html menu by Css3Menu.com
The Search for the Holy Grail Trading System : Part 2
The following case study is for educational purposes only and is no endorsement of any of the concepts described in this article. The results provided in this case study are not indicative of, and have no bearing on, any results, which may be attained in actual trading. Results of past performance are no guarantee of future performance. It should not be assumed that you would experience results comparable to that reflected by the results described in this article. No assurance is given that you will not incur substantial losses, nor shall Compuvision Australia Pty Ltd be held liable if losses are incurred.
In part 1 of this article we discussed a possible trading system that for all intense and purposes fitted the practical definition of a Holy Grail trading system. We then evaluated its performance over the worst possible conditions by simulating it over a 15-year interval using a broad range of securities from the ASX top 200. In the trading simulation we also used the worst case slippage and still found the system to be profitable in the long run with reasonable drawdowns that were contained within a maximum of 14%. In this article we focus on whether the outcome of this simulation was the result of a fluke selection of stocks or whether or not another trader trading the same system with the same trade database would produce similar or completely different results.
When trading a system using a portfolio of securities the outcome is no longer unique as it is when testing with one security alone. This is due to the number of different permutations and combinations of securities which can be traded at any given time. What this means is that Trader-A could pick a different set of securities than Trader-B and come up with two different sets of results even though they are trading the same trading system over the same period of time.
To understand this aspect more closely consider a trading a system, which generates entry triggers for six securities on the next trading day. Due to available trading capital limits, only two of these six possible securities can actually be traded. Which two of course depends on the trading algorithm in the simulator and the way it has been set up. Once two of these trades are entered the others can be thrown away since the entry triggers are no longer valid after that day. This also explains why all of the trading candidates in the trade database cannot possibly be traded in each simulation. However the simulator allows a random selection of securities from a group of securities, which have the same entry date thus modeling the variability of a trading system when trading with a portfolio of securities. This is a very important aspect which is usually overlooked when back testing a trading system since most back testers are usually only limited to the analysis of one security alone.
Monte Carlo Analysis
This non-unique outcome gives rise to the need for proper statistical analysis in order to thoroughly evaluate a trading system. By repeatedly simulating a trading system using a random selection of securities that meet the entry date and position sizing requirements we can build up a statistical profile of a trading system based on the portfolio used. Using the Enterprise Edition of TradeSim a Monte Carlo analysis was performed on the trading system in order to generate a comprehensive statistical profile of the trading system. TradeSim was setup to generate 20000 simulations, which was more than enough to thoroughly exercise this trading system as well as producing a more accurate statistical analysis. The results from this analysis are shown in the screen shots below. You will observe in the Monte Carlo report shown below that out of the 20000 simulations at no stage was there one simulation, which produced a loss - a most remarkable result!
In the Profit Distribution histogram below you can see how the net profit was distributed over the 20000 simulations ranging from 71% all of the way up to 7725% with an average profit of 1435% - a most exceptional outcome. One should note that the probability of achieving the highest profits is 5 in 20000 or (1 in 4000), in other words a fluke selection of stocks would have only achieved this outcome. The most important thing to note is that at no time was a loss produced indicating that this is a historically profitable trading system.
The winning and losing trades histograms shown below illustrate a higher number of winning trades compared to the losing trades even though we have used a random entry strategy which indicates the robustness of the exit strategy used.
We can summarize the above results as follows.
A final point to note is that all of the results apply to the past performance and there is no guarantee of future performance but the important point to note is that trading with a system such as this would ensure that the balance of probability would be in our favor and we could be fairly confident of producing a profitable outcome in the future providing that we stick judiciously to the trading rules. This is the biggest challenge! These results highlight the need to thoroughly test a trading system before any conclusions about its performance are made.