To return or not return - Trend spotting in the Swedish market

Abstract: The following essays checks whether the Swedish stock market (represented by OMXS30) supports the weak efficiency of the EMH by observing whether the stocks contain a unit root and whether one can use trading strategies to create abnormal profits. The AR(1) model of the stock prices have a unit root coefficient very close to 1, thus they are trend stationary but have a very long memory. This model suffers from autocorrelation (AC) and heteroskedasticity, (H) and thus a GARCH(1,1) model and an IGARCH(1,1) were used in order to rectify this. For the GARCH(1,1) model, only a few of the series suffered from AC and H, whilst the IGARCH was basically free from both. When using the strategy of only buying winners, we obtain positive returns for all periods and statistical significance for our 3- , 6- and 36 month periods on a 5% test-level. The results of our 3 month portfolios yields an average return of 3.79% , beating the market by 4,92% and shows a strong statistical significance on the 5 %-level. Our loser portfolios are only profitable during our shortest- (1 month) and longest (36 months) investment horizons and only significant on the latter. The returns on 3- and 6 months are both negative. In comparison to the market, the Sharpe-ratios for all strategies except the 3- and 6 month loser-strategy suggest that they are better investment alternatives