Forecasting Portfolio Risks in Futures Markets Using the GARCH-VaR Model

With the increase of financial transactions in recent decades, forecasting risk of financial products has become more and more challenging, though it's of great significance to financial companies and investors. Fortunately, the accumulation of huge valuable records from trading platforms has provided strong support for solving this issue. In this paper, we proposed a GARCH-VaR model for forecasting the risk of investors' portfolios. We consider both of the volatility of futures and the available fund for each account in GARCH-VaR. Specifically, we first preprocessed settlement prices and holding positions data, and then labeled the samples with whether extreme losses occur. In the second step, we calculated the volatility of all futures and their correlation coefficient matrix. Finally, we forecasted the risk probability of investors' daily positions. Experimental results on the transaction data of the futures company showed the effectiveness of our proposed approach.