Lisa-Maria Bieker Ironhack Data Analysis
The aim of this project was to disprove the Random-Walk-Theory. Therefore, three trading strategies were checked with different forex trading pairs (daily and minute data).
The following strategies were tested:
- Red green candle
- Reversed candle
- Moving average crossover
Trading pairs: USDJPY, EURUSD, GBPUSD, USDCAD
Presentation of the project was done with Google slides: link
- Jupyter Notebook
- Python: Pandas, Numpy, Seaborn, Matplotlib
- MySQL
- Tableau
- Statistical analysis
Data was used from these sources:
- Daily data
- Daily and minute data --> account needed to get data
All datasets contained the following columns:
- Date/time
- Open
- High
- Low
- Close
- Volume
- dropping of not required columns (volume etc.)
- conversion of data in ticks
- calculation of difference open - close prices per row
- calculation of classification and order type columns (long, short, none) depending on strategy per row
- calculation of profit per row
- preparation for figures: calculation of cumulative profit per row
- calculation of how many standard deviations result of strategy is away from expected profit value of 0 --> significance level for accepting strategy as valid is 3.9 standard deviations
- analysis of number of trades vs cumulative profit: Tableau and Matplotlib
Adapted red green strategy in combination with USDJPY minute data resulted in a profitable strategy with a standard deviation of 5.22. Under the given significance level and by usage of this strategy and dataset the Random-Walk-Theory could be disproved.