Convexity and Framework Integrity Scoring

Parts 1 through 5 defined what the framework believes and why: regime framing, intellectual lineage, macro thesis formation, tax architecture, and portfolio construction with its governance protocols. Part 6 is the execution kernel. It quantifies, monitors, and enforces the system defined earlier — and it introduces no new rules. Everything here operationalizes rules that already exist, through two scoring mechanisms and a weekly workflow.

The two mechanisms divide the work cleanly. The Convexity Integrity Score quantifies the quality of each position taken on its own. The Framework Integrity Score validates how those positions have been assembled into a portfolio. Every question the framework can ask about implementation resolves to one score or the other, and never to both. Position quality is measured first.

The Convexity Integrity Score is a 0–100 asset-level score answering one question.

CIS is strictly position-level and portfolio-context-free. The same position scores identically regardless of what else is held, which wrapper contains it, or how concentrated the portfolio has become. CIS does not evaluate diversification, correlation, concentration, or tax efficiency. Any portfolio-aware logic belongs to FIS or governance — never to CIS.

The score is built from four components, each answering a different question about the same position.

Weights are engine parameters, read from the framework’s source of truth rather than asserted per session. A thesis profile may shift them, but only within a bound of ±0.10 per component, and every shift is floored, ceilinged, and renormalized so the four weights always sum to 1.0. The four-component structure itself is Doctrine.

What the score deliberately leaves out matters as much as what it includes. Each exclusion has a home elsewhere in the system, and the table below names it.

All subscores are floats, not bucketed tiers. A position scores 78.3, not 75 or 80. Continuity preserves information: a score that can only land on round tiers hides small week-over-week movement, and small movement is what longitudinal tracking exists to see.

Score updates are confidence-weighted, and each update is clamped to a bounded move: low confidence permits ±3 points, medium ±5, high ±8 per update, and evidence derived through a proxy rather than observed directly permits ±6. Two situations warrant wider bounds, because a baseline is being established rather than adjusted — initial scoring permits ±20, and a material thesis change permits ±15. Structural rescores — a change in the scoring model’s version — and large corrections of model divergence may legally bypass the clamp altogether, so that a genuine re-basing is never suppressed by machinery built to resist mood.

Why clamp at all? The reason is behavioral as much as mathematical. Unclamped scores oscillate with mood and headline flow, and a single enthusiastic week could otherwise rewrite months of accumulated conviction in one sitting. The clamp forces large re-ratings to be earned across successive updates, each requiring fresh evidence.

Confidence is not asserted by hand: the engine derives it from the quality of the evidence source behind each component and aggregates the result, with structural caps by venue and size — over-the-counter listings cap at low confidence, microcaps at medium. The derivation keeps CIS decision-support rather than autonomous judgment: the score declares how sure the evidence permits it to be, and the practitioner decides what to do about it.

Structurally different assets cannot be scored honestly through a single lens, so CIS applies an archetype before it applies the rubric. The engine labels four structural archetypes — spot Bitcoin, ETF, levered ETF, and operating equity, the default — and routes scoring itself through six lanes: Bitcoin, crypto, ETF, equity, private-structural, and not-scorable. The archetype is labeling; the route is the lens.

Bitcoin routes to a monetary-TAM path, because that is what the asset is: a scarcity claim on a monetary market, not a business with a balance sheet. ETFs score the exposure thesis with reduced optionality, since a diversified wrapper inherits its theme’s upside but dilutes any single holder’s right tail.

What the archetype does not do is put a floor under the number: CIS is a pure weighted sum of its four components, with no post-aggregation floor of any kind, and there is no cash-like archetype — an asset held for stability is governed as Ballast by Part 5’s posture rules, not scored on a separate rubric here. The archetype defines the scoring lens for a position’s intrinsic characteristics — never for portfolio context or relative allocation, which remain out of scope for CIS entirely.

Scores resolve into four bands, labeled numerically by the engine and named here for reading. Seventy and above is Strong — the band carrying the accent; 60–69 is Moderate, rendered as monitor-gray in the live register; 50–59 is Caution; below 50 is Weak.

This register is shared: in the live engine, CIS and FIS read these same four bands. On FIS, 70 is the action line; on CIS, it is the Strong band’s floor — a band boundary, not an action trigger. That is the position half of the kernel. The other half scores the assembly.

A collection of high-CIS positions concentrated in a single sector, held in the wrong wrappers, carried without documented theses, is a poorly constructed portfolio regardless of individual quality. The Framework Integrity Score exists to catch exactly these failures. It answers: how well is the portfolio assembled relative to framework rules?

FIS does not grade the portfolio up from zero on vague notions of quality; it starts the assembled portfolio at 100 and deducts a named penalty for each violation it finds. The direction matters. A score built by addition invites argument about what deserves credit. A score built by subtraction indicts: every point below 100 is attributable to a specific, named departure from a specific rule. Fix the penalty → recover the points. The score is therefore not a verdict but a repair list — FIS never restates the rules it enforces, it measures compliance with rules defined in the referenced Parts.

Value-weighting is scoped precisely: governance and dead-capital penalties scale with position size, with a 0.2 percent floor and a 12 percent cap — larger positions generate proportionally larger penalties, but the cap prevents any single position from dominating the score — while allocation, concentration, and complexity contributions are flat.

Penalties land in five named buckets. The mechanism is easiest to see run once: the exhibit below walks an example portfolio down from 100 through five deductions, one per bucket, to the score it earns — 83, in the Strong band.

Deductions in the exhibit are illustrative; the buckets, their caps, and what each one detects are canonical — this is the live engine’s register.

Because the caps sum to eighty, FIS cannot fall below twenty — even a portfolio that violates everything at once retains an attributable floor.

Two accounting notes preserve continuity with earlier Parts. Tax architecture is no longer a standalone bucket: wrapper placement is scored inside Allocation as wrapper drift, so the Part 4 rules still bind — they simply bill through a different line. And posture drift is not a score bucket at all: the engine reports it as metadata alongside the score, and a separate live diagnostics surface warns when posture deviates more than 10 percent from target and fails beyond 20 percent.

One score for the position, one for the assembly, both read against the same register. The diagnosis comes from reading them together.

CIS and FIS are intentionally independent. A 95 CIS position in the wrong wrapper still generates FIS penalties; a perfectly constructed portfolio of 65-CIS positions may be soundly assembled but remains sub-core in conviction. FIS cannot increase position quality, and CIS cannot repair construction. Each score identifies a different dimension, and each requires its own response. The dividing line on both axes is the shared band line at 70 — the Strong band’s floor on each score, and on FIS also the action threshold. Crossing the two scores at that line yields four states, and the framework prescribes a specific response — and a specific repair order — for each.

CIS determines position sizing within the bands established by Part 5. It does not determine those bands — Part 5 governance does. Part 5’s Torque, Ballast, and Hype sizing tables carry the full CIS-to-allocation mappings by posture. The relationship: higher CIS earns a larger allocation within posture limits — 70 or above core sizing, 60 to 69 constrained, 50 to 59 starter, below 50 none. Exits are governed by posture rules, tripwires, and thesis evidence, not by the score band alone. Three signals govern the response to score movement.

Signals only matter if something runs them on schedule. The framework fixes the schedule.

The framework operates on a weekly cadence with event-triggered updates, executed Sunday evening or Monday pre-market. The live governance engine runs exactly this sequence each pass: ingest current state → CIS updates → FIS calculation → governance checks → action determination → decision logging. The loop closes weekly whether or not anything traded.

The CIS pass computes C, R, M, and E for each position under its archetype lens, clamps the delta, and logs the result. The FIS pass calculates penalties across the five buckets, value-weights the governance and dead-capital charges, and generates the attribution: FIS equals 100 minus the sum of penalties. The governance gates then check earnings proximity (T-5, positions above 3 percent), the live momentum gates — the 200-day and 50-day trend checks and the RSI band from Part 5, standing in for the doctrine’s three dimensions — tripwires, and posture drift.

Action determination follows fixed precedence: a tripwire demands immediate response; an earnings window trims to 3 percent; FIS below 70 fixes the top penalty; CIS drift beyond 10 resizes; no trigger means hold. Decision logging closes the pass, and the evidence surfaces are live: the decision log retains a rolling year of entries, and the TapeRead journal (the dashboard’s trade-notes record) and the governance audit page carry the qualitative record.

Four events interrupt the cadence rather than wait for it: a position moving more than 5 percent in a single day · material news · an earnings announcement · a tripwire condition met.

Between triggers, frequency limits keep the practitioner from confusing activity with diligence.

The cadence earns its keep less in the weeks it acts than in what happens when it stops running at all.

Implementation quality rarely fails loudly. It degrades through patterns that produce no obvious signal — a score nobody re-ran, an allocation the market moved without a single trade, a tax inefficiency too small to notice in any given month. Watch what those patterns do to a healthy portfolio in the year after reviews stop.

Five patterns account for most of the erosion. Each has a canonical diagnostic — a threshold that turns a slow, invisible failure into an explicit flag — and all five diagnostics run live in the dashboard’s longitudinal layer.

Without objective recalculation, narrative momentum substitutes for current analysis. That is why framework health is longitudinal, not episodic: every one of these failures still rounds to healthy at any single point in time, and becomes visible only through week-over-week tracking. What remains is the question of what carries that tracking in practice.

Parts 1 through 6 form a complete system. The framework governs behavior; practitioners must execute. The weekly burden is concrete: CIS review for every position, FIS recalculation after portfolio changes, earnings calendar monitoring, momentum tracking, posture drift detection, tripwire evaluation, and longitudinal tracking. None of it is individually difficult; all of it must actually happen, every week, for the measurements to mean anything.

Three implementation paths carry that burden.

Choose the implementation that makes systematic execution sustainable over the relevant horizon; the framework does not care which one, only that the loop actually runs.