Permasalahan yang dihadapi P.T. Bank X adalah belum digunakannya pengukuran capital charge yang lebih akurat alas risiko transaksi valas yang diakibatkan oleh fluktuasi volatilitas return sharf. Digunakannya Metode Standar dengan tarif 8% dari Posisi Devisa Neto yang dimiliki, mengakibatkan besarnya penyediaan modal (capital charge) lebih besar jika dibandingkan dengan kebutuhannya sehingga dapat merugikan bank.Metodologi penelitian ini adalah: mengumpulkan data kurs tengah harian dan Posisi Devisa Neto P.T. Bank X pada mata uang USD, SGD, EUR dan SAR. Periode penelitian selama 335 hail, confidence level VaR 95% dan holding period 1 had (spot). Menghitung return harian masing-masing kurs, melakukan uji stasionaritas, bila tidak stasioner di- derencing. Bila data stasioner dilakukan uji normalitas dan uji heteroskedastik. Aka data heteroskedastik forecasting volatilitas menggunakan EWMA dan LARCH. Dui model yang lebih valid dilakukan perhitungan VaR untuk menentukan capital charge. Kemudian dibandingkan dengan Metode Standar yang digunakan dalam pengukuran capital charge P.T. Bank X dan diuji validitasnya dengan backtesting Kupiec Test.Untuk pengukuran capital charge risiko shad, balk Model Standar maupun Model GARCH keduanya dapat digunakan karena berdasarkan perhitungan validasi dengan backtesting keduanya menunjukkan basil yang valid, tetapi capital charge Model GARCH lebih efisien. Untuk tujuan efisiensi dan efektivitas usaha, Model GARCH perlu diterapkan di P.T. Bank X.

---

The problem which is faced by P.T. Bank X is has not been used of measuring capital charge which is more accurate to the risk of valas transaction which is caused by the return sharf volatility fluctuation. The used of Standard Method with a value 8% from the owning of Net Open Posistion, cause the amount of capital charge is bigger if it is compared with the needs, so that it causes bank. loss.The research methodology are collecting daily exchange rate data and Net Open Position of P.T. Bank X for some currencies USD, SGD, EUR and SAR. The research period run in 335 days, confidence level VaR 95% and holding period I day (spot). Count the daily return in each exchange rate, doing stationary testing, if it cannot be stationer, it can be dif3:'erencing. If the stationer data is done such normality testing and heteroscedastic testing. If the data is heteroscedastic so forecasting for volatility use EWMA and LARCH from model which is more valid is done of counting VaR to determine the capital charge. Then it is compared with Standardized Method which is used in measuring capital charge P. T. Bank X and the validity is tested by backtesting Kupiec Test.For the measuring of capital charge for risk foreign exchange, either Standardized Model or LARCH, both of them can be used because based on the counting validation with backtesting, they show the valid result, but the capital charge with LARCH is more efficient. For the aim of efficiency and effectiveness of a bussiness, LARCH Model can be applied in P.T. bank X.