As of —

Quantifying the battle between carry-trade pressure and PBOC policy intent.

A three-layer pressure gauge for USD/CNY at the 2-year tenor — from the gross yield differential, through DXY-orthogonalised regression residuals, to the daily fixing's hidden defence posture. One number that tells you whether the line still holds.

Composite Policy Pressure

—

— / 100

—

LowModElevHighExtr

Coverage

The Carry-Trade Verdict Today

—

Verdict signal · cumulative track record

Verdict flips when:

—

PBOC Policy Stance Today

—

Layer One · Carry Feasibility

Carry Monitor — From Gross Spread to Hedged P&L

The "water pressure" of carry-trade capital: how much can you earn borrowing CNY and lending USD? We track both the raw 2Y yield differential and a CIP-implied hedged carry proxy that strips out forward-point costs. When hedged carry is positive and rising, speculative outflow pressure is structurally building — regardless of what the spot rate does today.

Method

Carry Decomposition

Raw Carry = US_2Y − CN_2Y (gross incentive) Hedged return = 100 × [ (1 + UST_1Y/100)×(F₁ᵧ/S) − (1 + Shibor_1Y/100) ] when CFETS 1Y outright F₁ᵧ is available ≈ UST_1Y + [100×(F₁ᵧ/S − 1)] − Shibor_1Y (same object, decomposed) Fallback (no F₁ᵧ) = Raw Carry − CIP_Basis% 2Y CIP-implied proxy MM Spread (1Y) = UST_1Y − Shibor_1Y (funding-rate analog)

Free APIs do not expose real swap-point quotes. We derive the hedging cost from Covered Interest Parity: the CIP basis measures how far the actual spot deviates from what arbitrage-free forwards imply. A positive hedged carry proxy means real arbitrage profit exists after hedging — historically rare and a sign of USD funding stress or capital-control friction.

Interest-Rate Differential — 2Y Tenor

Spread shaded by sign

Carry Pressure & On-Offshore Gap

Pressure metric vs CNH–CNY discount

Carry Decomposition — Raw vs Hedged

Hedged carry = raw carry minus CIP basis (forward-hedge cost proxy)

Money-Market Funding Layer — 1Y Tenor

Short-end funding rate differential (modern Libor-Shibor analog)

Roadmap

Real swap points, forwards & NDF

This build uses a CIP-implied hedged-carry proxy because public APIs do not expose onshore swap-point strips. The next upgrade is to layer in USD/CNY forward points, CNH forwards, NDF pricing, and multi-tenor (1M–2Y) hedged carry — separating onshore CNY from offshore CNH — so Layer 1 reflects tradable economics rather than only theoretical pressure.

Key Findings

Loading…

Layer Two · Mispricing

De-Noising With the Broad Dollar

A pure spread-vs-spot regression confounds two drivers: Sino-US rate differentials and broad-dollar moves. If the spread widens while DXY rallies, a stable CNY is rational, not intervention. We upgrade to multivariate OLS, regressing USD/CNY on both spread and DXY. The residual then isolates China-specific factors.

Only after the broad-dollar noise is filtered out does the residual become a clean reading of risk-premium and policy-intervention dynamics.

Method

Multivariate Specification

Single-variable (legacy): USD/CNY_t = α + β₁ · Spread_t + ε_t Canonical (this build): USD/CNY_t = α + β₁ · Spread_t + β₂ · DXY_t + ε_t Residual > 0 ⟹ CNY weaker than rate-differential AND broad-dollar predict Residual < 0 ⟹ CNY stronger than the model — likely defence or capital inflows CIP fair value (Layer 02·A): F_CIP = S_base × [(1+r_CN)² / (1+r_US)²] / [(1+r_CN_base)² / (1+r_US_base)²]

DXY — Broad Dollar Index

FRED Trade-Weighted, ~95% corr to ICE DXY

USD/CNY — Actual vs Multivariate Model

Residuals = mispricing signal

Residual De-Noising

Single-variable vs DXY-adjusted multivariate

Rolling OLS — β₁ & β₂

252d window: USD/CNY ~ spread + DXY

Model fit & residual z-score

In-sample R² and rolling (ε−μ)/σ (not pseudo-OOS)

Roadmap

Model stability & alternatives

Rolling coefficients and R² are a first diagnostic step. Further work: expanding-factor controls (e.g. terms-of-trade proxies), pseudo out-of-sample forecasts, and comparison vs. rolling OLS with shorter windows, ridge, or small VARs — published only when the specification passes stability checks rather than a single static β.

CIP Deviation

Spot − CIP-implied fair value

Key Findings

Loading…

Layer Three · Policy Intent

Decoding the Daily Fixing

Between Beijing 4:30 PM close and 9:15 AM fixing lies a full New York trading session. Overnight DXY moves print mechanical CNY changes that a naive model mistakes for intervention. We strip those out, then read what remains: the true defence posture of the central bank.

Roadmap

Policy toolkit monitor

Fixing bias is one observable lever. A fuller toolkit would add CNH liquidity / CNH Hibor, PBOC offshore bill issuance, FX risk reserve ratio changes, evidence of counter-cyclical factor use, verbal guidance, state-bank spot behaviour, and the CNY–CNH spread — wired in as data becomes available in this open pipeline.

Method

DXY-Adjusted Bias

Naive: Bias_raw = PBOC_Fix − CNH_Prev_Close Canonical (DXY-adjusted): α = rolling 252d β of CNY returns on DXY returns Expected_Fix = Anchor × (1 + α · ΔDXY_overnight) Bias = PBOC_Fix − Expected_Fix Negative bias → PBOC set stronger CNY than DXY-adjusted expectation → DEFENDING Positive bias → PBOC set weaker CNY than expectation → ALLOWING WEAKNESS Defense Intensity = − rolling_mean(bias, 20d) × 10000 [pips]

The Policy Triangle

PBOC Fix vs Onshore CNY vs Offshore CNH

CNH Funding Stress — PBOC Offshore Defence

CNH HIBOR 1Y − Shibor 1Y · spikes flag squeezes

Fixing Bias — Raw vs DXY-Adjusted

In pips · 20d & 60d trends

Scenario	Carry	Bias	Implication
Max Tension	High ↑	Strongly Negative	PBOC burning reserves to hold the line — watch for capitulation.
Managed Decline	High ↑	≈ 0	PBOC permitting orderly weakness.
Comfortable	Low ↓	≈ 0	No policy dilemma.
CNY Strong	Negative / Low	Positive	PBOC resisting appreciation.

Key Findings

Loading…

∑

Synthesis · Composite Trend

One Number, Three Forces

A weighted blend of all three layers. Watch sustained crossings into the 75+ region — historically the moments when something gives.

Composite methodology

Weights, standardisation & the 75 line

◷

Credibility · Historical stress lens

Episodes to read against the dashboard

This is not a formal event-study or backtest: public rebuilt series lack point-in-time feeds. Use the table as a qualitative checklist — did composite, fixing bias, hedged-carry proxy, and regression residual co-move ahead of known stress windows?

Window	Context	What to check on this site
2015	CNY reform + devaluation pressure	Composite vs carry percentile; fixing bias turning defensive; residual volatility.
2018	US–China trade war	Policy score vs spreads; whether mispricing layer leads spot once DXY is filtered.
2022	Strong USD / Fed cycle	DXY-orthogonalised residual — clean China-specific signal vs broad dollar.
2023	CNY near 7.30 defence narrative	Sustained composite > 75 with negative fixing bias; hedged-carry proxy trajectory.
2024–25	Managed defence / range-trading phase	Rolling β stability; residual z extremes; policy percentile vs carry percentile divergence.

⊞

Explore · Build a Chart

Roll Your Own View

Pick any combination of the underlying series. Rendered live in your browser — no server round-trip.

Frequency

Series

⊕

Reference · Variable Glossary

Every Field, Sourced

Each computed field, its formula, source, and unit. Click to expand.

Full Variable Glossary (48 fields)

↓

Export · Data

Recent Observations

Last 30 trading days · full series available below.

↓ Download CSV (full series) ↓ Download data.json ↓ Download Excel (.xlsx) ↓ Replication Notebook (.ipynb)

How to Cite

Erick Wei (2026). USD/CNY Macro-Policy Divergence Tracker — 2Y Focus. Online dashboard. Accessed —.