计算风险价值VaR

function [valueAtRisk simulatedPortfolioReturns] = monteCarloVar(securityReturns, securityPrices, portfolioMarketValue, portfolioWeights, numSim, varargin) if isempty(varargin{1}) method = 'portsim'; else method = varargin{1}; end % The number of observations, in our case we look 1 day ahead numObs = 1; % Expected mean and covariance of the returns expReturn = mean(securityReturns); switch method case 'portsim' expCov = cov(securityReturns); % % Simulate the returns of the portfolio numSim times % simulatedAssetReturns = portsim(expReturn,expCov,numSim,1,numObs, 'Exact'); % % % Make the return series two dimensional % simulatedAssetReturns = squeeze(simulatedAssetReturns); % simulatedPortfolioReturns = simulatedAssetReturns*portfolioWeights'; % Simulate the returns of the portfolio numSim times simulatedAssetReturns = portsim(expReturn,expCov,numObs,1,numSim, 'Exact'); % simulatedAssetReturns = portsim(expReturn,expCov,numObs,1,numSim, 'Expected'); % Make the return series two dimensional simulatedAssetReturns = squeeze(simulatedAssetReturns); simulatedPortfolioReturns = simulatedAssetReturns'*portfolioWeights'; case 'econometrics' sigma = std(securityReturns); correlation = corrcoef(securityReturns); X = securityPrices(end,:)'; dt = 1; switch varargin{2} case 'gbm' % Create the GBM object GBM = gbm(diag(expReturn), diag(sigma), 'Correlation', correlation, 'StartState', X); % Simulate for numSim trials [simulatedAssetPrices,T] = GBM.simulate(numObs, 'DeltaTime', dt, 'ntrials', numSim); %#ok simulatedAssetReturns = tick2ret(simulatedAssetPrices, [], 'continuous'); simulatedAssetReturns = squeeze(simulatedAssetReturns); simulatedPortfolioReturns = portfolioWeights*simulatedAssetReturns; case 'sde' % Create the drift and diffusion functions F = @(t,X) diag(expReturn) * X; G = @(t,X) diag(X) * diag(sigma); % Create the SDE object SDE = sde(F, G, 'Correlation', correlation, 'StartState', X); %Simulate for numSim trials [simulatedAssetPrices,T] = SDE.simulate(numObs, 'DeltaTime', dt, 'ntrials', numSim); %#ok simulatedAssetReturns = tick2ret(simulatedAssetPrices, [], 'continuous'); simulatedAssetReturns = squeeze(simulatedAssetReturns); simulatedPortfolioReturns = portfolioWeights*simulatedAssetReturns; end end valueAtRisk = -prctile(simulatedPortfolioReturns*portfolioMarketValue, [1 5]);