Are backtests on renko charts reliable? | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

RENKO Trading SYSTEM - RENKO Indicators 7 replies

New Renko Bar and Reversal Renko Bar Indicator Alert Help Please 5 replies

MTF indicators don't work during backtests? 6 replies

Question regarding automatic backtests 0 replies

Validity of Backtests in MT3? 1 reply

Attachments: Are backtests on renko charts reliable?
Exit Attachments

Are backtests on renko charts reliable?

  • #1
  • First Post: Jun 25, 2020 2:50am Jun 25, 2020 2:50am
Since I met renko charts I have been impressed and fascinated by studying them I have managed to find some profitable strategies. Recently I am approaching the world of expert advisors and automatic trading, therefore I immediately wanted to try automatic strategies based on renko charts. Reading various threads on the subject here I came across this guide (https://www.forexfactory.com/showthread.php?t=572375) where I learned how to perform backtests with renko. The thing that left me amazed was the exceptional results of the backtests. So good as to make me suspicious of the validity of these tests. Here is an example to talk with some data.

This is a very simple expert who opens buy positions when two bullish renko candles are formed and closes it when a bearish candle is formed, vice versa for the sell.

Attached Image (click to enlarge)
Click to Enlarge Name: asdad.JPG Size: 136 KB


Everything is too perfect to be true. I'm pretty sure that because I'm a noob in backtesting ea these results could be distorted by some mistakes of mine. I used dukaskopy data that I downloaded from quantum data manager, I followed this guide for renko backtesting (https://www.forexfactory.com/showthread.php?t=572375) and these are the backtest settings. I'll link the ea I've used here below.

Attached Image (click to enlarge)
Click to Enlarge Name: 6465.JPG Size: 84 KB



Or maybe the problem lies in the backtests of the renko charts themselves?
Please let me know what do you think


Attached File(s)
File Type: ex4 Renko EA.ex4   166 KB | 490 downloads
  • #2
  • Jun 25, 2020 3:48am Jun 25, 2020 3:48am
the quality of the data you are backtesting is very low. This you understand from the quality of the model. It should be at least 90%.
PS. Sono Italiano
  • #3
  • Jun 25, 2020 3:59am Jun 25, 2020 3:59am
Quoting gian97
the quality of the data you are backtesting is very low. This you understand from the quality of the model. It should be at least 90%. PS. Sono Italiano
Quality on 1m time frame (even if he is using renko) can't be higher than 25% on metatrader 4. This is how the platform works.

Just try to make a manual backtesting, scrolling the chart back manually and test if your stategy works fine.
  • #4
  • Jun 25, 2020 5:10am Jun 25, 2020 5:10am
Quoting jojofx
{quote} Quality on 1m time frame (even if he is using renko) can't be higher than 25% on metatrader 4. This is how the platform works. Just try to make a manual backtesting, scrolling the chart back manually and test if your stategy works fine.
Well, I didn't know this. Thanks so much!
  • #5
  • Jun 25, 2020 5:39am Jun 25, 2020 5:39am
Quoting jojofx
{quote} Quality on 1m time frame (even if he is using renko) can't be higher than 25% on metatrader 4. This is how the platform works. Just try to make a manual backtesting, scrolling the chart back manually and test if your stategy works fine.
Wrong!

This is only valid for the vanilla MT4 backtester. Unfortunately it is unreliable independent of the used timeframe on top of that! You can ignore the percentage value as it is always rubbish. Reason is that MT4 vanilla backtester interpolates data smaller than 1m. You are simulating whatever but no true movement of quotes.

The picture gets completely different when extending the Backtester with additional software.
  • #6
  • Jun 25, 2020 6:33am Jun 25, 2020 6:33am
I spent a lot of time going down this road, unfortunately renko backtests on mt4 doing it the way you've done it even with the best data are completely pointless. A fairly easy way to prove this point, run your EA on live charts for 1 week, at the end of the week run a backtest on that week and you will see the results are completely different to the live results.
Pursuit of happiness
  • #7
  • Jun 25, 2020 7:19am Jun 25, 2020 7:19am
Quoting Drolph
{quote} Wrong! This is only valid for the vanilla MT4 backtester. Unfortunately it is unreliable independent of the used timeframe on top of that! You can ignore the percentage value as it is always rubbish. Reason is that MT4 vanilla backtester interpolates data smaller than 1m. You are simulating whatever but no true movement of quotes. The picture gets completely different when extending the Backtester with additional software.
You type "wrong" but with your statement you just confirming my point
  • #8
  • Jun 26, 2020 2:44am Jun 26, 2020 2:44am
Quoting gian97
the quality of the data you are backtesting is very low. This you understand from the quality of the model. It should be at least 90%. PS. Sono Italiano
Ciao , I'm using dukaskopy data that I downloaded from quantum data manager, I read that they are good data, why do you think this is not so? In any case, what data do you use and where do you suggest me to find good data?
  • #9
  • Jun 26, 2020 2:52am Jun 26, 2020 2:52am
Quoting jojofx
{quote} Quality on 1m time frame (even if he is using renko) can't be higher than 25% on metatrader 4. This is how the platform works. Just try to make a manual backtesting, scrolling the chart back manually and test if your stategy works fine.
I don't understand, so the problem is about the strategy tester in MT4 and not the data? Because even with a manual backtest as you suggest the data remain the same. Always for the records, I used dukaskopy data from quantum data manager.
  • #10
  • Jun 26, 2020 2:56am Jun 26, 2020 2:56am
Quoting Drolph
{quote} Wrong! This is only valid for the vanilla MT4 backtester. Unfortunately it is unreliable independent of the used timeframe on top of that! You can ignore the percentage value as it is always rubbish. Reason is that MT4 vanilla backtester interpolates data smaller than 1m. You are simulating whatever but no true movement of quotes. The picture gets completely different when extending the Backtester with additional software.
Hi, where can I learn how to do a good backtest you can trust, what programs do you use?
  • #11
  • Jun 26, 2020 3:27am Jun 26, 2020 3:27am
Quoting EScalperi
{quote} I don't understand, so the problem is about the strategy tester in MT4 and not the data? Because even with a manual backtest as you suggest the data remain the same. Always for the records, I used dukaskopy data from quantum data manager.
Yes the data remain the same but you are sure, during the backtest, that the EA open trades when it is supposed to.
Or you can do as Poh suggested: let the EA run for one week, then do the backtest of that week and check if the trades match.
  • #12
  • Jun 26, 2020 3:31am Jun 26, 2020 3:31am
No back testing is reliable. you need to live test.

KP
Do your homework, follow the footprints of smart money
  • #13
  • Jun 26, 2020 3:39am Jun 26, 2020 3:39am
Quoting EScalperi
{quote} Hi, where can I learn how to do a good backtest you can trust, what programs do you use?
There are several solutions out there. I am using TickDataSuite.

My personal advise: If you want a serious solution, use a paid one. Nothing compared to the amount of money lost because of unreliable backtesting.
  • #14
  • Jun 26, 2020 10:11am Jun 26, 2020 10:11am
Quoting Poh
I spent a lot of time going down this road, unfortunately renko backtests on mt4 doing it the way you've done it even with the best data are completely pointless. A fairly easy way to prove this point, run your EA on live charts for 1 week, at the end of the week run a backtest on that week and you will see the results are completely different to the live results.
Agreed. It can be exhausting, but this is the way to be sure the data quality is good or not.
  • #15
  • Jun 26, 2020 10:21am Jun 26, 2020 10:21am
Quoting Poh
I spent a lot of time going down this road, unfortunately renko backtests on mt4 doing it the way you've done it even with the best data are completely pointless. A fairly easy way to prove this point, run your EA on live charts for 1 week, at the end of the week run a backtest on that week and you will see the results are completely different to the live results.
Yep you're right, if you backtest it a week back the result will be different it's because the anchor of the renko chart are changed since you only choose a week back.
Here's the things, renko chart created when a price traveled X pips from point A. The problem is where is this "Point A" will be?
Each time you load a renko chart on different mt4 your beick will slightly different since the anchor point ("Point A") is different.
One way to increase the reliability of renko backtest is to make a Fxt file from third party tick supplier such as Dukascopy. This way, your anchor point always be at a fix price.
Or the other way is to manually backtest using online chart such as tradingview since they have all the tick data on their side therefore the anchor point are at a fix price.

Just my experience so far exploring renko.
  • #16
  • Jun 27, 2020 4:15am Jun 27, 2020 4:15am
Quoting Evernewtron
{quote} Yep you're right, if you backtest it a week back the result will be different it's because the anchor of the renko chart are changed since you only choose a week back. Here's the things, renko chart created when a price traveled X pips from point A. The problem is where is this "Point A" will be? Each time you load a renko chart on different mt4 your beick will slightly different since the anchor point ("Point A") is different. One way to increase the reliability of renko backtest is to make a Fxt file from third party tick supplier such...
This is why I'm using Dukascopy data. And as you can see in every backtest the data confronting the same time period look exactly the same.

Attached Image (click to enlarge)
Click to Enlarge Name: image.jpg Size: 415 KB


I'm not sure everybody understood I'm not using Mql5 default data, but I'm using dukaskopy data. Because I'm reading alot of comments about the bad quality of my data.

Anyway I'm going to test others data sources, to be sure.
  • #17
  • Jun 27, 2020 4:22am Jun 27, 2020 4:22am
Quoting Drolph
{quote} There are several solutions out there. I am using TickDataSuite. My personal advise: If you want a serious solution, use a paid one. Nothing compared to the amount of money lost because of unreliable backtesting.
Ok', I'm gonna download the TickDataSuite trial version and confront the data with mine. Anyway, I'm using quantum data manager, as they declare they use dukascopy data, which I thought was good data.
  • #18
  • Jun 27, 2020 4:39am Jun 27, 2020 4:39am
Quoting jojofx
{quote} Yes the data remain the same but you are sure, during the backtest, that the EA open trades when it is supposed to. Or you can do as Poh suggested: let the EA run for one week, then do the backtest of that week and check if the trades match.
I checked with the visual mode and the tester run the ea perfectly fine.
  • #19
  • Last Post: Mar 22, 2022 5:29am Mar 22, 2022 5:29am
Quoting Drolph
{quote} Wrong! This is only valid for the vanilla MT4 backtester. Unfortunately it is unreliable independent of the used timeframe on top of that! You can ignore the percentage value as it is always rubbish. Reason is that MT4 vanilla backtester interpolates data smaller than 1m. You are simulating whatever but no true movement of quotes. The picture gets completely different when extending the Backtester with additional software.

what you mean by additional software?
0 traders viewing now