Hi guys I am seeking your opinions on how to plan for worst scenario in regards to losses. Say I backtested my system and the worst drawdown during that period was -$1000 (per quarter). Assuming my testing to be correct and results valid, what kind of drawdown should I plan for when actually live trading? p.s. I dont consider events such as the flash crash as valid seeing how they are too infrequent. I am looking for reasonable protection not protection against every possible thing. Thanks in advance.
Did a search for âworst case scenarioâ, and then quickly scanned thread titles to look for anything I remember as being possibly relevant to your question. But my memory isnât that good, so apologies if some/all of these are wide of the mark ⦠http://www.elitetrader.com/vb/showthread.php?s=&threadid=218614&highlight=worst+case+scenario http://www.elitetrader.com/vb/showthread.php?s=&threadid=217800&highlight=worst+case+scenario http://www.elitetrader.com/vb/showthread.php?s=&threadid=212824&highlight=worst+case+scenario http://www.elitetrader.com/vb/showthread.php?s=&threadid=210848&highlight=worst+case+scenario
Thanks. I am looking for personal experiences though. For example, your in experience how much drawdown would you plan for my given scenario?
what i am asking is what kind of discrepancy should i expect in terms of drawdowns between real world results and theoretical backtesting
In 08... you could have easily lost 50%, or more. There is no correlation, zero, zip, between "real world" and theoretical backtesting. Markets have no memory and cannot be predicted. Only false reassurances.
I think this will be highly dependent on your trading style and how important trade execution is to your strategy. Are you a scalper/short-term trader or trading on a higher time frame? In engineering, there is a term called a "design factor" which is a fudge factor that allows you to account for uncertainty. Typical values are 1.5 to 4, depending on how critical the "thing" is. In this case, your analysis has shown -1000 is your estimate, so if you use a factor of 2-4, this should at least keep you in the realm of reason.