Inception Report · 2026-05

1 · 区间回顾

The framework began publishing daily NAV on 2025-12-01 as a backtest / monitoring period; the first formal rebalance with binding falsification conditions takes effect on 2026-06-01. The 125 trading days that precede this whitepaper-aligned launch are reported here as an inception report rather than as a monthly review — no view has yet been falsified, because no view has yet been formally committed.

Group performance over the inception window (2025-12-01 → 2026-05-25):

Realised volatility for all three groups is within their respective targets (§2 of the whitepaper: Base ≤ 8%, Bull ≤ 12%, Bear ≤ 6%). The Bull − Base spread of +200bp is suggestive but, at this sample size, statistically indistinguishable from zero — a result that is itself worth reporting, because publishing such a spread as “view alpha” would contradict the sample-size discipline committed to in whitepaper §5.3.

In practice, this means the only claim we make at inception is that the framework is operating within risk budgets and has produced returns broadly consistent with the cross-asset exposures it carries. Any further claim about view skill is deferred to the post-2026-06-01 closed-decision sample.

2 · Macro regime: where we stand

Growth and inflation momentum are computed each month as 5-year rolling z-scores across a curated set of US indicators (whitepaper §6.1). The latest read for 2026-04 is:

The point falls in Q4 (growth ↓ · inflation ↑) — the stagflation-hedge quadrant in the Dalio (1996) taxonomy. The dominant signal pushing into Q4 is the negative-momentum reading on nonfarm payrolls (z = -1.12), which has the framework lean toward gold + cash + long duration; the breakeven inflation reading is the only positive contributor to the inflation axis, and core CPI / PCE both surprised softer than the 5-year trend.

In practice, this means the regime overlay (§6.2 of the whitepaper) tilts toward OW gold + OW duration + N equity + UW credit for the 6-month horizon — and these tilts are visible in the published 2026-06 weight vector (next section).

3 · Current factor exposures

Daily group returns since inception have been regressed on five macro factor proxies using a rolling 60-day OLS window (whitepaper §5.2). The latest betas, from a window of 114 observations:

Two readings stand out. First, the USD beta is large and negative across all three groups (Base -0.96), which is mechanically attributable to the ~25% offshore-USD exposure (SPY + QQQ + TLT + gold) and consistent with the framework’s stated v1 limitation that no FX adjustment is applied (§9.2). Second, the commodity beta is positive (Base +0.14) — the gold overweight is delivering the directional exposure the framework expects of it, with no sign that the gold sleeve is moving anomalously relative to its proxy.

The intercept α is, in basis-point-per-day units, +0.73 for Base, +1.46 for Bull, and -1.01 for Bear. Per whitepaper §5.3, no inferential claim is made about α with fewer than six closed monthly decisions; the intercept is published only as a disclosure artefact.

In practice, this means the USD path is, for the next several months, the single most consequential macro variable for the portfolio’s NAV. Adjusting this exposure is on the v2 roadmap (§9.2 — USD/CNY conversion); in v1 we accept the exposure as documented and treat the USD beta as the first item on each monthly review’s diagnostic checklist.

4 · Decision log status

The inception decision (id: rebalance-2026-06, confidence 3, horizon 1m) has been published with four falsification conditions:

1. Bull underperforms Base by more than 2% (view in A-share / gold has failed)
2. Bear outperforms Base by more than 2% (regime has shifted to Q3/Q4, Bear weight should rise)
3. All three groups underperform 60/40 by more than 1.5% (framework problem, not view problem)
4. VIX > 35 sustained for 5 trading days (liquidity event, view assumption invalidated)

None of the four conditions has triggered as of publication date (VIX month-end 16.8; full inter-group spreads within tolerances). The verdict field will be auto-populated from nav-history.json after 2026-07-01 by scripts/update_verdict.py; a manual lessons: annotation is required before the record is marked as closed.

In practice, this means the first formally closed decision arrives at 2026-07-01 month-end. The audit ledger published at /track-record will update automatically; the discipline that needs the human in the loop is the lessons paragraph — the only field the verdict-update script deliberately refuses to populate.

5 · Where I might be wrong

Three concerns are worth flagging explicitly before the formal launch:

• The Q4 regime classification is dominated by one indicator. The payroll z-score (-1.12) carries most of the weight in the growth axis this month. If the next NFP print revises strongly upward, the classification could flip to Q3 (growth ↓, inflation ↓) and the duration

  • gold tilt may underperform a more equity-friendly Q2 (growth ↑, inflation ↓) configuration. A regime overlay that depends on a single monthly print is, by construction, noisy.

• The 5-factor regression is high-R² but conditional on the inception window. R² of 0.47 for Base is informative but the window (114 days) spans a single regime. The first true test of the factor model is whether the betas remain stable across a regime transition; the v1 expectation is that they will not, and the rolling design is intended to surface this rather than hide it.

• No FX adjustment overstates the offshore portion’s contribution. The -0.96 USD beta on Base means that approximately 50% of daily NAV variation moves with the dollar; the v1 NAV understates this risk to a CNY-domiciled reader, because the offshore book is valued in local currency. v2 (§9.2) will fix this. Until then, the published returns should be read with a mental adjustment of ±200 bp/year depending on USD direction.

In practice, this means we will be watching the next two NFP prints, the stability of the 5-factor betas across the first regime shift, and the USD path. If any of these diverges sharply from the v1 framework’s assumptions, the next monthly note will report a framework revision candidate rather than a routine review.

6 · Outlook · June 2026

The 2026-06 rebalance keeps the inception weights largely intact: Base remains risk-parity neutral with the duration + gold overweight expressed through TLT (15%) + CGB (15%) + gold (10%); Bull adds A-share large-cap to 22%; Bear pushes cash to 25% + duration to 35%. The single view being tested explicitly is that the Fed will skip a July cut and the long end will hold a 4.4–4.8% range; the falsification condition is a 10Y UST breaking 4.85% on a closing basis. The companion view that A-share large-cap will outperform A-share mid-cap is held implicitly through the Bull tilt and will be reviewed at the August note.

In practice, this means the next 30 days are diagnostic in two ways: the July FOMC outcome will close or extend the first explicit view, and the July NFP revision will provide the second data point for the regime classifier. Both feed into the 2026-07-31 monthly note, which will be the first to publish a closed verdict on a real view.

Verdict on this note: itself unscored — inception reports are not counted toward the hit-rate ledger. The first closed verdict arrives at 2026-07-31.