Use when designing cross-asset macro strategy, analyzing central bank policy impacts, constructing regime-based allocation frameworks, interpreting global liquidity conditions, modeling inflation-growth scenarios, or building intermarket signals. Handles macro regime classification (expansion, contraction, stagflation, goldilocks), central bank reaction function modeling (Fed, ECB, BOJ, PBoC), global liquidity measurement (G4 central bank balance sheets, real rates, credit impulse), intermarket ratio analysis (equity/bond, copper/gold, HYG/TLT), risk-on/risk-off frameworks, currency war and competitive devaluation analysis, and macro scenario stress testing. Do NOT use for individual security selection (route to portfolio-signal-manager), trade execution (route to algorithmic-trader), single-asset analysis (route to the relevant asset skill), or trade journaling (route to trade-performance-analyst).
Use when designing cross-asset macro strategy, analyzing central bank policy impacts, constructing regime-based allocation frameworks, interpreting global liquidity conditions, modeling inflation-growth scenarios, or building intermarket signals. Handles macro regime classification (expansion, contraction, stagflation, goldilocks), central bank reaction function modeling (Fed, ECB, BOJ, PBoC), global liquidity measurement (G4 central bank balance sheets, real rates, credit impulse), intermarket ratio analysis (equity/bond, copper/gold, HYG/TLT), risk-on/risk-off frameworks, currency war and competitive devaluation analysis, and macro scenario stress testing. Do NOT use for individual security selection (route to portfolio-signal-manager), trade execution (route to algorithmic-trader), single-asset analysis (route to the relevant asset skill), or trade journaling (route to trade-performance-analyst).
Portability target: Spec-level. Runs on Claude Code, Copilot CLI, Cursor, Codex, Gemini CLI.
Skill library:skills/14-finance/
Ground Rules — Read Before Anything Else
#
Negative Constraint
Mechanical Trigger
Violation Response
R1
REFUSE to predict central bank decisions without (a) citing the most recent policy statement, (b) stating the market-implied probability, and (c) acknowledging the forecasting error distribution.
Trigger: output contains "the Fed will" or "the ECB will" or "the BOJ will" AND none of (policy statement date, market-implied probability, error distribution) are present
STOP. Respond: "Central bank forecasting requires: (1) most recent policy statement date, (2) market-implied probability from Fed Funds futures or OIS, (3) explicit acknowledgment that central banks surprise markets ~30% of the time. I will not issue a naked prediction."
R2
REFUSE to cite a single macro indicator without the release date and the consensus expectation. Context without benchmark is noise, not signal.
Trigger: output cites an economic data point (CPI, NFP, GDP, PMI, etc.) AND neither the consensus estimate nor the release date is present
STOP. Respond: "Economic data requires context: actual vs consensus vs prior. Without the benchmark, the data point is uninterpretable. Provide: [ACTUAL] vs [CONSENSUS] vs [PRIOR] as of [RELEASE DATE]."
R3
REFUSE to claim a macro regime without quantifying it against at least 2 independent indicators. A single metric does not define a regime.
Trigger: output asserts we are in a regime (expansion, recession, stagflation, etc.) AND fewer than 2 indicators are cited
Admit uncertainty. Macroeconomic forecasting is unreliable beyond 1 quarter. Any forecast beyond 3 months must carry an explicit confidence band and acknowledge the historical forecast error.
Trigger: output contains a macro forecast >3 months out AND no confidence interval is stated
STOP. Respond: "Long-horizon macro forecasts have wide error bands. 12-month GDP forecast RMSE is ~1.5%. State your confidence interval. I am adding: [90% CI: X-Y%] based on historical forecast dispersion."
R5
REFUSE to treat correlation as stable. Cross-asset correlations are regime-dependent and invert during crises.
Trigger: output implies a stable cross-asset relationship (e.g., "bonds rally when stocks fall") without acknowledging regime dependency
STOP. Respond: "Cross-asset correlations are regime-dependent. The stock-bond correlation was negative from 2000-2020 but turned positive in 2022. State: correlation in the current regime vs long-term average."
R6
Flag your knowledge cutoff. Macroeconomic data is released on a calendar schedule. If the most recent data point could have been released since your knowledge cutoff, flag it.
Trigger: macro data point cited without a release date AND the date could be past the knowledge cutoff
STOP. Respond: "[AS OF YYYY-MM-DD — VERIFY LATEST RELEASE]. The next release of [INDICATOR] is scheduled for [NEXT DATE]. Current value may be stale."
Anti-Hallucination Safety Protocol
Before producing any macro analysis, verify:
Admit uncertainty — All forecasts >3 months carry explicit confidence intervals based on historical RMSE
Flag your knowledge cutoff — Economic data tagged with [AS OF YYYY-MM-DD] and next release date
Never guess security — Do not fabricate economic data releases, central bank meeting dates, or policy decisions
Every macro data point cited with actual vs consensus vs prior
Every regime classification based on 2+ independent indicators
All cross-asset correlations labeled with regime context
Anti-Rationalization — No Excuses
Rationalization
Reality
"The macro outlook is clear from the data"
GDP is revised 3 times. The advance estimate has an average revision of 0.6%. The data you're looking at is not the final data. Macro is always murky — admit it.
"This time is different"
It never is. But the specific combination of factors IS always unique. Distinguish between "the cycle is dead" (false) and "the cycle has unusual features" (true).
"Markets are pricing in a recession"
Markets price probability distributions, not certainties. A 60% recession probability means 40% no recession. Don't round 60% to certainty.
"The Fed is behind the curve"
This phrase has been used every cycle since 1994. Sometimes it's true (2021), sometimes it's premature (2019). Back it with a specific r-star estimate and a timeline.
"Global liquidity is driving asset prices"
Liquidity is a causal claim, not a statistical one. Correlation between Fed balance sheet and SPX is high, but the causal mechanism (portfolio balance channel? bank lending channel? sentiment?) matters for timing.
Core Workflow
Phase 0: Regime Classification
1. GROWTH ASSESSMENT
|-- Real GDP growth: QoQ annualized (US), YoY (rest of world)
|-- Leading indicators: ISM Manufacturing PMI, Conference Board LEI, yield curve (10Y-2Y spread)
|-- Labor market: Nonfarm payrolls, initial claims, JOLTS quits rate
|-- Classify: [EXPANDING|SLOWING|CONTRACTING|RECOVERING]
|-- Complete when: Growth regime labeled with 2+ indicators cited, release dates tagged
2. INFLATION ASSESSMENT
|-- Headline CPI/PCE YoY and 3-month annualized momentum
|-- Core (ex-food-energy) vs supercore (ex-shelter) for direction
|-- Inflation expectations: 5Y5Y breakeven, Michigan survey, NY Fed survey
|-- Wage growth: Average hourly earnings, Atlanta Fed wage tracker
|-- Classify: [ABOVE-TREND|TRENDING-DOWN|AT-TARGET|BELOW-TARGET|DEFLATION]
|-- Complete when: Inflation regime labeled with momentum direction and expectations cited
3. POLICY STANCE
|-- Policy rate vs r-star estimate (Holston-Laubach-Williams, or NY Fed)
|-- Real rate: policy rate - core PCE inflation
|-- Central bank balance sheet: expanding / contracting / steady
|-- Forward guidance: dot plot, SEP, press conference language analysis
|-- Classify: [RESTRICTIVE|NEUTRAL|ACCOMMODATIVE|EMERGENCY]
|-- Complete when: Policy stance labeled with r-star comparison and balance sheet direction
4. REGIME SYNTHESIS
|-- Map growth + inflation into macro quadrant:
| ┌──────────┬──────────────┬──────────────┐
| │ │Growth >Trend │Growth <Trend │
| ├──────────┼──────────────┼──────────────┤
| │Infl >Tgt │ OVERHEATING │ STAGFLATION │
| │Infl <Tgt │ GOLDILOCKS │ RECESSION │
| └──────────┴──────────────┴──────────────┘
|-- Refine with policy stance overlay
|-- Label: [REGIME: OVERHEATING|GOLDILOCKS|STAGFLATION|RECESSION] [CONFIDENCE: HIGH|MEDIUM|LOW]
|-- Complete when: Regime quadrant labeled with all three inputs documented
Phase 1: Central Bank Reaction Function
1. FEDERAL RESERVE
|-- Dual mandate scorecard: inflation gap + employment gap
|-- Taylor Rule implied rate vs actual rate → hawkish/dovish gap
|-- FOMC dot plot: median vs distribution → dispersion of views
|-- Minutes language analysis: "some participants" vs "most participants" vs "all participants"
|-- Complete when: Fed reaction function modeled with next-meeting probability distribution
2. ECB
|-- HICP inflation by country: core vs periphery divergence
|-- Wage negotiations tracker (IG Metall, etc.)
|-- Transmission Protection Instrument (TPI) readiness
|-- Governing Council hawk/dove lineup by nationality
|-- Complete when: ECB stance mapped with country-level inflation divergence noted
3. BANK OF JAPAN
|-- CPI ex-fresh food vs 2% target: consecutive months above target
|-- Spring wage negotiations (Shunto) outcomes
|-- Yield Curve Control (YCC) band adjustments: 10Y JGB tolerance
|-- BOJ balance sheet: % of GDP, % of JGB market owned
|-- Complete when: BOJ normalization timeline estimated with trigger conditions
4. PEOPLE'S BANK OF CHINA
|-- Credit impulse: change in new credit as % of GDP
|-- Property sector indicators: new home prices, developer financing
|-- CNY fix vs market expectations → policy signal strength
|-- Reserve Requirement Ratio (RRR) and Medium-term Lending Facility (MLF) rate
|-- Complete when: PBoC stance labeled [EASING|NEUTRAL|TIGHTENING] with credit impulse cited
Phase 2: Global Liquidity Measurement
1. G4 CENTRAL BANK BALANCE SHEETS
|-- Fed + ECB + BOJ + PBoC: aggregate balance sheet in USD terms
|-- 3-month change (annualized) → liquidity direction
|-- Compare to global market cap growth → relative liquidity
|-- Complete when: G4 liquidity impulse computed and direction labeled
2. REAL RATES & FINANCIAL CONDITIONS
|-- US 10Y TIPS yield (real rate benchmark)
|-- Goldman Sachs FCI, Chicago Fed NFCI, Bloomberg FCI
|-- Credit spreads: IG OAS, HY OAS, leveraged loan spreads
|-- Dollar (DXY): directional impact on EM financial conditions
|-- Complete when: Financial conditions index trend compared to Fed's implicit target
3. GLOBAL CAPITAL FLOWS
|-- TIC data: foreign flows into US assets
|-- EPFR fund flows: equity vs bond vs money market
|-- EM portfolio flows: IIF tracker
|-- Complete when: Capital flow direction quantified for last 4 weeks
Phase 3: Intermarket Signal Construction
1. RATIO ANALYSIS
|-- Equity/Bond: SPY/TLT → risk appetite. Rising = risk-on.
|-- Copper/Gold: Cu/Au → industrial demand vs safe haven. Rising = growth optimism.
|-- HYG/TLT: credit risk appetite vs duration safe haven
|-- Cyclicals/Defensives: XLY/XLP within equities
|-- Complete when: 4 intermarket ratios computed and trend direction labeled
2. CROSS-ASSET CORRELATION MATRIX (REGIME-SPECIFIC)
|-- Equities-Bonds: negative (traditional) or positive (inflation regime)?
|-- Equities-Commodities: 0-30% normal, >50% in supply-shock regimes
|-- Equities-Dollar: typically negative (strong dollar = EM stress = risk-off)
|-- Gold-Dollar: typically negative; positive = risk-off with dollar strength (unusual)
|-- Complete when: Current correlation matrix compared to historical regime averages
3. VOLATILITY LANDSCAPE
|-- VIX, MOVE (bond vol), CVIX (currency vol), OVX (oil vol)
|-- Vol ratio: VIX/MOVE → equity vol relative to rate vol
|-- Vol term structure: contango (calm) vs backwardation (stress)
|-- Complete when: Cross-asset vol landscape mapped with stress signals flagged
┌──────────────────────┐
│ Two central banks on │
│ diverging policy paths? │
└──────────┬───────────┘
│
┌──────────▼──────────┐
│YES │NO
▼ ▼
┌──────────────────┐ ┌──────────────────┐
│ EXPRESS VIA: │ │ Trade the consensus │
│ - Long diverging │ │ direction: │
│ currency pair │ │ - Short-end rates │
│ (long hawk, short │ │ futures │
│ dove) │ │ - Steepener/flattener │
│ - 2Y rate spread │ │ on single curve │
│ position │ └──────────────────────┘
│ - Cross-market │
│ equity exposure │
│ (overweight hawk │
│ country equities │
│ if growth-driven) │
└──────────────────┘
Liquidity Impulse → Risk Positioning
┌──────────────────────┐
│ G4 liquidity impulse │
│ increasing? │
└──────────┬───────────┘
│
┌──────────▼──────────┐
│YES │NO
▼ ▼
┌──────────────────┐ ┌──────────────────┐
│ LIQUIDITY TAILWIND │ │ LIQUIDITY HEADWIND │
│ Size up risk. │ │ Size down risk. │
│ │
│ Is real growth also │ Is growth also │
│ accelerating? │ decelerating? │
└──────┬─────────┬─────┘ └──────┬─────────┬─────┘
│YES │NO │YES │NO
▼ ▼ ▼ ▼
┌──────────┐ ┌────────────┐ ┌──────────┐ ┌──────────┐
│ MAX RISK-ON│ │ SELECTIVE │ │ MAX RISK- │ │ LIQUIDITY │
│ Risk assets│ │ RISK-ON │ │ OFF │ │ ONLY │
│ full size │ │ Overweight │ │ Full hedge│ │ Hedge + │
│ │ │ quality │ │ │ │ wait │
└──────────┘ └────────────┘ └──────────┘ └──────────┘
Gotchas
Gotcha
Cost
Fix
Classifying macro regime from a single indicator — using ISM Manufacturing alone to declare "recession" while services PMI, payrolls, and consumer spending are still expanding. The manufacturing sector is ~11% of US GDP; services is ~77%. A manufacturing recession is not an economic recession.
$50K-$500K in misallocated portfolio positioning. Going defensive 6 months early costs 10-20% in foregone returns during the late-cycle melt-up.
Triangulate with 2+ indicators from different sectors. Always cross-check manufacturing surveys with services surveys and hard data (payrolls, retail sales, industrial production).
Assuming the stock-bond correlation is always negative — the 60/40 portfolio's diversification benefit relies on negative stock-bond correlation. But the correlation flipped positive in 2022 (both fell together) and again in Q3 2024. In inflation-driven regimes, bonds do NOT hedge equities.
$100K-$2M in unexpected drawdowns. A "balanced" 60/40 portfolio lost ~17% in 2022 because both assets fell. This is worse than a 100% equity portfolio in some inflation scenarios.
Check the stock-bond correlation regime quarterly. If 90D correlation >0, reduce bond allocation or replace with TIPS, commodities, or trend-following for diversification.
Quoting economic data without consensus context — reporting "CPI came at 3.2%" without stating consensus was 3.1%. The market moves on SURPRISE, not level. A 3.2% CPI with 3.5% consensus is dovish; a 3.2% with 3.1% consensus is hawkish. Same number, opposite trade.
$20K-$200K in wrong-direction trades. A macro trade based on headline alone has ~50% chance of being wrong because the market already priced the consensus.
Always report: [ACTUAL] vs [CONSENSUS] vs [PRIOR]. The surprise = actual - consensus. The momentum = actual - prior. Both matter for different trade horizons.
Treating the yield curve inversion as an infallible recession signal — the 10Y-2Y curve has predicted 8 of the last 5 recessions (false positives in 1966, 1998). It inverted in July 2022; if a recession started in late 2024, the lag was 30 months. The average lag since 1970 is 12-18 months but the range is 6-36 months.
$100K-$1M in opportunity cost from premature de-risking. Selling equities when the curve inverts and waiting for the recession costs 20-50% in foregone returns if the recession takes 2+ years to materialize.
Use the yield curve as ONE input alongside credit spreads, labor market, and leading indicators. Don't time the recession — size positions for a range of outcomes. The curve tells you the direction; it doesn't tell you the timing.
Confusing liquidity with solvency — the Fed cutting rates doesn't fix a solvency crisis. Rate cuts in 2001 (dot-com) and 2007-08 (GFC) didn't prevent massive equity drawdowns. Liquidity (rate cuts) addresses liquidity problems; solvency problems (overleveraged balance sheets) require debt restructuring.
$200K-$5M if positioned for "Fed put" recovery in a solvency crisis. Buying the dip aggressively after rate cuts in a solvency event can lead to 40-60% further drawdowns.
Diagnose: is this a liquidity event (funding markets stress, money market spreads) or a solvency event (corporate defaults rising, household debt service spiking)? Liquidity events → buy aggressively on central bank action. Solvency events → wait for debt restructuring before buying.
Overweighting the most recent data point — the last NFP or CPI print dominates the narrative while the 3-month and 6-month trend tells the real story. A single hot CPI print in a disinflationary trend is noise; treating it as a regime change causes whipsaw.
$30K-$300K in whipsaw losses from overreacting to single data points. Macro funds that trade every NFP/CPI as a regime change underperform those that smooth over 3-6 months.
Always compute 3-month and 6-month moving averages. Don't change the macro regime classification on a single data point unless it's confirmed by 2+ other indicators.
Proactive Triggers
#
Trigger Condition
Auto-Response
P1
Macro data point cited without consensus → "CPI at X%" with no comparison
[FIX] Add: [ACTUAL] vs [CONSENSUS] vs [PRIOR]. Tag release date.
P2
Regime declared from single indicator → "We are in a recession because PMI..."
[STOP] Require 2+ independent indicators before regime classification.
P3
"The Fed will..." prediction → no market-implied probability cited
Macro scenario analysis, recession probability, capital markets outlook
Board presentations, strategic planning, capital allocation decisions
Error Recovery
Symptom
Root Cause
Fix
Lesson
Portfolio positioned for recession but no recession materializes
Overweighted yield curve inversion signal without considering lag variability
Size for a probability distribution, not a binary event. Use a "recession watch" allocation (reduced risk, not eliminated) until hard data confirms.
Recession signals have wide timing error bars. Trade the probability, not the certainty. Scale positions gradually as confirming evidence accumulates.
Cross-asset hedge fails during stress event
Correlation assumed based on long-term average rather than current regime
Check rolling 90D correlations before constructing hedges. In inflation regimes, use commodity overlays and TIPS instead of nominal bonds for equity hedges.
Correlation is regime-contingent. The hedge that worked last cycle may amplify losses this cycle. Update correlation assumptions quarterly.
Macro view correct but timing wrong
Confused direction (6-12 month view) with timing (this week/month)
Layer your macro view: structural (3-5Y), cyclical (6-18M), tactical (1-3M). Only tactical views drive entry timing. Use technical confirmation for entries within the macro direction.
Macro gives you direction, not timing. Use shorter-horizon tools (technicals, positioning, flows) for entry timing within the macro framework.
Central bank surprise causes portfolio loss
Treated market-implied probability as the only scenario
Always model: baseline (market-implied), hawkish surprise, dovish surprise. Size positions so that a 1-standard-deviation surprise doesn't cause outsized losses.
Central banks surprise markets ~30% of the time. Position sizing must account for the distribution, not the point estimate.
What Good Looks Like
Good — Macro Regime Update:
"Regime: GOLDILOCKS [CONFIDENCE: MEDIUM]. Growth: ISM Manufacturing 50.9 [ACTUAL] vs 49.5 [CONSENSUS], expanding. ISM Services 53.8 [ACTUAL] vs 52.0 [CONSENSUS], expanding. Inflation: Core PCE 2.6% YoY [ACTUAL] vs 2.7% [CONSENSUS], trending down; 3-month annualized 2.1%. Policy: Fed Funds 4.25-4.50% vs r-star ~2.5% [Holston-Laubach-Williams estimate] → restrictive. FCI loosening: GS FCI at 99.2, down from 100.5 3 months ago. Risk-on/risk-off composite: +5 (RISK_ON). Asset allocation overlay: overweight equities, overweight credit, underweight cash. Key risk: inflation re-acceleration from commodity price pass-through. Next data check: CPI [DATE], FOMC [DATE]."
Bad — Vague Macro Commentary:
"Macro looks pretty good right now, growth is solid and inflation is coming down. The Fed will probably cut rates soon. Stay long risk assets."
Verification Guardrails
Before delivering macro analysis, verify:
Every economic data point cited with [ACTUAL] vs [CONSENSUS] vs [PRIOR] and release date
Regime classification based on 2+ independent indicators
Central bank forecasts include market-implied probability and acknowledge ~30% surprise rate
All forecasts >3 months carry explicit confidence intervals
Cross-asset correlations tagged with current regime context
"Good" vs "Bad" examples provided for macro update format
Liquidity analysis includes balance sheet change rate, not just direction
Risk-on/risk-off score computed from all 6 dimensions
Scenario probabilities sum to 100% with asset-level impacts documented
Cross-skill coordination table populated with upstream and downstream skills
Deliberate Practice
Exercise 1: Regime Classification Drill (15 min)
Take the most recent ISM Manufacturing, ISM Services, CPI, NFP, and GDP data. Classify the current regime using the 2x2 growth-inflation matrix. What's the confidence level? What indicator would make you change the classification?
Exercise 2: Central Bank Reaction Function (20 min)
For the next FOMC meeting: compute the Taylor Rule implied rate. Compare to the current rate. Read the last meeting minutes — count "some" vs "most" vs "all participants." What's the hawk/dove dispersion in the dot plot? Model the probability distribution for the next decision.
Exercise 3: Risk-On/Risk-Off Composite (15 min)
Score the current environment on all 6 RO/RF dimensions (liquidity, real rates, credit spreads, economic surprises, vol, dollar). Compute the composite. Does it agree with the VIX level? If not, what's the divergence telling you?
Exercise 4: Intermarket Ratio Analysis (10 min)
Plot SPY/TLT, Cu/Au, and HYG/TLT over the last 90 days. Are they all pointing in the same direction? What's the single ratio you'd watch if you could only watch one?
Exercise 5: Scenario Stress Test (20 min)
Build a 3-scenario macro outlook (baseline 60%, upside 20%, downside 20%). For each, define the GDP path, inflation path, and Fed path. What's the equity market impact in each? What's the single best hedge for your downside scenario?
Verification
[Data Timestamp] — Verify every macro data point (GDP, CPI, PMI, rates) includes the specific release date or [AS OF YYYY-MM-DD] tag.
[Calculation Transparency] — Verify any cross-asset ratio, regression, or scenario projection shows the calculation steps and raw inputs.
[Regime Classification Evidence] — Verify the regime call cites specific indicators (yield curve slope, ISM vs services PMI, credit spreads) rather than general claims.
Pass criteria: All checks pass before delivering output.
Anti-Hallucination
Admit uncertainty. If you cannot determine the correct approach, ask — do not guess.
Flag your knowledge cutoff. If this project uses tools or patterns you have not seen, state your assumptions.
Never guess security. If work touches auth, payments, or PII, route to security-reviewer.
[VERIFIED] — Confirmed against official documentation or published standards
[COMMON-PRACTICE] — Widely used in the industry
[INFERRED] — Reasonable extrapolation from general principles
[UNKNOWN] — Requires verification against specific context
