MT5 Custom Backtest report generator

Check out this free MT5 tool that creates a custom backtest report and includes some of the key metrics MT5 leaves out on purpose.

Looks great, but when I run a backtest and then run the script "Backtest Performance Report", I get the following error in Experts Tab:

Error: Cannot open input file MQL5\Files\backtest.html
Error code: 5004

When I select File/Open Data Folder and navigate to MQL\Files, I can clearly see the file "backtest.html"

I have the following files in MQL\Files directory:

backtest.html
backtest.png
backtest-holding.png
backtest-hst.png
backtest-mfemae.png
testergraph.csv

I wonder if anyone else is having this problem.

Oh, I see what you've done. When I watch your video around the 2min:18sec mark, you have got the files in the following path:

MQL5\Files\MQL5\Files

So, I have now created another 2 subfolders shown above and it finally works. You somehow must have copied the path twice in your code.

And guess what? IT ACTUALLY WORKS! What a find that was!!! Probably saved you a lot of time there LOL :-)

It looks great. You little Ripper!!!!

Michael,

From your Backtest Performance Report, in the Monthly Performance Table, I notice you calculate the Monthly Percentage the following way:

Monthly Profit / Initial Starting Balance.

If the EA your testing increases the lot size as time goes on, then the results are going to be significantly skewed in favor of much higher meaningless percentages.

I think the Monthly/Yearly Percent should be calculated the following way:

Monthly Profit / Balance previous Month. And
Yearly Profit / Balance previous Year.

I think that's how MT5 calculates it for live trading history. What do you think?

To make sure the Monthly Percentages make sense for the duration of a backtest, I decided to try backtesting my EA consisting of a Portfolio of Rene Balke's famous strategies using a fixed balance size of $50K AUD. This means that no matter what size the Balance is during the backtest, all trade volumes are calculated based on a $50k balance size. So, the Portfolio consists of his Go Long (buy at start of day, sell at end of day), Turnaround Tuesday with MA (buy on Monday, sell on Tuesday at end of day) and some Range Breakout strategies (Gold and UJ) and here are the results (using much lower risk management than Rene does):

@Codemaster75 thanks so much I apologise for not responding sooner. Give me a couple of hours and I will address all your comments.
Once again thanks for the feedback. Knowing the tool will be used is all the encouragement I need to make it great. I have been working on some further functions so yes let me address the issues first.

Thanks for pointing out this error. I knew my configuration was going to cause trouble at some stage. I have made a correction to it using this approach, it should work fine now but do let me know if anyone has a similar issue.


To correctly find the right Files folder I have included some extra code to make sure it is found.
You don't need to worry about this code it should do the heavy lifting for you.

// Find the last occurrence of MQL5\Files in the path
string searchPattern = "MQL5\\Files";
int lastIndex = StringFind(terminalPath, searchPattern, 0);
int nextIndex; // Keep searching until we find the last occurrence
while((nextIndex = StringFind(terminalPath, searchPattern, lastIndex + 1)) != -1) {
lastIndex = nextIndex;
}
// If we found MQL5\Files in the path
if(lastIndex != -1) {
// Extract everything up to the last MQL5\Files
string basePath = StringSubstr(terminalPath, 0, lastIndex + StringLen(searchPattern));
return basePath + "\" + fileName;
}
// Fallback: Just use MQL5\Files
return "MQL5\\Files\" + fileName;
}

I have also fixed up the way the code is calculating monthly performance in the table.
It now checks the equity value at the beginning of the month and divides the net profit by this amount after adjusting for deposits or withdrawals.
For the yearly total, it is the cumulative return, which is not simply summing the 12 months. The cumulative return for the year takes compounding into effect.

I have included a few new charts into the report I would love some feedback.

Hi Michael,

If your monthly formula adjusts for deposits and withdrawals, does it still work if after a couple of years, you withdraw all your deposits so that you are only trading your profits? Is it the same formula that Metatrader use? I think forex factory do something similar in their trade reports.

I notice your monthly PNL chart contains 31 bars and never anymore. What do the x axis numbers represent? Maybe when you've finished adding new features and you get more feedback from other people, you can explain all these fascinating statistical features in another video.

Michael, what would you say is the most important statistic in these generated reports and why?

Michael, "Having run an incubation platform and program for 10 years", what kind of statistics would impress you if someone was able to generate your reports on a live account of more than 6 months?

I've been playing around with Python, and I haven't touched it in years, but I am very impressed with how easy it is to read in a comma seperated file (.csv) and stick it into a data frame. What I've done is first written an MQL5 script to extract the trading history from my live account and store the results in a comma seperated file with header "Close D/T,Symbol,Type,Volume,Gross P/L,Comm,Swap,Net P/L,Duration MSC,Magic,Comment". Next, I just write code in python "df = pd.read_csv ('Report.csv')" and with a bit of Perseverance, I can code any results I want.

Michael, what I don't know how to do yet is produce amazing interactive html reports like you have done above. In your opinion, what is the best way to go about learning how to produce these marvelous html reports with cascade style sheets in python? Is there some kind of python package I can use? or is it better to learn how to do cascading style sheets properly so that you can understand all the tags?

Not only do I want to be able to create reports like you do, I want to go that little bit extra and display results for each of the systems in my EA portfolio. In MT5 a unique system can be defined by Symbol, Magic Number/Expert Id and/or Comment. So, I could easily have another input text file containing these values read in from python code to extract even more amazing interactive html reports. Does this sound a bit far-fetched? For example, I could have a report that looks at all the large drawdowns in my systems (maybe use a graph) and see when they happened, then draw some conclusions from that. Or I could have a look at the correlation of daily/weekly returns from my systems and create a correlation table. I know less is more, but a picture does speak 1000 words :-)

Hi @codemaster75 I think it will be easier to respond like this.

If your monthly formula adjusts for deposits and withdrawals, does it still work if after a couple of years, you withdraw all your deposits so that you are only trading your profits?
At the start of each month it takes the starting equity for the calculation. How that equity is made up is not a concern. So short answer it shouldn't make a difference.

I notice your monthly PNL chart contains 31 bars and never anymore. What do the x axis numbers represent? Maybe when you've finished adding new features and you get more feedback from other people, you can explain all these fascinating statistical features in another video.

31 represent the days in a month. So day 1 is the first of the month and the 31st is the last possible day of a month. This is helpful to see if PNL is around month end/beginning.

Michael, "Having run an incubation platform and program for 10 years", what kind of statistics would impress you if someone was able to generate your reports on a live account of more than 6 months?

Watch this space. I will provide more commentary around the stats once I have settled on the final version.

I love Python I would prefer to stay in one language. However, there is also so much heavy lifting done by the MT platform that it offers the small guy a great chance to get up and running very fast. However, with speed comes accidents. So one needs to be careful.

I have been operating quant based strategies and run software fintech companies for years but seldom did any coding. I have now reached a point where my model has changed. No longer am I employing 20+ people and getting paid last, I am now doing everything myself. There are natural tradeoffs but for the moment I am loving the learning that comes with coding all these ideas.

My one tip that I find most helpful is my AI partner Claude. Lets keep the dialogue going in here so others can learn and I will share more and more.
On my website signal2noise dot news I share some python code.

Finally I have made some improvements to the backtest report generator. I have more to come.

I have just added a new Monte Carlo feature.

I have made a number of improvements to the Backtest Report Generator.

I have added a score and some alert warnings.

Hi mate is this still available on MQL5??? How can I find it??