The implementation of statistical arbitrage strategy on the JSE top 40 index option

Abstract

This study implemented the statistical arbitrage strategy on financial time series. This strategy is known as an extended horizon trading prospect used for generating profit that is riskless and is further designed for exploiting tenacious incongruities. The study used statistical arbitrage strategy to explore its efficiency on selected stock markets in pricing relative risk in highly correlated and cointegrated stock options traded on the Johannesburg Stock Exchange (JSE). Statistical arbitrage avoids the dual premise dilemma associated with traditional tests of market efficacy. This is due to autonomous description of model symmetry and the fact that the said strategy has a mismatch with market competency. The study explored the methodology for investigating statistical arbitrage and provided the findings that determined whether or not value trading policies constitute opportunities for statistical arbitrage. The data used are related to four sectors: the financial, industrial, mining and retail that trade on the JSE. These data are for stock market comprising closing prices of option stock for 40 companies of the four sectors traded on the JSE. The data was collected daily from 04 January 2010 to 31 December 2015. A total of 1500 observations with one year approximately having 250 trading days excluding weekends and public holidays was used. The study followed the methodology of two broad components: simple rolling regression and Engle and Granger cointegration technique. A two-step approach, namely correlation analysis and cointegration technique, were used in formulating potential pairs portfolio on the basis of the profitability indicator evaluated in-sample. The findings from correlational analysis revealed thirty nine pairs under financial sector, two pairs under industrial sector, five pairs under mining sector and six pairs under retail sector. The selected pairs were stationary at first log difference. The findings of cointegration analysis confirmed only eighteen pairs under financial sector, one pair in industrial sector, two pairs in mining sector and three pairs in retail sector. Since the results revealed a mean reversion, statistical arbitration strategy was implemented. The findings revealed that the financial sector perfom1ed better than the other three sectors. The implementation of pairs trading to historical volatility revealed several possibilities of trade in the financial sector. Therefore, the investors profit more from trading the identified pairs of stocks in the financial sector. The study contributes to the existing literature in historical volatility modelling and stochastic volatility in the field of index option in emerging markets. This study further provided a base for future researchers conducting studies on emerging markets, more specifically in South African context.