Human behavior tends to react in akin ways, hence creating opportunities to implement profitable strategies. Our trading strategy identifies historical similarities in the development of the percentage relation between EUR/USD. By comparing a current pattern with historical ones, the system provides the user with trading recommendations.

The system’s main feature is the prediction algorithm, which averages the future outcomes of the best matching historical patterns. If the relation is expected to increase, a long position is recommended to be taken, and vice versa. In the case of the expected return being less than the calculated risk, the recommendation is to not make a trade.

By testing various parameters for the pattern prediction, the algorithm’s ability to recommend the correct trading decision was optimized. The incorporated parameters affecting the trading analyzing process are the following: measuring if the predicted outcome deviates enough to recommend a trade; and requiring a minimum total similarity between the current investigated pattern and the historical ones.

As of yet, the trading strategy has generated vastly positive returns. During the start of November 2018 until the beginning of February 2019, the daily excess return was 0.11 percent, while carrying a low risk. Furthermore, the beta was negative due to the fact that it hedged itself to the high volatility of the S&P 500 – independently of the index’s movements, the strategy provided consistent returns. At last, the system gave the correct recommendation in almost 65 percent of the performed trades.

Analysts: Alfred Bornefalk and Felix Persson

Historical-Repetitiveness-Alfred-Bornefalk-Felix-Persson-2019_final-1