From theory to practice
The experiment started a month ago (testing of a portfolio of experts on the demo account) has brought the first fruits, namely: the results of the work of online experts and, most importantly, the first profit, albeit virtual.
There were some unfortunate mistakes on my part. When registering a demo account, the automatically generated password was not written down. Subsequently, when the account was transferred to another computer, access to the "accumulated treasures" became impossible. At that time the balance had grown by 700 dollars and there was open position on GBPJPYwhich gave a floating profit of about $200.
The only way out was to switch to a new account 2046102. To maintain the purity of the experiment, the initial balance of the new account was set equal to the funds (Equity) of the old account and immediately opened a position on GBPJPY. Of course, the correspondence "cent to cent" did not turn out, but the error was no more than 10 dollars. As a result, part of the profit received was not reflected in the reports of the old and new accounts, but was taken into account in the balance sheet.
But let's move on to analyzing the work of each Expert Advisor separately. To get a separate list of trades for each strategy, just use the script HistoryToFilewhich is covered in the same issue of the journal, section "Script Assistants".
To generate four independent reports, you will need to run the script four times with different settings. For example, generating a report on GBPUSD will require entering the string GBPUSD into the SymbolName parameter. Although there is another way. Since each strategy operates with its unique MagicNumber identifier, you can select the necessary trades by this value. The SARPlusAlligator strategy has MagicNumber 10002. Therefore, the SymbolName parameter can be left empty, and in the MagicNumber to enter 10002. The result will be the same in both cases.
It is easiest to leave the file name for report upload as it is, report.csv, and rename it after creating the file. The file appears in the experts\files folder of the terminal. You can open it directly in Excel and save it later as an xls file.
The resulting list of transactions per currency pairIf the results of testing strategies in the tester for the same period, you can compare them with the results of testing strategies in the tester. Such a comparison will show how much we can trust the results of testing and in which direction, better or worse, the deviation occurs in real life.
Parabolic SAR strategy and USDJPY currency pair
During four weeks of trading 11 deals were actually made, which brought a profit of 269.85 dollars (2395 pips). Five deals turned out to be unprofitable, and three of them followed one after another and in total formed a balance drawdown of 134.88. By and large, this value can be considered as the maximum drawdown on this segment of history.
The other six deals were profitable. The series of profitable deals also reached three profits in a row. This series brought a profit of 373.63 dollars and it should be noted that these are the last three deals. So there is every reason to continue the series.
Compared to the strategy test in the tester, there are no special differences. At least the number of trades completely coincides. It is clear that closing and opening prices do not coincide up to a point (see Fig. 1).
The biggest differences are deals #7 and #11, they are marked in red font color in USDJPY.xls file. The real state of affairs played a role here. It consists in the fact that at night I have no possibility to control uninterrupted work of my computer (every 24 hours I want to sleep eight hours for some reason), which sometimes leads to the loss of control over the account by Expert Advisors.
By a happy coincidence, the EA inoperability exactly during the period of making these deals gave additional profit, as both deals were opened later and at a better price.
Thus, according to the version of the strategy tester, the net profit should have amounted to 216.65 dollars (52.20 less than the real result), and the maximum drawdown 204.01. Since the drawdown in the tester is calculated taking into account the drop of funds per each trade, it makes no sense to compare it with the drawdown that was calculated on the basis of online data on a series of losing trades.
SARPlusAlligator strategy and GBPUSD currency pair
The real number of deals was 10. The cumulative profit was at the level of 175.52 dollars (1768 points). There were quite few losing deals - four, and the maximum series of losses at the moment can boast only number 2. Nevertheless, the maximum continuous loss was reached by one deal - 83 dollars.
Profitable series of deals did not go far - three profits in a row. They gave a continuous profit of 123.24 dollars.
The results obtained in the strategy tester differ again for the worse (see Figure 2)
First, we can see a discrepancy in the number of deals, in the tester there are two more. The matter here is again in night deals. The deal of 26.11.2009 05:00 buy was not actually made. The second deal does not appear in the online report yet, as it is an open position, which makes no sense to take into account, but the tester has already taken it into account.
Secondly, night openings led to the difference of time and prices of opening or closing deals, which also influenced the final result. However, this influence is mainly in our favor.
According to the tester's version, the trade should have turned out that way: net profit 69.01 (106.51 less than the actual figure) dollars versus a maximum drawdown of $223.70 dollars.
Stochastic_V2Filtr strategy and GBPJPY currency pair
The biggest inconvenience when switching from one account to another was caused by an open position on the GBPJPY currency pair. I had to restore the current position immediately upon account opening. That is why in the Excel-file of the report the transaction of 04.12.2009 buy is synthetic, as it is taken from the reports on two accounts.
As it turns out, this particular one strategy at the moment made the most tangible contribution to the deposit growth. By means of 16 deals it was earned 889.15 dollars (7800 points). The most part of deals (ten) gave a loss.
At all this a series of losing deals looks very impressive - 5 in a row. But the total damage they caused is relatively small - 166.75. And the biggest loss was brought by one deal - 222.56 dollars.
The series of profitable deals was shorter than the losing one - only two deals. The maximum continuous profit was again made by one deal - 490.18 dollars.
The strategy tester gave these results (see Figure 3)
We have to state another discrepancy in the number of deals. However, again it is necessary to reject the last deal, as it is not closed yet and in the end it can bring both profit and loss.
The first discrepancy refers to the "night shift" - the deal of 17.11.2009 00:00 buy. The second discrepancy chronologically follows the first one - the deal of 17.11.2009 04:00 sell.
The other differences between real trading and the tester are the opening and closing prices of trades.
The tester's results were again worse than the real thing: a net profit of 692.31 (196.84 worse than the real thing) dollars and the maximum drawdown $590.02.
ZigZagStatistic strategy and USDCHF currency pair
The strategy's contribution to the overall success amounted to the second largest value - $297.49 (3016 points). Eight trades were spent to achieve the result, of which only two were unprofitable.
It is incorrect to speak in this case about a series of losing trades, as the strategy is based on compensation of failure of the first open trade by an additional trade. That is, by the current moment it can be considered that all entry points in deals justified themselves and did not lead to achievement stop order level.
The strategy tester, in contrast to the previous three cases, gave a more optimistic picture (see Figure 4):
In this case it is again about night deals: 18.11.2009 00:00 sell and 19.11.2009 00:00 buy. Otherwise, there are no significant differences between the tester and online.
Without missing two deals and according to the version of the strategy tester, the Expert Advisor's work should have led to the following results: net profit 374.02 (76.53 better than the real indicator) against the maximum drawdown of 129.22 dollars.
Conclusion
If we add up the gains that all four strategies have produced, we get a total gain of $1,632.01, while the current balance of the demo accounts is 18,621.64. As you can see, a little more than 10 dollars were lost due to the switch to another account, which is not a significant amount.
As a percentage, the profit is quite decent - 9.6%. This indicator can be considered more than successful, but we should not be happy about it yet. There are several reasons for this:
1- The historical interval is still too short for serious conclusions and the result may turn out to be mere luck.
2. None of the strategies fell into the zone of the highest maximum drawdown, from which we can conclude that the times of being in a serious drawdown are still ahead of us.
3. Since the target annual profit is estimated to be 50%, one or more of the following months will likely produce a much smaller profit or no loss at all.
4. The success of strategies can be conditioned by the end of the calendar year, as the market follows a well-established track. This rut most often changes its direction with the beginning of a new year.
Attached files:
Test.zip - detailed results of strategy testing over the past four weeks.
USDJPY.xls - list of real deals made as a result of the work of the Parabolic_V3 Expert Advisor during the last four weeks.
GBPUSD.xls - list of real deals made as a result of the SARPlusAlligator Expert Advisor's work during the last four weeks.
GBPJPY.xls - list of real deals made as a result of the work of the Stochastic_V2Filtr Expert Advisor during the last four weeks.
USDCHF.xls - list of real deals made as a result of the work of the ZigZagStatistic Expert Advisor during the last four weeks.