Hello everyone,
over the last month I've created a FX D1 portfolio. It currently consists of three systems: Two newly developed trendfollowing systems on eurusd and gbpusd and a mean reversion system on eurusd that existed before. The aim was to show the efficiency of creating a system portfolio. Every system had a in sample period of five years and an out of sample period of two years. They have also passed robustness tests like the monte carlo simulation. All results are fixed lot backtests on Dukascopy tickdata.
Eurusd - Trendfollowing
Equity
Statistics
Gbpusd - Trendfollowing
Equity
Statistics
The portfolio of both trendfollowing systems looks promising too. The drawdown is slightly higher than the drawdown of the Gbpusd system but it's lower than the one of the Eurusd system meaning the the larg losses are not highly correlated - but the wins add up:
Equity
Statistics
Eurusd - Meanreversion
Equity
Statistics
In the last step I've added the meanreversion system to the portfolio and compared it to every system and to the trendfollowing system:
Three system Portfolio
Statistics
This portfolio is the most profitable combination of the systems shown here. The wins of every system are added up. Nevertheless the very low drawdown of the meanreversion system also leads to a drawdown than is lower than the one of the trendfollowing systems and the trendfollowing portfolio. This is not self-evidend. The drawdowns could have added up leading to an inefficient portfolio. By combining trendfollowing and mean reversion systems the probability increases that drawdowns / drawdown periods are not correlated or even negatively correlated.
