This paper examines the feasibility of modelling and predicting the implied volatility surface of index options. Based on the seven-factor model of Chalamandaris and Tsekrekos (2011), we propose a nine-factor model which explicitly models the option contract moneyness in the corner regions. We find evidence that this model best describes the surface in terms of in-sample fit. Its forecasting performance does, however, fail to improve on previous literature. Both the seven-factor model and our proposed model significantly improve the forecasting accuracy as compared to the model of Goncalves and Guidolin (2006), which has previously been regarded as the best deterministic model for describing the implied volatility surface. We analyse the economic value of our deterministic models and find that they deliver risk-adjusted abnormal returns using a simple delta-hedge trading strategy before transaction costs. Additionally, we show that trading short-term In-The-Money (ITM) call options and Out-Of-The-Money (OTM) put options is most profitable.