Mean-variance preference and CAPM.docVIP

Mean-variance preference and CAPM 1. Hypothetical Worlds of Finance Assets and Portfolios Firstly, we define a simple world: There is a finite set of states representing uncertainty: , and there is a finite set of assets . We also have a vector of current asset price Thereof, we have a payoff matrix, in dollars Thus, we may define a return matrix, in percentage, as Secondly, we define a portfolio consisting units of assets from this simple world. To do so, we need 1) units of each assets in the portfolio, 2) amount invested in each asset, 3) relative portfolio weight in each asset (the portfolio weight must be complete, i.e. sum up to 1) Thus we define 1) 2) dollar amount invested in each asset as units of this asset times the current price of the asset, or . Also we may define the payoff of such an asset as , where is the payoff in state 1 and so on. In reality often we have budget constraint, w, at any given time so that the total amount invested may not exceed this budget constraint. Thus we have, and the weight of individual asset is . (a negative weight signifies a short position). According to 3) above, the weights must sum up to 1, . Note: Portfolio weights are often manipulated to find optimal portfolios. Dollar amount invested are often used to find arbitrage opportunities. 2. CAPM (Capital Asset Pricing Model) We have already defined the expected return vector of asset i, We may now define the variance-covariance matrix of the expected return as: We can see that such a matrix is symmetric on both sides of diagonal, on which there is the collection of variances. The determinant of this matrix must be positive. Also we have a vector of portfolio weights and Thus we may now define the expect portfolio return as , which is a linear function, and portfolio variance as , which is a quadratic function. Optimal Portfolio that Minimizes Variance This is a special case in which the agent is variance