PORTFOLIO CONSTRUCTION INTERMEDIATE · LESSON 14 / 24 ~6 min read

Sector rotation: where the next dollar goes.

Picture a dinner plate. You can't eat five servings of mashed potatoes and call it a balanced meal — you'd be sick by the third one. Same with sectors. A portfolio with 60% in tech is not "diversified across ten names." It's a concentrated tech bet wearing the costume of diversification. The framework caps each sector at roughly 25-30% of risk-weighted exposure, and when the next high-conviction candidate is in a sector that's already at cap, the answer isn't "yes, but smaller." The answer is "no, find a different sector — or sit on hands until something rotates out."

What "diversification" actually requires

Most retail traders treat diversification as a count problem. Twenty names = diversified. The framework treats it as a correlation problem. Twenty names that all move together when SPY drops 3% is one position dressed as twenty. Two names in two genuinely uncorrelated sectors is more diversified than ten names spread across one mega-sector.

The mechanical rule the framework enforces: no single GICS sector above ~25-30% of risk-weighted exposure. Risk-weighted, not dollar-weighted — a 15% Sports Cars position with a 6% stop carries more risk than a 30% Anchor position with a 2% stop. The cap counts the risk, not the dollars.

The sector cap math

For each open position, risk dollars are shares × |entry − stop| (Lesson 5). Sum those across all positions in a given sector and divide by total portfolio risk-dollars. That's the sector's risk-weighted concentration. The default cap:

sector_risk_pct = Σ(shares × |entry − stop|) per sector
                ÷ Σ(shares × |entry − stop|) total

default cap: 25-30%  (configurable via SECTOR_CAP_PCT)

Above the cap, the broker pre-flight chain refuses new entries in that sector at preview time. The error message names the sector, the current concentration, and the cap. Override exists; like every other override, it gets logged.

Rotate-IN and Rotate-OUT

The dashboard's Sectors card surfaces two specific kinds of chip:

Rotate-OUT chip. A sector currently at or above cap with at least one position weak enough to be a candidate for trim or full exit (Grade C/D, R:R degraded, structure broken). The chip names the position(s) that are pulling the sector concentration up — those are the candidates for the rotation.
Rotate-IN chip. A sector currently below cap with at least one watchlist candidate at Grade A and 13 pillars green. Where the next dollar can go without violating the cap.

The two chips together form a directional read: trim the weak position in the over-cap sector, redeploy into the strong candidate in the under-cap sector. The framework doesn't auto-execute the rotation — it surfaces the trade. The trader makes the call. But the rotation is rarely controversial when the data lines up: a Grade D tech name in a 32% tech sleeve is exactly the position that should free up dollars for a Grade A energy name in a 6% energy sleeve.

⌬ Sector cap checker

Tech 32% over cap

Healthcare 14% room

Financials 12% room

Energy 6% room

Consumer 9% room

Industrials 7% room

Total allocated80%

Sectors over cap1 (Tech)

VerdictRotate-OUT trigger

Tech is at 32% — above the 30% cap. New tech entries refused at preview. Energy (6%), Industrials (7%), Consumer (9%), Financials (12%), Healthcare (14%) all have room. The framework would surface a Rotate-OUT chip on the weakest tech name and Rotate-IN candidates from the lowest-concentration sectors with watchlist passes.

Drag any sector — the verdict updates. Cap default is 30%; over the line, that sector turns red and refuses new entries. The framework's actual cards do the same with risk-weighted dollars; the slider here just uses % of capital for clarity.

Why the cap is 25-30%, not "minimize concentration"

The cap isn't "be as diversified as possible" — that strategy converges on owning the index, which the framework already loses to most retail traders by virtue of fees and over-trading. The cap is "concentrate where the conviction is, but not in only one sector." Three Grade A Sports Cars in semiconductors is allowed up to ~25-30% of risk; that fourth Grade A semi name pushes the sleeve over cap and gets refused even if it's the strongest setup on the list.

The reason: when the semiconductor cycle turns (and it always does), three positions move as one. Stop-out on one is usually stop-out on all. A 30% sector cap means a sector-wide 25% pullback costs roughly 7.5% portfolio max — survivable. A 60% sector cap turns the same pullback into 15% — doom-loop territory.

Sector classification — one note

GICS sector classification has edge cases. NVDA is "Information Technology" but its revenue is increasingly enterprise infrastructure with macro sensitivity that looks more like industrials. AAPL is "Information Technology" but consumer-discretionary in nearly every operational read. The framework uses GICS as the default but allows manual reclassification when the operational truth diverges. Two positions classified GICS-tech that actually move on different macro inputs aren't really one sector for risk purposes — but the default starts conservative.

The rotation playbook

When the Sectors card shows a Rotate-OUT/Rotate-IN pair, the disciplined sequence:

Verify the weak position is actually weak, not just temporarily underperforming. Grade C+ with R:R degraded and structure broken — that's weak. Grade B with a normal pullback isn't.
Verify the strong candidate has actually passed all 13 pillars, not just the surface ones. Audit grade, R:R floor, sector concentration in target band, macro tape friendly, sovereignty cap room, drawdown gate clear.
Execute the trim first, the entry second. Order matters: closing the weak position frees risk-budget for the new entry. If you do it backwards and the new entry fills before the old one closes, you're temporarily over both sleeve and sector caps.
Log it as a rotation in the journal — not as two unrelated trades. The performance attribution depends on knowing the trim and the new entry are linked.

What if the rotation has nowhere to go

Some weeks the watchlist returns no Rotate-IN candidates — every sector below cap has nothing passing the 13 pillars. The honest answer is: don't rotate. Trim the weak position to free risk-budget, send the dollars to HCF (Lesson 13's Hard Cash Floor), and wait. The Sports Cars sleeve dropping from 47% to 41% is fine; the HCF growing from 32% to 38% is fine. Sitting on more cash than usual until a strong setup appears is a feature of the framework, not a bug. Capital that isn't deployed isn't earning return; capital that's deployed badly is losing return. The math prefers the first.

The real lesson

Diversification isn't "lots of stocks." It's "lots of stocks across uncorrelated sectors with no single sector at cap." The Sectors card's Rotate-OUT and Rotate-IN chips are the framework's mechanical version of that rule — surfacing exactly when one sector is too crowded and pointing at the specific underperformer that should free up the dollars. The rotation isn't a market-timing call. It's a concentration-management call. Run it weekly during the Friday close ritual; let the chips fire when they fire; don't force rotations that don't have a destination.