I hate drawdowns as much as you do, and after touching a 70 something percentage return since the house account started trading in August of 2016, we are going through a bit of a drawdown at the moment.
Can drawdowns be avoided?
If I knew of a rule that would reduce loses, then I would incorporate that rule into AlgoLab's SuperSystem trading system, but unfortunately, I have not found any rule that is beneficial and that passes my statistical tests which is why the AlgoLab house account is never paused or changed. Examples of some of those rules that I have tested include tracking the equity curve itself, and pausing trading when the equity value drops below a moving average of that equity value (ie: when your equity starts to drop below a certain level), using a volatility indicator to pause trading when volatility reaches either above or below a certain value, etc.
This is not to say that pausing your AlgoLab by using your own discretion doesn't work - it's just that I have not found a concrete rule that works consistently and passes my statistical tests. You have probably noticed one of our AlgoLab funded account customers with the alias name "passtime". Passtime seems to have avoided some of our drawdowns in the past, and is avoiding the current drawdown. Although previous tests showed that Passtime's return was no higher than the AlgoLab house account over the same period of time, Passtime's drawdown vs profit ratio was better. Is Passtime's performance is due to luck, or an innate ability to time AlgoLab? This is difficult to answer, and I certainly hope it is due to an innate ability because we may all be able to benefit from overriding the system rules from time to time by using our discretion based on observation of market activity. Only time, and more data will tell.
I thought you might find this interesting. Following is a list of monthly total profits from a SuperSystem backtest from 2007 to now. You can see that the month of August is a typically very profitable month, and January is not. Since none of this data is statistically significant, I cannot incorporate any of this into a rule in SuperSystem.
$100,000 of capital .075 risk backtest from 1/1/207 to 7/26/2017
A good example of the dangers of mistaking casual observations for statistically significant causal factors, is since SuperSystem started trading in Aug of 2016, most of the trades that closed on a Friday, closed at a loss. Sure enough, when I added a rule to the system to close all trades on Thursday afternoon, and to never enter any new trades on a Friday, this greatly improved the profits over the 1 year period since AlgoLab started trading in Aug of 2016. However, it had no effect at all when expanded to the entire 10 year historical backtesting period.