Three years of live trading
RobinVOL was released in July/2012. Now it is a good moment to look back to see what has happened and how RobinVOL has performed.
Backtest without compounding effect
A backtest without compounding effect allows a much easier performance analysis, as recent trading activity hides any problematic periods in the past. To remove the compounding effect the “Balance” axis is configured with a logarithmic scale:
The first thing to notice is that the drawdown period experienced in 2014 is something that had already happened in the past (as already stated in the statistical analysis). The second important thing is that since its public release in 2012 it has been behaving as expected.
Backtest with compounding effect
On a backtest with compounding effect, recent trades have a much bigger effect on the shape of the curve as the lot size on recent trades is bigger (as long as it is a winning backtest). This is the reason why in any compounding backtest, recent activity usually seems to be more volatile.
In this image we can see the period used to research default settings. Only data up to 2009 was used for this. Since then, there has been 5 years of out of sample and live trading with very good results. It is interesting to see that the drawdown period in 2014, on a backtest with compounding effect seems to be much more important than the one that happened in 2007, but the truth is that both periods are very similar as you can see on the previous image.
What can we expect in the future?
RobinVOL behavior has been very good on unseen data, mixing very good winning periods with periods of plain or light drawdown that lasts for some months in the worst case. Obviously nothing guarantees that the future will be the same, but considering that RobinVOL survived to many different market conditions and is currently at new equity highs, nothing points to a change in behavior in the near future.