Part A — the portfolio for one investor, derived

as of 2026-08-23. Working analysis behind /portfolio. Not personalised advice; every figure is a function of stated inputs.

The investor. Roughly equal nominal thirds in a Roth, a traditional tax-deferred account and a taxable brokerage account. Multi-decade horizon. Contributing 5–15%/yr. Wants ETFs simple to hold. Willing to run a rule-based strategy only with real understanding of it. US federal taxes; state tax excluded and additive.

Sources: .claude/scratch/claims-ledger.md, docs/research/portfolio-recommendation.md, docs/research/final-construction-test.md, docs/research/trend-weight-under-uncertainty.md, docs/research/structural-and-tax-edges.md §8, src/content/shelf.ts, src/content/placement.ts, src/lib/placement.ts, .claude/scratch/market-scan-2026.md, .claude/scratch/stacking-thesis-verdict.md.

---

1. The construction

Taken from the construction Experiment 016e scored (docs/research/final-construction-test.md, spec 3a86ef6f…, run cd2fb4b9…). Not a new invention.

The table below is the 25% arm, which is what was tested and is not what is recommended. §2 takes the wrapper to 30% and VTI to 19%, which is the vector /portfolio publishes and the one every figure record on the site carries. Nothing else in the construction changes, so §1’s fund-by-fund reasoning applies to both.

Capital weights — the only vector anyone types

Weighted gross fee computed here is 33.4 bp; 016e’s frozen run reports 33.9 bp for the same arm (small fee-grid differences). Net-of-lending weighted cost of 31.3 bp/yr is derived here from the eight funds’ Form N-CEN medians in final-construction-test.md §2 plus VTI 1.16 and VXUS 1.43 from structural-and-tax-edges.md §6.1.

Notional exposure — a derived audit quantity, never typed

RSST delivers 1.072 of US equity and 1.00 of managed-futures notional per dollar of capital (Form N-PORT 2026-04-30; src/content/shelf.ts).

Equity notional is 101.8%. Bonds: zero. That is a position, not an oversight, and the equity-versus-bond split is a larger decision than everything else on this page — 60/40 to 90/10 was worth +127.1 bp/yr against 485 bp of tracking error (B3).

What the evidence can and cannot separate

The construction tournament scored four candidate constructions against a leverage-matched control: recommended +2.20, AVUV-instead-of-VTV +2.35, previous recommendation +1.92, investor’s original eight +2.49 pp/yr — a spread of 0.57 inside detection floors of 2.75–3.33. Ranking them is reporting noise, and the repository says so.

So the fund choices below are made on cost, spread, turnover and holdability, and are stated as such:

• VTV over AVUV. Unresolved on both admissible windows (−0.15 pp/yr [−0.68, +0.34] against a 0.68 floor). Picked on cost: 2.70 bp net against 24.54, and 8%/yr turnover against AVUV’s higher figure. AVUV’s extra exposure over VTV is 87% size, on a premium of +0.33 pp/yr against a 2.47 floor.
• RPV rejected, not on exposure. It buys HML +0.369 over VTV while selling RMW −0.204 and UMD −0.173, at 42%/yr turnover against VTV’s 8%. Negative under all four premium scenarios.
• SPMO not added. Better than MTUM on every knowable dimension (13 bp, 44% turnover) and +0.02% a year at a 5% weight against a −0.14% to +0.18% range, with an active leg +0.626 correlated with IDMO’s, which the portfolio already owns.
• AVDV kept. +0.28 pp/yr [+0.05, +0.56] against a 0.29 floor — short of resolution by 0.01, positive in the full window and all seven declared sub-periods, reproduced independently at +0.29 on the unlevered pair. Its net cost is 30.03 bp, not the 36 bp headline.
• RSST over CTAP, MATE and JPFP. Not on return, which cannot separate them, and no longer on the funding rule (RSST δ −0.07, CTAP −0.027, MATE −0.159 all keep the whole gap). On age, size, counterparty and spread. See §4.

Two arithmetic facts that carry the construction

1. The funding rule is worth more than any sleeve, and it contains no sleeve property. a_p − σ_p² ≈ +2.44 pp/yr for a 100%-equity base. RSST’s δ of −0.07 keeps essentially all of it; a standalone managed-futures fund keeps none. This is why the trend leg is bought through a stacked wrapper rather than by selling equity.
2. Stacking has a hard ceiling. At the ρ = 0.435 measured among this portfolio’s own active sleeves, an unlimited stack of 55%-likely bets reaches 0.576 and stops. Eight tickers were worth 3.71 effective independent bets. This portfolio is about 93% substitution, so its edge is a weighted average bounded by its best single sleeve, not a sum. Only the RSST line is financed.

---

2. The 25% versus 30% trend decision

The trade, stated

The five points come out of VTI. Because RSST carries 1.072 of US equity per dollar, US equity notional barely moves (65.8 → 66.2) and the whole change is +5 points of trend.

Growth. The higher weight wins by 0.50 pp/yr [0.23, 0.77] against a 0.39 floor — the only whole-portfolio comparison in either tournament that clears its own resolution, negative for the 25% arm in all seven declared sub-periods (−0.19 to −0.94). The tournament is explicit that this is a leverage result, not a construction one: at the panel’s realised 9.83% US equity premium, more notional wins.

And it is not the comparison in the table above. 016e’s recommended_vs_original scores RSST 25 / VTI 24 / VTV 15 / VXUS 16 / AVDV 10 / IDMO 5 / AVES 5 against the investor’s original eight-fund proposal — RSST 30, VTI 20, AVLV 15, DFIV 10, VEA 10, IDMO 5, IEMG 5, AVES 5 — so the pair differs in four holdings as well as in five points of the wrapper, and no arm in the programme holds these seven funds at 30%. The reason to read the gap as the wrapper: the two tilt complexes, stripped of trend and of leverage, score +0.7996 and +0.79 pp/yr against the same unlevered control on the same 427 months, leaving the wrapper as what moved. That is an argument and not a paired measurement — the two figures come from two tournaments and carry no interval of their own — so it cannot bound the fund-list contribution. 016e’s open question 4 asks for the matched pair, and it has not been run.

Holdability. Abandonment probability over thirty years at a −20% relative-drawdown trigger runs about 17% at 30% against 11% at 25% at the premium prior’s median, and 66.7% at 30% if the premium is gone entirely. Holdability and return are a bet on the same parameter — the same risk twice, not a diversification of it.

Price of the step. The 0.25 → 0.30 step buys +15 bp/yr of expected growth for +8.2 points of abandonment probability, about 1.81 bp per point with capitulation priced inside the path. Every step from 0.15 to 0.40 still buys more than 1 bp per point; the elbow that used to sit at 0.20 was an artifact of a units error and no longer exists.

Four routes bracket the weight and 25% is the only one all four admit: variance minimisation 21.6% [10.3, 32.8]; growth subject to holdability 15–25%; minimax regret 25% (robust 20–30%); construction tournament no interior optimum.

Recommendation: 30%

For this investor. Three reasons, in order of weight.

1. The comparison that favours 30% is the only one in the programme that clears its own floor, even though it does not isolate the weight. Nearly everything else here is unresolved. Discarding the one resolved measurement in favour of four routes that disagree with each other is choosing the weaker evidence — but it is one arm short of the comparison it is being asked to be, and that belongs in the reason rather than in a footnote.
2. A contribution stream is what carries a position through a drought. The abandonment model’s trigger is a relative drawdown, and the mechanism that defeats it is new money arriving at a low price. This investor contributes 5–15%/yr, which is 2.5 to 7.5 times the largest rebalancing rotation the portfolio needs.
3. Zero is the worse extreme. Over the panel’s one flat-to-negative equity decade (1999-03…2009-02, equity −2.55%/yr) a 30% overlay contributed about +9.5 pp/yr above its own complement and turned that decade from −2.55%/yr to about +0.05%/yr, against +0.21 pp/yr in ordinary decades.

Choose 25% instead if you would sell it. A portfolio held beats a better one abandoned, and the honest test is whether a decade of the sleeve contributing nothing while equities rise would end with you selling. If the answer is maybe, take 25% — every route admits it, and the growth you give up is half a point a year against a 6.0% tracking error you would never be able to attribute anyway.

One caveat that cuts against 30%. The resampled probability that the overlay is the deeper drawdown is 6.9% at 30% of notional and doubles from 10.8% to 18.9% between 58% and 60%. That is a cliff, not a ramp, and it is why the weight cannot simply be raised further.

---

3. The placement problem

This is the new work. Method is src/lib/placement.ts (taxableCostBp − shelteredCostBp), applied at fund level exactly as research/src/portfolio_edge/studies/investor_placement.py does. Arithmetic reproduced in scratchpad/placement.mjs.

priority = Box1a yield × [q × cgFraction + ordinary × (1 − cgFraction)]
           − creditable foreign tax yield

The second term is the whole of the correction, and it is zero for everything except a foreign holding. §408(e)(1) exempts an IRA from tax, so there is no US tax for §901 to credit against and a §904 numerator of zero: foreign withholding is paid and permanently lost inside a traditional IRA and a Roth alike.

3.1 Inputs

Three of eight yields are filed for this exact construction; three are derived and flagged. Every derived assumption pushes its fund toward the taxable account, so the plan is conservative in the direction it is uncertain. Yields mix windows, which is the input a placement ranking is most sensitive to and none of these is point-in-time.

3.2 The ranking, at three brackets

Basis points per dollar of shelter capacity. Wrapper on the recognised basis.

Every international fund outranks every US equity fund at every rate, and VTI is last at every rate. This is the opposite of “hold international in taxable for the foreign tax credit,” and the reason is that the credit is small and the yield gap is not.

Sheltering all four international funds destroys 7.45 bp/yr of foreign tax credit permanently (repository figure for the eight-fund original: 8.81 bp/yr). Sheltering them buys back far more: VXUS alone is worth 64.91 bp per dollar sheltered at 23.8%, against VTI’s 25.39. The credit is not close to deciding it.

The break-even is exact. q* = u·w·y_i/(y_i − y_d) puts the developed break-even at 10.52%, below every positive US qualified rate. The US schedule offers 0%, 15%, 18.8% and 23.8%, so developed ex-US always belongs in shelter ahead of US equity. That is a fact about the bracket schedule, not about the funds. The 0% bracket is a trap: §904 limits the credit to US tax on foreign-source income, and there is none, so a 0%-rate investor forfeits the withholding in both locations and the credit argues for neither.

A high-turnover momentum fund at a 5% weight outranks almost everything else. IDMO designates 25% qualified dividend income against 105% portfolio turnover, and its December distribution of $0.68417 of short-term and $0.27579 of long-term gain per share takes its taxable distribution to 4.40% of net assets with only 25.6% at capital-gain rates. The ETF in-kind shield does not survive 105% turnover. Nothing about its 5% size or its “momentum” label suggests it should be a placement priority, and it is second in the queue at every bracket.

The order is stable across the bracket range. Only AVDV and VXUS change places, and at 18.8% they are 0.01 bp apart. Treat them as tied and order them by whatever is operationally simpler.

3.3 The wrapper, and why its unresolved input does not stall the decision

RSST routes its trend leg through a wholly-owned Cayman controlled foreign corporation whose net income is included in the fund’s taxable income each year. Its undistributed ordinary income went from 1.40% to 8.56% of net assets in one year while the fund distributed 0.33%. Two readings:

• Distributed — count only what shareholders were taxed on: 33.72 bp/yr at 23.8%. Corroborated independently by the fund’s own prospectus, 17.17%/yr before tax against 16.85% after taxes on distributions, a 32 bp gap.
• Recognised — count the 9.27% of net assets the fund recognised: 361.78 bp/yr.

Computed here for this construction at 23.8% and f = 1, at the recommended 30% wrapper (25% figures in brackets):

34 to 1 at both weights. The repository’s figures for the eight-fund original are 1.12–8.54 bp/yr against 42.62–89.88 bp/yr — ten to one at every bracket and every open-menu fraction. Both readings agree: shelter the wrapper, and leave the measurement open. The review trigger is the fund’s next December distribution.

3.4 The plan — account by account, at equal nominal thirds

Shelter capacity is 66.7% of the portfolio. Assume f = 1, the whole tax-deferred third in a rollover IRA with an open menu. Stated at the recommended 30% wrapper; the 25% variant is beside it and moves nothing except how much VTV spills into taxable.

At a 30% wrapper the traditional third is 90% one fund. That is the arithmetic warning already on record: a 30% wrapper consumes nine tenths of a one-third pre-tax account. The 2026 mandatory-Roth catch-up rule makes it worse over time — SECURE 2.0 §603 is effective for 2026, and a participant whose 2025 FICA wages from the sponsoring employer exceeded $150,000 must make every catch-up dollar as designated Roth, removing $8,000 to $11,250/yr of pre-tax capacity. The pre-tax account this wrapper lives in grows more slowly every year, for exactly this investor.

Roth and traditional are not interchangeable. The recurring drag is identical in both — zero for everything except the forfeited foreign credit, which is also identical. What separates them is that Roth growth is never taxed, so the Roth’s advantage is proportional to expected return, while a traditional balance is shared with the government at the withdrawal rate. Exactly: swapping growth factor A in a Roth of size R for factor B in a traditional of size T at withdrawal rate t changes terminal wealth by (R − T(1−t))(A − B). At equal nominal thirds, a 24% withdrawal rate, 30 years, a 30% sleeve swapped and a 1 pp/yr expected-return gap, that is 1.96 bp/yr. It is a forecast, and it is not free: holding the same after-tax allocation, moving a sleeve to the Roth raises your share of its dispersion by the same factor it raises the mean.

Which is why the wrapper goes in the traditional and the tilts go in the Roth. The naive rule — shelter the highest drag — points the other way and puts the wrapper in the Roth. The drag is the wrong instrument for that question, because it is the same number in both accounts.

A tax-deferred balance is not the investor’s money. At a 24% withdrawal rate, $100,000 of traditional IRA is $76,000 of investor wealth. Equal nominal thirds are Roth 37.84% / traditional 28.76% / taxable 33.41% after tax — the taxable account is the larger constraint after tax, not the smaller. The ranking above is stated per dollar of shelter capacity precisely to sidestep that.

3.5 What the plan is worth

Three benchmarks, and their answers do not add.

Computed at the recommended 30% wrapper. The 25% figures are 121.44 / 39.43 / 33.75 / 6.41.

The first is not a saving anyone can capture, because nobody holds everything taxable. The second is the value of getting the order right, and the repository’s own scope note says a pro-rata control is infeasible for an investor whose shelter is partly a captive employer plan, so measuring against it overstates the result. The booked figure against a feasible control is −2.04 / +6.66 / +5.41 bp/yr at f = 0 / 0.5 / 1 at the top bracket, falling to −0.40 / +2.56 / +2.04 at 15%.

So the honest total value of placement is +2 to +7 bp/yr against the investor’s own counterfactual, and it is negative if the tax-deferred third is wholly captive. Not the 121 bp headline. Four lines were withdrawn to get there: a rebalancing hurdle avoided is not a saving; lot selection is mutually exclusive with never selling; the wrapper’s undistributed accrual is conditional on a distribution decision the fund has not made; and fee and fund-structure lines are measured against a cheap index and belong to a different benchmark.

The one line that is larger than the whole ordering decision: never having to sell in the taxable account is worth about 14 bp/yr. Realising 10% of standing gain a year costs 41.5 bp of the 84.1 bp deferral at a 30-year horizon.

3.6 Where a captive employer menu breaks it

A typical employer 401(k) menu holds a broad US index fund, a developed ex-US one and an emerging one, and nothing else this portfolio owns. Of the seven lines here, only VTI and VXUS can go in it. No employer plan offers a return-stacked ETF, an Avantis systematic fund or a single-factor momentum ETF.

Write f for the share of the tax-deferred third sitting in a rollover IRA. The unconstrained plan shelters VXUS at 16% of the portfolio, so the employer plan is free while it is no larger than that:

1 − 0.16/0.3333 = 0.52

The menu binds below f = 0.52 for this construction. The repository derives 0.55 the same way for the eight-fund original, which shelters VEA 10 + IEMG 5 = 15%.

At f = 0, computed here at the recommended 30% wrapper:

VTI, last in the queue at every rate, is forced into shelter at 17.3% while AVDV and AVES are evicted to taxable. That is the exact inverse of the ranking, imposed by a fund lineup rather than by any tax fact. Cost, computed here at 23.8%: 9.12 bp/yr at a 30% wrapper and 6.00 bp at 25% — larger than the whole booked placement edge either way. The repository’s figure for the eight-fund original is 9.09 bp, which this reproduces.

Two consequences.

• Consolidating an old employer balance into a rollover IRA buys the whole f = 0 to f = 0.5 improvement for the cost of a form. It is the cheapest lever available.
• At f = 0 the wrapper has no choice but to take the Roth, because the traditional third cannot hold it. That is a second and separate cost of a captive plan, and it is the one place the plan above cannot be executed as written. The Roth is 33.3% against the wrapper’s 25%, so it fits — but at a 30% wrapper the margin falls to 3.3 points, and above 33.3% the wrapper spills into taxable and the weight itself becomes the thing to reconsider.

3.7 Rebalancing feasibility

Restoring the target without a taxable sale is reachable iff every fund’s taxable holding is at or below its portfolio target weight (exact, not a heuristic). At a 30% wrapper: VTV 14.3 against a 15 target, VTI 19.0 against 19.0. VTI sits exactly at target, so it has zero headroom, and the constrained direction is selling US equity to buy international. That rotation needs roughly two points of the portfolio a year; contributions of 5–15%/yr cover it 2.5 to 7.5 times over.

The clean fix is to leave one or two points of headroom on VTI — hold 18 in taxable and 1 in the Roth at a 30% wrapper, or 23 and 1 at 25%. The repository’s joint solution costs 0.28 bp/yr for a 1 pp headroom band. A forced taxable sale costs hundreds of times a spread.

---

4. The two new findings, checked against the recommendation

CTAP’s 33 bp bid-ask spread

capital-efficiency-and-breadth.md concluded that neither the funding rule nor the fee table separates RSST, CTAP and MATE. The market scan found CTAP’s 30-day median bid-ask spread is 0.33% against RSST’s 0.09% (Rule 6c-11(c)(1)(v) disclosures, read 2026-08-22). A round trip in CTAP costs about 66 bp; a one-way purchase adds 33 bp to a first-year all-in cost of about 0.81%.

Does it change the recommendation? No — it confirms it, and it retires a caveat. The recommendation already names RSST. The spread finding removes the last reason to reopen the comparison: the three wrappers were tied on δ and tied on all-in fee, and the tie is now broken by 24 bp/yr on a one-year hold, on a shelf whose entire fee dispersion is 18 bp. For an investor who rebalances annually, CTAP is the dearest of the three, not the cheapest.

What it does change is a review trigger. CTAP’s waiver lapses on 2026-12-04, taking it to about 0.99% all-in. If the waiver is renewed, its spread tightens with age and size, and its 82.48%-of-net-assets bilateral swap exposure to one bank falls, the comparison is worth reopening. Not before.

Fidelity’s $100 per ETF purchase

Fidelity began charging $100 per trade on ETFs from issuers refusing platform fees on 2026-06-01; Schwab has confirmed a comparable programme for year-end 2026. On a $10,000 purchase that is 100 bp on day one — larger than the entire annual expense ratio of every fund here except the wrapper, and about three times AVDV’s whole annual net cost.

Does it change the recommendation? It changes how you execute it, not what you hold — and it is the one open item that could change what you hold.

The exposure is structurally asymmetric. Vanguard, iShares, Invesco and American Century (Avantis) are not plausibly at risk. Tidal, which issues RSST, is exactly the profile that is — a boutique with little fee revenue to hand over. No shelf issuer appears on any published list, but Fidelity’s live service-fee list returns HTTP 403 to automated fetches and could not be read, so this is unverified rather than cleared.

Three concrete consequences for a contributing investor:

1. Check your broker’s service-fee list before the first purchase, for RSST above all.
2. A per-trade fee is fatal to monthly contributions and survivable annually. $100 twelve times a year on a $200,000 portfolio is 60 bp/yr. Once a year on the same portfolio is 5 bp. If RSST carries the fee at your broker, either move that line to a broker that does not charge it, or buy it once or twice a year and direct monthly contributions at the index funds.
3. This is a reason to prefer the wrapper over a standalone managed-futures fund a second time. One line rather than two means one exposed purchase rather than two.

One more thing the scan turned up that is not a change yet. RSIT — Return Stacked International Stocks & Managed Futures, inception 2026-05-06, 0.98%, $68.53m, spread 0.15% — is the international twin of RSST, with a base leg near 1.00 and δ ≈ 0.00. It lands exactly where this construction is thin. It is three and a half months old with no N-PORT and no measurable loading, so nothing can be promoted from it. Revisit when it has 24 filed months.

---

5. The failure modes

What losing looks like

The failure mode is not a crash. It is a decade of quiet monthly underperformance against the most familiar comparator.

Every one of those is inside the history this construction is built from. None of them is a tail scenario.

Add the portfolio-level picture: maximum drawdown −50.3% against the levered control’s −64.6%, longest run under water 42 months, and a worst decade against its own control of −1.15 pp/yr for seventeen years post-2009, against +8.15 pp/yr through the 1999–2009 flat equity decade. The asymmetry is the wrapper’s.

The reframing fact

At this portfolio’s tracking error, thirty years of holding it cannot establish whether it worked.

• Against a leverage-matched cheap index: +2.20 pp/yr [+0.05, +4.57] against a 2.83 floor — 59 years to separate, at 6.0% tracking error.
• Against a same-split cheap core: tracking error about 400 bp/yr, of which 372 bp is the trend overlay. The thirty-year detection floor at 400 bp is 93 bp/yr, against a central edge of 92 bp. The portfolio is designed so that thirty years of holding it cannot settle the question.
• The arithmetic is exact and horizon-free: T = (z·s/e)². The same 50 bp edge reaches 90% confidence in 24 days at 10 bp of tracking error and 105 years at 400 bp. And every probability computed that way is an upper bound, because it assumes the edge is known.
• An estimated edge imposes a second ceiling no horizon removes: P(T) → Φ(e/τ). Path noise washes out; error in the mean does not. Thirty years converts an unsignable premium into a longer wait for the same coin flip.

The load-bearing precommitment: the comparator must be leverage-matched. Comparing a levered portfolio with an unlevered index credits the leverage to the strategy on the way up and blames the strategy for it on the way down.

And a monitoring rule that follows from it: do not set a performance review for a diversifying sleeve. At a 30% weight such a sleeve underperforms in 43.8% of resampled ten-year histories even when its premium is positive, so a performance trigger removes it for doing what it was bought to do. Require three consecutive readings. A bar must be coarser than the instrument that reads it — the trend-weight page proposed removing the sleeve below a 0.70 loading, and the first measurement returned 0.681 on an interval of [0.406, 0.955] that spans the bar twice over. The defensible bar is the wrapper’s break-even against a standalone fund, 0.19 to 0.27 at a 25% weight, and the measurement clears it comfortably.

The one thing on this page nobody has measured

Three of seven lines have no measured factor exposure of any kind. The wrapper’s trend loading rests on 31 filed months, roughly one market regime, on an interval that cannot tell one dollar of delivered trend from four fifths of one. Its financing spread is disclosed by no issuer, because futures financing lives in the basis, and it decides the sign of the overlay’s contribution. 52% of the managed-futures ETFs listed in 2019 had stopped filing by the end of 2025. RSST is under three years old.

---

6. What would change this

Ordered by how much it moves the answer.

1. Your own numbers. The rollover share f of the tax-deferred third — it moves the booked placement line from −2.0 to +6.7 bp/yr and decides whether AVDV can be sheltered at all. Your maximum tolerable drawdown and months underwater, which is the single most valuable missing input in this repository and is not a research task. Your bracket.
2. The equity-versus-bond split, which this construction sets at 100% equity by omission and which is worth more than every tilt combined.
3. RSST’s trend loading refreshed at 48 filed months, around 2027-09.
4. The fund-level financing spread, which decides the sign of the overlay’s contribution.
5. RSST’s next December distribution, which resolves the recognised-versus-distributed reading behind the placement’s conditional line.
6. Whether your broker charges $100 per purchase on the wrapper. Unverified; HTTP 403.
7. CTAP’s waiver on 2026-12-04, and its spread and counterparty concentration after it.
8. RSIT at 24 filed months, if the international sleeve is ever to be financed too.

---

Verified, derived, assumed

Verified from the repository. Every tournament figure, interval, detection floor and sub-period in §1, §2 and §5, from final-construction-test.md and run cd2fb4b9…. Every filed tax characteristic for RSST, VTI, IDMO and AVES, and the 23.8/18.8/15 priorities for those four, from structural-and-tax-edges.md §8 and src/content/placement.ts. Net costs from 50 fiscal years of Form N-CEN. Wrapper structure and δ from Form N-PORT 2026-04-30. Spreads and the broker programme from .claude/scratch/market-scan-2026.md.

Derived here, and reproducible from scratchpad/placement.mjs. VTV, VXUS and AVDV priorities at all three brackets. The f = 1 and f = 0 fill orders for the seven-fund construction at both wrapper weights. The 34-to-1 wrapper regret ratio. The 0.52 menu-binding fraction. The 9.12 bp menu cost at 30% and 6.00 at 25%. The 7.45 bp of forfeited credit. Weighted net cost of 31.3 bp at a 25% wrapper and 36.2 bp at 30%. The notional decomposition and the 126.8 gross.

Assumed. VTV, VXUS and AVDV yields are trailing-twelve-month figures from an aggregator read 2026-08-23, not sponsor filings. VXUS’s and AVDV’s qualified fractions are blends of VEA’s and VWO’s filed figures, not their own. AVDV’s withholding rate is VEA’s. VTI’s, VTV’s and the wrapper’s qualified fractions are assumed at 1.00. State income tax is excluded and additive; it compresses every gap without reordering them. Equal nominal thirds are assumed; after tax they are 37.84 / 28.76 / 33.41.