Turnover Risk Measure Constraints

PortfolioOptimisers.set_risk_constraints! — Method
set_risk_constraints!(
    model::Model,
    i,
    r::TurnoverRiskMeasure,
    opt::Union{AbstractProgrammeAllocationSet, RiskJuMPOptimisationEstimator},
    ::AbstractPriorResult,
    args...;
    prefix,
    kwargs...
) -> VariableRef

Add the epigraph of the turnover of a TurnoverRiskMeasure to the model, and return its risk variable.

One norm-one cone bounds the risk variable from below by the $L^1$ distance of the weights from the reference weights r.w, scaled by $k$. The risk expression of the measure is that variable. Outside a ratio problem $k = 1$. In a ratio problem the weights are homogenised, and the row bounds $k$ times the turnover of $\boldsymbol{w} / k$. The model reads no moment, so it accepts any prior result and does not read it.

JuMP formulation

Variables

  • w: read, $\boldsymbol{w}$.
  • k: read, $k$.
  • turnover_risk_<i>: created, $t$.

Expressions

  • turnover_r_<i>: $\boldsymbol{w} - \boldsymbol{w}_{0} k$.

Constraints

  • cturnover_r_noc_<i>: $(s_c t,\, s_c (\boldsymbol{w} - \boldsymbol{w}_{0} k)) \in \mathcal{K}_{1}$, which is $t \geq \lVert \boldsymbol{w} - \boldsymbol{w}_{0} k \rVert_1$.
  • turnover_risk_<i>_ub: $s_c (t - u k) \leq 0$, when r.settings.ub is a number.

set_risk_bounds_and_expression! registers the bound row and adds $t$ to risk_vec when r.settings.rke is true. Each name carries the index i of the measure and the namespace prefix.

Where:

  • $t$: Epigraph variable of the turnover, the risk of the measure.
  • $\boldsymbol{w}$: Portfolio weights vector $N \times 1$.
  • $\boldsymbol{w}_{0}$: Reference portfolio weights vector $N \times 1$, the weights from which the turnover is measured.
  • $k$: Variable that scales the budget in the homogenised problem.
  • $s_c$: Constraint scale. It multiplies both sides of a row, so a positive value leaves the feasible set unchanged.
  • $\mathcal{K}_{q} = \{(t, \boldsymbol{x}) : t \geq \lVert \boldsymbol{x} \rVert_{q}\}$: Norm cone of order $q$. $\mathcal{K}_{2}$ is the second-order cone.
  • $u$: r.settings.ub.

Relaxation

The encoding is not exact: the entries below bound the quantity instead of reproducing it, and the bound is tight only under the condition stated here.

  • The cone bounds $t$ from below only, so $t$ is the turnover only when the row is tight.
  • When the objective minimises the risk, the row is tight. A risk that the objective does not minimise can lie above the turnover, but the bound row then still holds the turnover, because the turnover is at or below $t$.

Arguments

  • model::JuMP.Model: The JuMP optimisation model.
  • i: Constraint index for unique variable and constraint naming.
  • r::TurnoverRiskMeasure: Turnover risk measure. Its w holds the reference weights.
  • opt::RiskConstraintOwner: The owner of the risk constraint, a risk-based JuMP optimisation estimator or a programme Allocation Set.
  • ::AbstractPriorResult: Prior result. The model does not read it.

Keyword arguments

  • prefix::Symbol: Model State namespace of the build (default: empty, i.e. the bare key).

Returns

  • turnover_risk: The epigraph variable $t$.

Related

source