I think it is very clear that in Forex trading we should expect all possible scenarios – especially the negative ones – to come true. All strategies have a probability – no matter how small – of having a draw down period that exceeds any threshold. Every system you trade has a given probability to have a 20, 30 or 99% draw down so it is always possible to reach this value at some turn of the market. In order to quit using the strategies when a draw down that is considered to be outside of the regular statistics of the system happens we use Monte Carlo simulations as a way to limit the possible “worst case” we are prepared to endure. However – as the above implies – there is a probability to reach this worst case scenario and in reality this value will be reached on a given percentage of your trading accounts before they get to “risk-free” territory (when the worst case scenario doesn’t imply a loss of the capital principal).
Trading a system thinking that one scenario will never materialize is nothing but naive. Whenever you trade an algorithmic strategy it is clear that there is a real possibility that your account will end up reaching what you deem to be your worst case statistical point although obviously you hope this will not happen before you reach the “risk-free” zone. Although in most cases where systems are properly designed this happens you must be aware that a given percentage of systems will always reach a Monte Carlo derived worst case scenario and you need to be prepared if the system you chose happens to be on that list. How do you prepare to and deal with a Monte Carlo worst case scenario? Here are some tips for you.
1. Don’t have only one account trading one portfolio or system. One of your best defenses against a Monte Carlo worst case scenario from both a psychological and financial point of view is to have several different accounts with evenly distributed capital trading different systems such that reaching a worst case scenario on one will not imply that you “failed” on your journey as an algorithmic trader. When you have one single account with a single portfolio and you reach a Monte Carlo worst case you will get a very hard blow because you will believe that there was something wrong with the way in which you did things. Most likely you were just unlucky, holding the portfolio that failed and went to its worst case. Several account and several portfolios diminish the effect of the rare but inevitable Monte Carlo worst case.
2. Top up the account that got its worst case and start with another algorithmic implementation. When a worst case is reached it is usually a bad idea to load a portfolio right back into that account even if its within adequate capital requirements. The reason is that while that account’s balance shows that loss there will be a strong psychological effect to attempt to “recover” which will cause you to load more risky implementations in the hope to get “over water” when you top the account to start a new portfolio it gives an air of freshness that makes the process of trading the entirely new portfolio a lot easier.
3. Use portfolios with varied levels of robustness. I believe that diversifying through robustness is a good practice. I usually have a given percentage of my accounts dedicated to portfolios which were developed with only a 10-20% out-of-sample test which have not been extensively live tested, another percentage goes to portfolios developed as the above but which have already proved themselves through real market conditions, another one goes to 10 year implementations developed without optimization and another one to 10 year developed, 10 year out-of-sample developments (total 20 year tests). Certainly there are many ways to measure robustness but it is good to diversify accounts so that you have some accounts that return less, draw down more frequently but are much less likely to reach Monte Carlo worst cases.
4. Next time, half risk. After you go through a worst case scenario you may be tempted to load a new portfolio that has even more risk than your previous implementation. My advice is that when you hit a worst case you should start a new portfolio with half the risk. The reason to do this is that you will most likely have a psychological hard time after any Monte Carlo worst case and starting a new implementation with half the risk of your previous one allows you to get profits/losses in a way which is less psychologically demanding. Trading with less risk is always easier to do and therefore after a worst case it is a good idea to half the risk of the replacing implementation.
5. Remember statistics avoid despair. Perhaps the most important thing when a worst case scenario is reached is to remember that the worst case exists for a reason. This value was derived from Monte carlo simulations and holds real statistical value so it should be treated as it was meant to be treated. When this value is reached do not continue trading, increase your risk or fall into any “gambling like” practices. Remember that when you decided to trade you knew that there was a worst case point at which you would stop trading that implementation and when it’s reached there is nothing to do but STOP, take a step back, reanalyze a new implementation and load it.
Certainly dealing with worst case losses is psychologically hard but it is a realistic aspect of trading which you will most likely have to face in the future as your journey in algorithmic trading catches speed. As I said before it would be naive to consider that none of the systems you develop will fail as the probability to fail is ALWAYS there. Therefore what you need to do is diversify, keep a focus on statistics and when a worst case happens move to a new implementation with less risk. Having different accounts with varied levels of robustness also helps as you will naturally have more confidence on a set of “core strategies” with higher levels of robustness, even if their return is lower and their tendency to go into draw downs is higher.
If you would like to learn more about my work in automated trading and how you too can learn to develop and analyze systems through Monte Carlo simulations please consider joining Asirikuy.com, a website filled with educational videos, trading systems, development and a sound, honest and transparent approach towards automated trading in general . I hope you enjoyed this article ! :o)