99% backtesting. Where and how?

Hi traders

Yes, I am a bit confused about backtesting. I always hear that you have to had 99% backtesting quality in order for you to get somehow accurate results for your strategy or EA that you like to backtest. And I also hear much about "quality data" and "not quality data", and also I find some paid services for that.
So, now can anyone explain for me these things out of what you know?
Because as much as I know about backtesting and out of what I am doing is, to go to the download center of my mt4 platform and download the available data provided for the instrument I want and then begin the game of playing around with it. And of course during the backtest I set it to every tick option.
My broker is XM-Global-limited, and as I can see they provide real data. So if anyone that know something different please share your knowledge.
So, please clerify for me if I do anything wrong here, because I had spend months backtesting tons of ideas and strategies and all showed that they have to go to the recycle bin.

Another issue I have faced with the gold data. Whenever I try to download gold history data it tells me that there are no data available. I have tryed different brokers mt4 platform with same issue. So whats going on here? Is it not possible to backtest gold or what?

And lastly merry christmas for you all that celebrate christmas. And looking forward to hear from you.

I'm using mt5, so may be my answer will not be accurate...
I have noticed that if I have less than 100% tick data, the results of testings are different.
How can I get tick data? First I have bought it from tickstory. But you need it only if you want to test 10 years back for example.
Second way I just go to View-Symbols and download tich data from my broker (icmarkets). It gave me 100% tick data for most currencies for 2-3 years without any problem.

Just tried to download tick data for gold. It is ok.


About recycle bin for strategies: this is ok, because every strategy you will find have only math calculated entry rules, and "magic/luck" exit (like static SL, RR etc). That's why they fail.

p.s. I don't know may be you know it already but do not backtest using close values. Only real ticks. Because if you will use close you need to have EA that close after candle close. But if you have for example dynamic TP/SL the difference will be huge.

MT4 is not able to perform tickdata backtesting by default. It can only utilize M1 bars at best. So it interpolates M1 data for timeframes smaller than that. That means price changes below the M1 timeframe are artificial here.
Considering this a real tickdata backtest is not possible using MT4 without additional software/enhancements. You can roll a dice here as well.

There are paid solutions to extend the MT4 terminal to handle real tickdata including variable spread and even slippage simulation based on several statistical methods.

Tickstory is indeed one. I am using Tick Data Suite instead. You get what you pay for. In the end it is way more expensive to lose money due to inaccurate backtesting.

Hello,

Does anyone have access to DXY tick data?

Thanks

I've spent the last 3-4 weeks backtesting my strategy using TradingView replay mode with FXCM data. Is this data not accurate then?
I used it to backtest my strategy from 2018, 2019 and 2020.

Thanks for all your inputs.
So far I am still confused. Can I just trust my xm-broker's history data as long as it is for the last three years or what? How can I know these data are not quality data and these are?
I mean how to find out?
And why the data from 10 ago is not accurate when they are downloaded from the mt4 historycenter?

Sorry, hope some one that can explain these things for me.

If you doubt the veracity of the data or its relevance, you should of course check everything, especially if it in any way negatively affects your work. And for that purpose, I would recommend that you use some independent resource so that you can compare the graphs and make sure that the data really does match reality. And this is something that you really have to do, even if you initially chose a brokerage company with an excellent reputation, because everything can change and everything can affect how you analyze the market and how you navigate through it. So you shouldn't be too trusting, it's a dangerous business.

This is of course a common problem, which deserves a separate consideration, because many traders face some difficulties precisely because they receive inaccurate or completely dubious data that does not make sense at all, which leads to serious losses and, consequently, to complete disappointment in this business. And this is a big, I would even say huge problem.
Sometimes this is done intentionally, sometimes it is due to the fact that the brokerage company invests little money in the development of the trading platform or analysis tools, and it all works very poorly.
I agree that this needs to be checked in advance and it really is of key importance to any trader.
I hope that you are doing just fine.

its a nice description with information , really got so many fine lines, thanks again for the nice post you have done.

Hi Mate,

For your need, please have a look at
https://forextester.com

I have used it before.

Regarding the History data, if you have licenced version its easy to fetch the data. Or if you have data in csv format, you can import it. ( see the attached image)

Few things I personally liked are

- fundamental flag, for one my strategies, I wanted to check with respect the FOMC's economic projection. It helped me.

- From strategies perspective , you can import from Mt4/mt5 if you have your own EA. You can use their own EA builder to built and test simple strategies. I can feel from your explanation that you might have some serious complex strategies to test.

Whenever I try to download gold history data it tells me that there are no data available. I have tryed different brokers mt4 platform with same issue. mcdvoice myhoneybakedfeedback

can you back test Ea too with trading view?

Hey guys,
I sometimes see posts where people say they've done backtests on ICMarkets for several years back. But when I try to test using "every tick based on real ticks," I only get full tick data for the current year (2025.01.01-2025.30.07). As soon as I go back to, say, 2024, the backtesting quality already drops to around 33%.
I even reached out to ICMarkets support and they told me they don't have real tick data that far back.
So how are others doing proper multi-year tick-by-tick backtests? Am I missing something, or is it just not possible?
Thanks in advance if anyone can clarify!