MATLAB: Can I make global leverage constraints on a portfolio using a Portfolio object

Financial Toolbox

I would like to set a leverage constraints on the entire portfolio (max sum of positive weights, max sum of negative weights). However, using the MATLAB function "setBounds" functionally distributes the desired leverage boundary value to every asset, allowing the portfolio leverage to achieve upper bounds equivalent to 'desired leverage' * 'n assets'.

More generally, I would also like to be able to set group level upper and lower bounds such that the max sum of positive weights can be bounded by some value U, and the max sum of negative weights is >= L.

How can I do this?

Best Answer

If the initial portfolio is zero (i.e. InitPort = 0 or []), the global leverage constraints can be set as follows.

Instead of using "setBounds", use "setOneWayTurnover". The syntax for this is:

obj = setOneWayTurnover(obj,BuyTurnover,SellTurnover,InitPort,NumAssets);

The global constraint can be implemented by setting the "BuyTurnover" as the max of sum of absolute values of positive weights and "SellTurnover" as max of sum of absolute values of negative weights. Again, the "InitPort" has to be zero or empty in this case. This is because the current version of the portfolio class only allows construction of the leveraged portfolio when the initial portfolio weights are zero.

Please refer to the following documentation on "setOneWayTurnover".

https://www.mathworks.com/help/releases/R2016a/finance/setonewayturnover.html

Related Solutions

MATLAB: Does the Portfolio class have an equivalent to ABS2ACTIVE (in PORTOPT) to set constraints in the Financial Toolbox 5.0 (R2012b)

There are two approaches to this problem:

Method 1: Using the ‘setInequality’ method.

If you have A and b matrices such that "Ax <= b" with portfolio "x" (i.e. weights) and a reference portfolio "x0", the active constraint has the form

     A*y <= b - A*x0

where "y" are the active portfolio weights. These must satisfy a budget constraint defined as:

     sum(y) = 0

The sum of the active weights "y" is zero because y = x - x0, and the sum of the weights for both "x" and "x0" is 1.

The constraints can be set up directly in the Portfolio object with

    p = Portfolio;
    p = p.setInequality(A, b - A*x0);
    p.setBudget(0, 0);

and then you set up your additional information to form a proper portfolio optimization problem (means, covariances, etc.)

Method 2: Using the output of ABS2ACTIVE

When you use the output from ABS2ACTIVE to determine the constraints, the procedure would be

    p = Portfolio;
    c = abs2active([A, b], x0);
    p = p.setInequality(c(:,1:end-1),c(:,end));
    p = p.setBudget(0, 0);

MATLAB: Is it possible to specify per-asset turnover constraints for Portfolio objects in the Financial Toolbox

In R2012a, there is no direct way of specifying per-asset turnover constraints using the 'Turnover' property of the Portfolio object or using the setTurnover method. As you have already noticed, these are intended for average turnover constraints.

However, you could use the individual upper and lower bounds on the portfolio weights to set per-asset limits. For example, if you want asset i to stay within 5% of the current level, set LowerBound(i) = 0.95*InitPortfolio(i), UpperBound(i) = 1.05*InitPortfolio(i).

Best Answer

Related Solutions

MATLAB: Does the Portfolio class have an equivalent to ABS2ACTIVE (in PORTOPT) to set constraints in the Financial Toolbox 5.0 (R2012b)

MATLAB: Is it possible to specify per-asset turnover constraints for Portfolio objects in the Financial Toolbox

Related Question