Risk-On and Risk-Off: A Cross-Asset Regime Diagnostic

Diagnose risk-on and risk-off behavior with synchronized equity, breadth, credit, volatility, rates, currency, funding, and liquidity evidence while separating the underlying shock from an informal market label.

For educational purposes only; not investment advice. Investing may result in loss.

Direct answer

Risk-on and risk-off are informal, retrospective descriptions of changing cross-asset co-movement and risk-bearing conditions. They are not securities, official regimes, causal explanations, or mechanical trading signals. Risk-on usually describes broader willingness or capacity to hold equity, credit, liquidity, duration, currency, commodity, or leverage risk; risk-off describes a contraction in that willingness or capacity.

No single market defines the label. Diagnose a dated interval using synchronized total returns, equity breadth, option-implied volatility, option-adjusted credit spreads, nominal and real rates, inflation compensation, currencies, funding, leverage, and liquidity. Then identify the shock. A disinflationary growth scare and an inflation shock can both hurt equities while moving government bonds in opposite directions.

Mechanism and measurement

Freeze the observation. Specify the information cutoff t₀, end time t₁, time zone, session, market holidays, currency, price or total-return basis, hedge status, and data vintage. Do not combine one market’s close with another market’s later reaction and call the result contemporaneous.
Map the shock before the label. Separate growth, inflation, monetary policy, fiscal or sovereign risk, credit, geopolitics, commodity supply, positioning, margin, and market-functioning shocks. Record surprise versus expectation, first mover, transmission path, and plausible alternatives rather than assigning causality from co-movement alone.
Measure equity participation. Use broad and regional total returns, equal- and capitalization-weight indexes, sectors, styles, and point-in-time breadth. With advancers A, decliners D, and eligible constituents U, calculate advance share = A ÷ U and net breadth = (A − D) ÷ U; unchanged, missing, halted, and stale securities need explicit treatment.
Measure credit, volatility, and funding. Use matched-maturity option-adjusted spreads rather than corporate yields alone. Treat VIX as a non-directional, annualized, constant 30-day expectation derived from SPX option quotes, not current fear or a forecast of sign. Add funding spreads, repo, cross-currency basis, issuance, margins, haircuts, dealer capacity, and market depth.
Decompose rates and currencies. Split nominal yields into real-yield and inflation-compensation changes where the data permit, inspect curve and term-premium behavior, and distinguish duration return from credit return. Interpret a currency against a named basket or pair and consider the shock’s geography, dollar-funding demand, policy path, hedge flows, and reserve or safe-asset role.
Normalize without hiding units. For indicator i, history fixed before t₁, mean μᵢ, standard deviation σᵢ, and orientation sᵢ equal to +1 when higher means more stress or −1 when lower means more stress, define zᵢˢᵗʳᵉˢˢ = sᵢ × (xᵢ − μᵢ) ÷ σᵢ. A weighted dashboard can use Z = Σ(qᵢ × zᵢˢᵗʳᵉˢˢ) ÷ Σqᵢ, but must retain components, weights, missing-data rules, winsorization, and version.
Test persistence and portfolio relevance. Compare intraday, 1-day, 5-day, and 20-day windows, rolling correlations, reversals, and event timestamps. Translate the observed moves into actual portfolio contributions, hedge basis, liquidity, financing, and scenario loss; a regime label by itself is neither a forecast nor a position-sizing rule.

A classic broad risk-on pattern can combine positive equity total returns and breadth, narrower credit OAS, lower implied volatility, easier funding, and stronger cyclical assets. A classic risk-off pattern can reverse those signs and increase demand for liquid or safer claims. These are tendencies. Under an inflation or sovereign shock, equities and long-duration government bonds can fall together; under a U.S.-centered credibility shock, the dollar can weaken; during forced deleveraging, assets normally called defensive can be sold to raise cash.

The OFR Financial Stress Index is one example of a documented composite, using credit, equity valuation, funding, safe-asset, and volatility categories. It is a global market-based stress snapshot, not a universal risk-on probability, and its publication lag, revisions, regions, transformations, and component definitions matter. A custom dashboard should not borrow the name or interpretation of a published index without reproducing its methodology.

Worked examples

Broad but provisional risk-on evidence: In a 500-security universe, 375 advance, 100 decline, and 25 are unchanged. advance share = 375 ÷ 500 = 75.0000% and net breadth = (375 − 100) ÷ 500 = 55.0000%. If the equity total-return index gains 2.0%, high-yield OAS moves from 360 bp to 340 bp, VIX moves from 24 to 20, the 10-year nominal yield rises 8 bp, and a defined dollar index falls 0.7%, the synchronized bundle is consistent with broad risk-on behavior. VIX changed by (20 ÷ 24) − 1 = −16.6667%, but neither that calculation nor the label proves persistence or cause.
Inflation-driven risk-off with falling bonds: An inflation surprise is followed by equities at −2.0%, high-yield OAS from 350 bp to 365 bp, VIX from 20 to 25, and the 10-year yield from 4.10% to 4.25%. VIX rises (25 ÷ 20) − 1 = 25.0000% and the yield rises 15 bp. For a bond with modified duration 7.5, the first-order price effect is −7.5 × 0.0015 = −1.1250%. Equities and Treasury prices both fall, yet the wider credit spread and higher volatility remain consistent with risk-off conditions.
Transparent standardized dashboard: Suppose daily equity return is −1.80% versus history μ = 0.05%, σ = 1.20%, so its stress-oriented score is −1 × (−1.80% − 0.05%) ÷ 1.20% = 1.5417. OAS change is +18 bp with σ = 8 bp, VIX change is +4 with σ = 2.5, and a funding spread changes +6 bp with σ = 3 bp; their stress scores are 2.2500, 1.6000, and 2.0000. Equal weighting gives Z = (1.5417 + 2.2500 + 1.6000 + 2.0000) ÷ 4 = 1.8479. That is a model-specific standardized reading, not an 84.79% probability.
Portfolio result is not the regime label: A portfolio starts the interval at 50% equities, 30% investment-grade bonds, and 20% cash. Returns are −3.0%, −1.2%, and +0.1%; beginning-weight contributions are −1.5000 pp, −0.3600 pp, and +0.0200 pp, for portfolio return = −1.8400% before fees and flows. Calling the interval risk-off does not explain security selection, duration, currency, credit, derivatives, intraday rebalancing, or the portfolio’s exact loss.

Risks and verification checklist

Define t₀, t₁, time zone, session, holidays, release timestamps, and whether the analysis was available in real time.
Align cash, futures, options, credit, rates, currency, and commodity observations to comparable clocks; flag stale or asynchronous closes.
Use total return where distributions matter and state price, gross, net, local-currency, base-currency, and hedge conventions.
Freeze a point-in-time equity universe and handle delistings, halts, unchanged securities, missing prices, corporate actions, and index reconstitutions.
Compare capitalization-weighted, equal-weighted, breadth, sector, style, size, regional, and single-stock evidence instead of relying on one headline index.
Use option-adjusted credit spreads with matched government curves; separate spread change from risk-free-rate and duration effects.
Distinguish investment-grade, high-yield, leveraged-loan, sovereign, emerging-market, and structured-credit composition and liquidity.
Record VIX level, change, 30-day horizon, option-quote timestamp, term structure, skew, and the difference from realized volatility and tradable VIX derivatives.
Decompose nominal yields into real yields and inflation compensation when possible; inspect curve, term premium, supply, auctions, and policy expectations.
State whether bond movement is a yield change, price return, or total return and use duration and convexity appropriate to the instrument.
Name each currency pair or basket, weights, fixing, session, and return direction; avoid treating every dollar move as the same signal.
Review secured and unsecured funding, repo, commercial paper, cross-currency basis, margins, haircuts, collateral, redemptions, and dealer balance sheets.
Measure liquidity with bid-ask spreads, depth, price impact, failed trades, issuance, ETF premiums or discounts, and executable size rather than volume alone.
Separate growth, inflation, policy, fiscal, sovereign, credit, geopolitical, commodity, positioning, and mechanical-flow hypotheses.
Keep raw units beside standardized scores and freeze lookback, frequency, mean, volatility, sign, weights, caps, and missing-data policy.
Avoid look-ahead and revision bias in composite indexes, macro releases, constituent histories, credit data, and vendor transformations.
Test 1-day, 5-day, and 20-day persistence, rolling correlations, lead-lag sensitivity, event exclusions, and alternative endpoints.
Distinguish description from prediction; validate any forecast out of sample with turnover, slippage, financing, borrow, taxes, and decision latency.
Map observations to actual holdings, derivatives, currency hedges, duration, credit, leverage, liquidity, and counterparty exposures before acting.
Version the data snapshot, code, thresholds, overrides, label, confidence, competing explanations, and later outcome so the diagnosis is reproducible.

Common misconceptions

“Risk-on means stocks rose; risk-off means stocks fell.” The label describes a broader and imperfect cross-asset pattern, not one equity return.
“Treasuries and the dollar always rally in risk-off.” Inflation, supply, sovereign credibility, shock geography, and funding flows can reverse either tendency.
“VIX is a fear or direction index.” It is a methodology-defined, non-directional measure of annualized 30-day SPX option-implied volatility.
“A standardized stress score is a probability.” A z-score depends on the chosen history, transformation, orientation, weights, and distribution.
“Once identified, a regime predicts the next return.” The label summarizes a dated sample and can reverse before a trade is executed.

Authoritative sources

Financial Stability Report - Federal Reserve framework and evidence on valuation, borrowing, leverage, funding, and near-term vulnerabilities.
VIX Methodology - Cboe definition and calculation conventions for constant-maturity SPX option-implied volatility.
Financial Stress Index - OFR description, publication lag, regions, categories, and downloadable global stress data.
Financial Stress Index Indicators - OFR component series, transformations, and category mapping.
ICE BofA US High Yield Index Option-Adjusted Spread - Federal Reserve Bank of St. Louis access to the high-yield OAS series and metadata.
Interest Rate Statistics - U.S. Treasury nominal and real yield-curve data and methodology resources.
Foreign Currency Funding Risk and Cross-Border Liquidity - BIS analysis of foreign-currency liquidity shortages and global dollar funding risk.
Global Financial Stability Report - IMF surveillance of global market, credit, leverage, liquidity, and financial-stability conditions.