THESE AIN'T YOUR GRANDDADDY'S TESTING TECHNIQUES - COMPARE LIVE AND BACK TESTING RESULTS OF ABBEY 24 EAS PORTFOLIO PART2

In part 1, we compared 20 trades of the EURUSD made from the Abbey 24 EAs portfolio from October 8, 2015 to November 3, 2015. In this blog, we will be comparing the next 20 trades from November 2, 2015 to December 1, 2015. As you can see some of the trades overlap each other from the different opening and closing times. In this post, we will be comparing the EURUSD live and the back testing results of the Abbey 24 EAs Portfolio. You can follow and perform the same tests by downloading the EA from the link below: 

https://www.mql5.com/en/market/product/12965

For these new twenty trades, we notice that 19 out of 20 trade entries were perfect. The time difference between the backtesting and real-time trades was within 8 seconds. There was one outlier, and it entered  the market 49 minutes later. This is a good example of real market trading whereby, the slippage limit prevented us from entering the market at our desired price. There could be internet connections, broker requotes, and etc. This can be examined further by diving into the log files. I think that there is an opportunity for the App developers to provide a one-click automated reporting to compare the live results with the backtesting results with some error identification on what may have caused the 49-minute delay.   

For the exit timing of these 20 new trades, we notice that 17 out of 20 trades were perfect. There were three exits that are questionable, and I have highlighted them in yellow. The first yellow exit is off by 1 hour between live trading vs. backtesting and is considered acceptable. In this case, the real time trading result is less than the backtesting result by 13 pips. This is not a significant amount to a swing system; however, it is a considerable profit reduction if it is a scalping system. For the second highlighted yellow exit, this may be another outlier since the trade exited five days ahead of the backtesting results. We really don't know what happened here. We notice that the profit is 45 pips less compared to the simulated backtesting results. For the third highlighted yellow exit, our trade exited with takeprofit 12 hours earlier whereas the backtesting result did not take profit and exited the trade with 24 pips less. 

When we compared the backtesting with the live trading results for these 20 trades, the total pip is 1,309 vs. 1,291 (18 pips different).  In this case, we can set a rule of thumb that our backtesting results will always be better than the real time trading results.

Ideally, we should compare 100 trades to obtain a large enough sample. In the next post, we will compare the next 20 trades. It is better for us to review 20 trades at a time so that we don't get the information overload. You can follow and perform the same tests by downloading the EA from the link below: 

https://www.mql5.com/en/market/product/12965