trade — the edges are fades and the middle is nothing. This is not a prediction of direction. It is which side of the market you are willing to be on today.
- **The rule** — Compare all five sessions, not only the most recent pair. **Value up**: at least 3 of the 4 transitions move the value area higher (VAL and VAH both above the previous session's), and the fifth session's VAL is above the first session's POC. **Value down**: the mirror. One counter-step is allowed. **Balance**: neither trend rule passes, and the price band shared by all five value areas is at least 50% of the widest value area's width, with that percentage read at two decimals. This is the drill's definition of substantial overlap; individual POCs can still move. Sequences that meet none of these conditions are mixed and are excluded from this three-answer drill.
- **What it doesn't say** — The extremes are not value. A session's high or low is where price was rejected, not where it was accepted; a higher high with value sitting still is a probe, and a lower low under a rising value area is a pullback. Read the blue, not the grey. Where the last close sits against today's value is the Value module; whether price is being accepted out there is Acceptance. This says where value has been going, and so which side you are willing to be on.
- **M** — The median high-to-low range of the last 20 bars of the fifth session. 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

Five value areas, left to right. Are they stepping up, stepping down, or landing in the same band?

## The rep

Five session profiles side by side, oldest on the left, each with its value area shaded and its POC marked. No prices, no labels, no dates. Frozen; the fifth session is complete. M is stated. The question, in full: **"Five sessions, oldest on the left. What is value doing?"** Up, Down, or Balance, with a confidence.

## The reveal

The chart again with the first session's POC drawn across and the five POCs joined; the five POCs listed with the shift between each in M; the transitions counted; for trends, the fifth session's VAL (or VAH) against the first session's POC in M; for balance, the common band's width in M, the widest area's width in M, and the overlap percentage; the verdict; and one takeaway line. Templates, N the absolute shift from the first POC to the fifth in M and k the number of transitions in the migration direction:

- Up: **"Value migrated up across five sessions: {k} of four transitions moved higher, and the fifth value area is above the first POC. The first-to-fifth POC shift is {N} M higher."**
- Down: **"Value migrated down across five sessions: {k} of four transitions moved lower, and the fifth value area is below the first POC. The first-to-fifth POC shift is {N} M lower."**
- Balance: **"The five value areas share a substantial price band. Neither migration rule passes; this is balance across all five sessions, even though individual POCs can move."**

Every graded rep ends on one of these lines, describing the five visible sessions. Then the sixth session's profile is revealed, with where its POC landed against the fifth's in M. The synthetic continuation does not determine correctness. No win percentage is shown anywhere.

## Answer key

For each of the five sessions, the profile: each bar's volume spread evenly over its high-to-low range into rows of 0.1 M anchored at the middle of that session's range. POC: the row with the most volume. Value area: from the POC outward, one row at a time, the heavier neighbour first, until 70% of the session's volume is in. VAH: the top of the highest value row; VAL: the bottom of the lowest.

Four transitions, session 1 to 2 through session 4 to 5. A transition is `up` when `round2((VAL[i+1] − VAL[i]) / M) > 0` and `round2((VAH[i+1] − VAH[i]) / M) > 0`; `down` when both are `< 0`; `flat` otherwise.

`UP = (ups ≥ 3) AND round2((VAL[5] − POC[1]) / M) > 0`; `DOWN = (downs ≥ 3) AND round2((POC[1] − VAH[5]) / M) > 0`. These conditions retain priority.

`commonWidth = max(0, min(VAH[1..5]) − max(VAL[1..5]))`; `widestWidth = max(VAH[i] − VAL[i])`; `commonOverlapPct = round2(100 × commonWidth / widestWidth)`. `BALANCE = NOT UP AND NOT DOWN AND commonOverlapPct ≥ 50`. Otherwise the result is `mixed`, rejected by the generator. Touching edges have zero overlap. Adjacent overlap alone does not establish a common five-session band. The percentage uses unrounded widths; displayed M widths are rounded separately.

Reported: the five POCs and their shifts in M, the transitions, the fifth session's edge against the first POC for trends, or the shared band, widest area, and overlap percentage for balance. Saved rows and logs include the rule revision and overlap evidence.

The three answers come up a third each (config). The extremes carry nothing the value areas do not: session highs, lows and ranges overlap across the cases, and at least a quarter of the UP and DOWN reps hold a session whose high or low steps against the migration.