This is the delta-gamma-vega-theta approximation pnl for option portfolio. It is less compute resource consuming than fully revaluation the portfolio and caculate pnl. Is it the only advantage of this method? We could search much papers about "delta-gamma-theta approximation" in google. I'm a bit confused that why vega was missed in these papers, as vega is a very important risk factor for option. And also it would be more complicated if take volatility dynamic risk into account for pnl. Seems the local volatility model (Dupire) is more used for FX option market, while SABR is more used for rates option. Not sure misunderstanding of this.