Economic structure
12named factors
A macro view of market risk
Market-derived proxies cover equities, rates, currencies, credit, commodities and volatility, including the shape of the rates curve and regional equity exposures.
Risk
Portfolio construction,risk and capital.
Economic and latent factors connect forecasts to portfolio exposures. Scenario analysis examines how drawdowns, margin and changing risk conditions affect capital requirements.
Explore the capital framework01 / Understand the exposures
Risk model & exposures
Different markets can express the same underlying risk. The model combines interpretable economic factors with latent components to examine shared exposures, diversification and the risk left unexplained.
Latent structure
3PCA components
Common risk beyond the named factors
Principal component analysis extracts shared movement from the returns remaining after economic-factor effects have been estimated.
Specific risk
The remaining asset-level variation
Residual variance completes the covariance estimate. Portfolio analysis separates contributions from common factors and asset-specific risk.
Inside the risk modelFactor definitions, rolling exposures and covariance estimation
Economic factors
The configured set comprises Equity, Rates, Rates Slope, Rates Curvature, USD, Credit, Energy, Metals, Agriculture, Europe, Japan and Volatility. These are market-based measures of economic risk.
Exposure estimation
Rolling ridge regressions estimate exposures. Named factors are residualised in a defined order; PCA then identifies common structure in the economic-factor residuals before the combined model is fitted.
Covariance & specific risk
Factor volatility and correlation are estimated separately, with shrinkage applied to correlation. Asset covariance combines factor exposures, factor covariance and diagonal residual variances.
Factor definitions, estimation windows and latent components influence the result. PCA components describe statistical structure; their economic interpretation can change through time.
02 / Shape the portfolio
Portfolio construction
Forecasts express the opportunity; the risk model reveals how positions interact. Construction examines the trade-offs between expected return, concentration and exposure, with portfolio composition and capital allocation treated as distinct decisions.
What to hold
Translate forecasts into relative weights. Assess the opportunity alongside asset concentration, factor risk, diversification and gross exposure.
How much to hold
Examine the scale that the portfolio can support under drawdown, margin and risk scenarios, with explicit capital assumptions and policy buffers.
Explore construction & diagnosticsStaged optimisation, concentration and allocation trade-offs
Preserve the opportunity
A staged construction method starts with forecast return under a variance constraint. Later stages seek less weight concentration, factor risk, factor concentration and gross exposure while retaining specified qualities of earlier solutions.
Inspect the risk allocation
Diagnostics examine factor exposures and risk contributions, common versus specific risk, effective breadth, diversification, netting and gross and net exposure.
Separate direction from scale
A concentrated opportunity can warrant a smaller allocation. Risk and capital analysis can inform portfolio scale while preserving the underlying forecast ranking and composition.
03 / Translate into instruments
Implementation & liquidity
A research exposure must be mapped into the instruments held. Implementation studies examine how underlying contracts, shared positions, trading conditions and margin affect the portfolio that can be expressed.
Contracts & shared legs
Map relative-value positions into their underlying futures. Examine overlapping legs, portfolio netting and the effects of whole-contract sizing.
Liquidity & trading costs
Study volume, open interest, intraday participation and settlement-window conditions alongside turnover and cost assumptions.
Margin & exposure
Relate gross notional exposure to modeled margin demand. Carry the margin assumptions into the capital analysis alongside the return paths.
Explore the implementation assumptionsOutright futures, return definitions and margin proxies
A common risk basis
The current risk-model configuration covers an outright futures investment universe across equities, rates, currencies, energy and metals. Relative-value exposures can be examined through their underlying legs.
Consistent experiment definitions
Position timing, return units, contract sizing and costs need to describe the same experiment. Changes to these assumptions can alter apparent diversification and portfolio outcomes.
Modeled margin demand
The current capital studies use per-instrument margin proxies against absolute exposure. These estimates are scenario inputs, rather than broker-specific margin calculations.
04 / Examine capital resilience
Scenarios & capital resilience
Volatility describes one dimension of risk. Capital research also examines losses along the path, margin demand and the sensitivity of portfolio decisions to different risk conditions.
Survival-capital research
Drawdown and margin,
evaluated together.
Historical replay and block-bootstrap simulations examine combined capital demand along each path. The timing matters: a drawdown and a margin requirement can place pressure on capital at the same moment.
- At each pointDrawdown + modeled margin
Measure both demands at the same point in the scenario.
- Within each pathPeak combined demand
Retain the largest combined requirement encountered.
- Across the scenariosA distribution of capital needs
Examine tail quantiles, safety buffers and exposure capacity.
The calculation uses contemporaneous demand, rather than adding a path’s separate maximum drawdown and maximum margin.
Return paths
Historical replay & simulation
Replay contiguous history and sample blocks of returns and margin together. Compare portfolio variants using a consistent set of capital assumptions.
Risk states
Different market conditions
Select historical model snapshots spanning measured risk conditions. Re-examine portfolio composition and its capital implications under those states.
Model sensitivity
Risk-model perturbations
Vary factor volatility, residual risk, correlations and exposures. Re-solve portfolio weights to examine how strongly decisions depend on the estimated model.
Inside the capital & scenario frameworkPath construction, risk perturbations and interpretation
Drawdown & margin paths
Return and margin histories use the same sampled observations. Portfolio weights remain fixed within each path; modeled margin aggregates absolute weights against instrument margin fractions.
Path requirements can be pooled across supplied portfolio variants. Selected tail quantiles, safety multipliers and capital floors inform a modeled gross-exposure limit.
Ten model perturbations
Paired scenarios increase or decrease factor volatility, residual volatility and overall volatility; move factor correlations towards or away from their shrinkage target; and shrink or expand exposures.
Rebuilding covariance and comparing the resulting portfolio directions reveals model sensitivity. These are deterministic diagnostic scenarios.
Reading the results
Historical risk states, resampled return paths and model perturbations answer different questions. Results depend on the history, fixed-weight assumption, margin proxies and scenario choices.
Survival capital is a modeled buffer estimate. It is not a guarantee of sufficient funding in every future market condition.
05 / Challenge the estimates
Monitoring & review
Risk estimates change as markets and exposures evolve. Comparison and calibration diagnostics examine whether the model still describes the portfolio, and how much confidence to place in its use for construction and scale.
Model agreement
Compare covariance, correlations, exposures and the balance of common and specific risk across model variants.
Calibration & diversification
Examine realised versus estimated volatility and diversification, including shortfalls relative to lagged model expectations.
Portfolio stability
Inspect weight dispersion under model perturbations, factor concentration and changes in effective breadth.
Explore diagnostics & reproducibilityRisk snapshots, model comparisons and research review
Recorded model states
Risk snapshots retain exposures, factor covariance, specific variances, dates and run metadata. This provides a defined basis for comparisons and downstream portfolio studies.
Separate diagnostic views
Model disagreement, calibration, diversification and optimisation stability provide distinct perspectives. Research examines how these measures might inform portfolio scale.
Review in context
Diagnostics are interpreted alongside the forecast, investment universe, estimation history and capital assumptions. The current work develops and evaluates these methods within the research platform.