Optimization of Portfolio of federal loan bonds and RE-PO trades

Abstract

Within the framework of the considered model, the Investor can make transactions for the buying and sailing of federal loan bonds (OFZ), as well as direct and reverse REPO deals secured by OFZ. Transactions are made for liquidity management and increasing interest income. This paper discusses the problem of constructing an optimal portfolio of such transactions. The paper considers the approach for the generation of scenarios for OFZ price changes, the mathematical formulation of the optimization problem, the assessment of its dimension depending on the number of assets and the number of scenarios, numerical experiments on historical data and the construction of an efficient portfolio frontier. The generation of scenarios for OFZ price changes is implemented using historical modeling of a parametric zero-coupon yield curve. The optimization problem criterion is the conditional value at risk (CVAR) risk measure. Constraints on the average return and self-financing of the portfolio are taken into account. As a result, the method of portfolio rebalancing without additional investment, the purpose of which is to minimize risk for a given profitability, is proposed. Numerical experiments are based on historical data about active traded OFZ in 2014-2020. The model is close to the real world: it takes into account commissions, repo discount, bid-ask spreads, trade volumes. Numerical results show that a trading strategy based on the introduced model is more profitable on average than investments in individual OFZ with comparable risk. Note that this effect is shown for strategies with high average profit constraints.