Linus Svensson, present his master's thesis
	
Contrarian Trading Strategies on Extreme Price Changes in the Foreign
Exchange Market
 
Extreme price changes are examined for high-frequency data in the
foreign exchange market with events being identified by estimating
distributions empirically. The relationship with factors such as
fundamental news and jumps in the price process is evaluated. Jumps
are formally detected by utilizing the nonparametric volatility
measures realized variance and bipower variation. Following the
hypothesis that an extreme move should have a subsequent reversal
effect, a trading model is developed that follows such a contrarian
trend. Model parameters are found through a combination of historical
statistical analysis and economic intuition. The model acts on an
intraday scale and generates economically significant returns for both
in- and out-of-sample data. Evaluating trading results, it is
suggested that trades related to both news and a jump are more
profitable.