One could trade a portfolio of non-correlated instruments, using same strategy. That would improve the drawdown.

.... and your curve fitting ;-)

Check your backtest and see if it's matching. If it matches, check your overfitting probability

Live by the priciple "your worst draw down is yet to come".

So, do your backtest, take the max dd and multiply it by 2. Adjust your trading size accordingly so you do not reach that level. If you still do (i.e. losing twice as much as the worst in your backtest), stop trading (if your investor hasn't pulled the money) and go back to the drawing board

@Marc ,actually I did come close to twice the backtested drawdown and after about 2 weeks it did recover completely while I just watched ,I only trade 1 lot in each symbol ,so adjusting trade size is not an option ,but adjusting trading capital by using twice the max dd should work in my case

Having success using a "Bayesian" approach to sizing, updating on real-time exits. Key metric is 95%ile/5%ile win/loss (reward/risk), driving partial-Kelly size. A big loss compresses R/R ratio & sizing fairly quickly, only allowing larger size as results envelope improves.

Mithoon, by "adjusting the trading size" I meant the trading size at the start of your live program, not when you are about to hit the maxdd. If you live by the principle "your worst dd is yet to come", then your trading size should be such that you can survive a backtest max dd times 2. I am glad to see you have recovered !!

Depends a lot on the time range and quality of your backtest data and how long you have been trading the system. There is a big difference if it happens after 1 month of trading and you have back-tested your system for 6M, 1 year, 5 years or 20 years.

I would check strategy vulnerabilities and how strategy and market behaved in comparison to your historical data and if there was any fundamental difference. Depending on findings I would then adjust or not adjust trading size accordingly.

>"do your backtest, take the max dd and multiply it by 2."

One should use this rule cautiously since it's results vary drastically in relation to backtest length. I multiply expected annual profit by 2 and take this as probable maximum drawdown. Of course this is also very rough but it prevents me from using strategies with overly high market exposure. Naturally it depends all on the individual strategy and how it generates returns, but many developers claim that their strategy can have high leverage without risk of big drawdown due to "special sauce"™. However, I have noticed the tendency that every XX years "special sauce"™ doesn't work for some obscure reason. ;)

Perhaps a volatility filter would help. Whether it is related to the vix or just a plain vanilla volatility indicator. In my experience, large drawdowns are caused mostly by market action that is out with the relative normal ranges that work best with my systems. Aug 24th 2015 limit down day being an extreme example in the futures/equity markets.

If you hit your Max Draw Down stop trading and go back to the drawing board. Experimenting with as system that just failed makes you vulnerable in your judgement and most people are likely to take more illogical risks in whatever they do to solve the issue. Learn from the experience. Get an emotional distance to your trading system and then start fresh.

Personally, I use also other parameters to check if max dd must end my strategy. For example, frecuency on trading, avg trade etc... that can give u and idea if this DD is due to to a change of market model, or it can be inside of your confort zone. I use calmar ratio as my warning alarm, and then I check the rest of parameters... But of course my Money management of each system inside a portfolio give me time and margin to solve this situations. Have a nice week everyone.

I expect every strategy to hit its historical max DD at some point. When it does, I double check that it is executing as designed and not due to an error. If not, then I evaluate market conditions to confirm they are still within those planned for use with this strategy and ensure there was not a logic / design flaw that I missed. If not, I generally keep trading but with reduced position size until results are back within historical norms. If it hits my pre-determined 'uncle point' (beyond its historical worst case DD) then the market has spoken and it's time to stop trading it and determine which assumption(s) failed.

if you don't plan your exit points before even making a trade this means your problem run depest than backtesting/strategy/. you need to plan exit points the same way you plan entry points and position sizing before even making a move. one trade is made you should stick to the plan (run minor changes due to macroeconomic events).

I would also suggest evaluating position size and determine if you are taking too much risk relative to your expectancy.

Unfortunately the maximum drawdown value as well as any other statistical performance metric cannot say anything about the health of the strategy. In order to understand whether it's time to switch it off or keep trading it's essential to do two things: 1) to understand what caused the maximum drawdown in your backtest and 2) to understand whether the present market regime is similar to that known worst historical period. If the reason which causes the drawdown today is different, or even worse — unknown, you should stop trading the strategy immediately. Most of the time it's possible to do even before the maximum drawdown is reached.

@Alex Krishtop ....I am glad to see your advice ,thank you very much !!!