Forex Tester 3, better then MT4 back testing? | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Is Trading Forex And Making Money Better Then Sex? 27 replies

MT4: Create objects in Tester in new build MT4 0 replies

Better Then MT4 - Any Suggestions? 0 replies

Testing a semi-automatic EA with a strategy tester 1 reply

2009.03.11 15:41:15Tester memory handler: tester stopped because not enough memory 19 replies

Attachments: Forex Tester 3, better then MT4 back testing?
Exit Attachments

Forex Tester 3, better then MT4 back testing?

  • #1
  • First Post: Dec 29, 2018 4:48am Dec 29, 2018 4:48am
Hi all,

How do you think the Forex Tester 3, simulation software? Is it better then back testing by MT4?

I read the information from their website, seems like good tool to analyse trading strategies, test EA...

Any comment will be appreciated.

Happy trading and best wishes for 2019!
  • #2
  • Dec 29, 2018 5:47am Dec 29, 2018 5:47am
Quoting cocozone9898
Hi all, How do you think the Forex Tester 3, simulation software? Is it better then back testing by MT4? I read the information from their website, seems like good tool to analyse trading strategies, test EA... Any comment will be appreciated. Happy trading and best wishes for 2019!
Hi,

Well, it's difficult to say without trying it, but beware of one thing that very often goes unnoticed by a lot of traders, if that tool is not capable of using historical data from the broker you will be trading with, then don't use it because you will be disappointed, I can tell you this is of upmost importance !
Simplicity is the Ultimate Sophistication.
  • #3
  • Dec 29, 2018 5:56am Dec 29, 2018 5:56am
Quoting broketrader
{quote} Hi, Well, it's difficult to say without trying it, but beware of one thing that very often goes unnoticed by a lot of traders, if that tool is not capable of using historical data from the broker you will be trading with, then don't use it because you will be disappointed, I can tell you this is of upmost importance !
So without the historical from the brokers, the simulation software will be junk?

Thank you for the comment.
  • #4
  • Dec 29, 2018 6:07am Dec 29, 2018 6:07am
Quoting cocozone9898
{quote} So without the historical from the brokers, the simulation software will be junk? Thank you for the comment.
That's my opinion, yes !

I have been testing EAs for years in the past, getting freely available and often good quality data 10 years back and then importing it into my MT4 platform.
Then, spending time to back test, correct and fine tune just to discover later that my EA wasn't working on my real account and me wondering why ! Until the day when by luck I discovered that my EA was behaving differently with my broker data and with the historical data from an external source !

Since that day, I develop and test my EAs exclusively with my broker's data and now have consistent results. I have an EA trading in a live account and my backtest trades are almost to the pip like the real ones.
Simplicity is the Ultimate Sophistication.
  • #5
  • Dec 29, 2018 6:16am Dec 29, 2018 6:16am
Quoting broketrader
{quote} That's my opinion, yes ! I have been testing EAs for years in the past, getting freely available and often good quality data 10 years back and then importing it into my MT4 platform. Then, spending time to back test, correct and fine tune just to discover later that my EA wasn't working on my real account and me wondering why ! Until the day when by luck I discovered that my EA was behaving differently with my broker data and with the historical data from an external source ! Since that day, I develop and test my EAs exclusively with my...
Hm..sounds promising..!

I believe as a newbie like me, got a long way to run...

Cheers!
  • #6
  • Dec 29, 2018 6:20am Dec 29, 2018 6:20am
Quoting cocozone9898
{quote} Hm..sounds promising..! I believe as a newbie like me, got a long way to run... Cheers!
Don't worry, it's a long journey for everyone !

Good luck !
Simplicity is the Ultimate Sophistication.
  • #7
  • Dec 29, 2018 6:29am Dec 29, 2018 6:29am
A final word to illustrate, look at the picture below:

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 39 KB


At the window on the top are current open trades on my live account, you can see the window is the Terminal and Trade tab.
At the window on the bottom are the trades given by the tester, you can see it's the Tester window and Result tab.

The trades in the terminal correspond to the trades 17 to 21 and if you look at their entry price, they correspond except for the first and last one (2 and 1 pip difference)

If you can have that kind of precision with your tool and your broker real trades, then you can use the tool but I doubt about it.
Simplicity is the Ultimate Sophistication.
  • #8
  • Dec 29, 2018 6:51am Dec 29, 2018 6:51am
Quoting broketrader
A final word to illustrate, look at the picture below: {image} At the window on the top are current open trades on my live account, you can see the window is the Terminal and Trade tab. At the window on the bottom are the trades given by the tester, you can see it's the Tester window and Result tab. The trades in the terminal correspond to the trades 17 to 21 and if you look at their entry price, they correspond except for the first and last one (2 and 1 pip difference) If you can have that kind of precision with your tool and your broker real trades,...
Yes, I am sure there will be no tool for such precision. However, just trying to find a way to try and test some EAs, and then try to modify or said adapt to use it.

Certainly, there is no perfect EA to trade, but a good method should be realistic, right?
  • #9
  • Dec 29, 2018 7:02am Dec 29, 2018 7:02am
Quoting cocozone9898
{quote} Yes, I am sure there will be no tool for such precision. However, just trying to find a way to try and test some EAs, and then try to modify or said adapt to use it. Certainly, there is no perfect EA to trade, but a good method should be realistic, right?
Yes, and that certainly starts with using real data from your broker.
Simplicity is the Ultimate Sophistication.
  • #10
  • Dec 29, 2018 8:31am Dec 29, 2018 8:31am
Quoting broketrader
if that tool is not capable of using historical data from the broker you will be trading with, then don't use it because you will be disappointed, I can tell you this is of upmost importance !
I couldn't agree more - obtaining and testing one's own broker data is absolutely critical.

@cocozone9898

I purchased Forex Tester 2 a few years ago, but try as I did, couldn't get on with it. Unless they've changed their system, I found that importing historic data into the program was a non-intuitive tedious process. The standard data accessed via the program appeared to only resolve to 2 x decimal places, and (if memory serves me correct), if one needed more accurate data, it needed to be purchased at extra cost.

Indicators were an issue too. I was working with a group of traders at the time testing supdem indie, an MT4 indie based on the ZigZag indicator. The native version of this indicator would not run in Forex Tester. We did make some changes and ended up with a version that worked, however, it ran so slowly and froze computers that it wasn't possible to obtain any meaningful test results.

There's very little help on offer inside Forex Tester, and if one runs into issues the program will divert the user to an Internet based forum. I found these forums so expansive and slow to get any useful help in a reasonable amount of time that I gave up on the program altogether. The drawbacks with testing on MT4 seemed minimal by comparison, and on that basis, I went back to testing on MT4.

Forex Tester may be useful in terms of playing/rewinding price in order to observe how the market ebbs and flows, but to be honest, this can still be achieved using one's own broker data (albeit without the rewind function) on MT4. Knowing what I know now, if I were faced with buying Forex Tester again, I honestly would not spend the money because I personally do not feel the product is worth it.
  • #11
  • Dec 29, 2018 8:56am Dec 29, 2018 8:56am
Quoting simond2002
{quote} I couldn't agree more - obtaining and testing one's own broker data is absolutely critical. @cocozone9898 I purchased Forex Tester 2 a few years ago, but try as I did, couldn't get on with it. Unless they've changed their system, I found that importing historic data into the program was a non-intuitive tedious process. The standard data accessed via the program appeared to only resolve to 2 x decimal places, and (if memory serves me correct), if one needed more accurate data, it needed to be purchased at extra cost. Indicators were an issue...
Thank you so much for the information.

I do like to learn from you pro guys, with very useful information and advises. Certainly, I hope and will share some in the near future!

Cheers!
  • #12
  • Dec 29, 2018 9:19am Dec 29, 2018 9:19am
Quoting broketrader
{quote} Hi, Well, it's difficult to say without trying it, but beware of one thing that very often goes unnoticed by a lot of traders, if that tool is not capable of using historical data from the broker you will be trading with, then don't use it because you will be disappointed, I can tell you this is of upmost importance !
  • #13
  • Dec 29, 2018 4:17pm Dec 29, 2018 4:17pm
Just remember that the testing stage doesn't really give you vital information that you need to know if the EA is going to work. Demo testing informs you of bugs and provides the real clues into the profitability of a EA. I run them for 3 months in demo and longer if I need to change a few settings. The testing stage often misleads you into thinking the EA is better than it actually is. I find the Draw down is often twice the amount indicated within the testing stage if you are running multiple charts within a account.
Lastly I often test the same EA on both Pepperstone & IC Markets at the same time and get almost the identical results ( maybe 1% difference ). The testing stage is only designed to tell a trader whether a EA COULD be profitable in a Demo account.

My tip is open many demo accounts so you can run a few versions of a EA at the same time. I have 9 Demo accounts and at least 8 of them are either testing EA or running them on Demo.
Trading thin liquidity at the boundary of the charts
  • #14
  • Dec 29, 2018 5:38pm Dec 29, 2018 5:38pm
Quoting RickM
I run them for 3 months in demo and longer
@RickM -- I agree that Demo testing informs you of bugs -- but 3 months or longer?

If you agree that broker data is important then you also understand how important it is to periodically recalibrate with recent broker data -- right? But if you are testing in a Demo account for long periods of time then when you get ready to move to a Real account the indicator settings are stale and need to be recalibrated to take into account changes in the data patterns.

How about testing in a Demo account for, say, a week -- to flush out any bugs or crashes. And then move to a Real *microaccount*. The microaccount serves a couple of purposes -- (1) it's a Real account and, so, performance doesn't merely mimic a Real account -- it is a real account; (2) even if it loses, the losses are minimal and are dwarfed by what you learn from trading your Real account.
  • #15
  • Dec 29, 2018 9:05pm Dec 29, 2018 9:05pm
Quoting RickM
Just remember that the testing stage doesn't really give you vital information that you need to know if the EA is going to work. Demo testing informs you of bugs and provides the real clues into the profitability of a EA. I run them for 3 months in demo and longer if I need to change a few settings. The testing stage often misleads you into thinking the EA is better than it actually is. I find the Draw down is often twice the amount indicated within the testing stage if you are running multiple charts within a account. Lastly I often test the same...
@RickM Thanks for the opinion.

Agree totally, I am doing quite the same way at the moment.

I installed 4 MT4 terminals(IC Market, FXTM, Go Market, IG) on my NAS(Network Attached Storage, in stead of Server), working with same EA, with different setting. It does take sometimes to trial and error!
  • #16
  • Dec 29, 2018 9:19pm Dec 29, 2018 9:19pm
Quoting robots4me
{quote} @RickM -- I agree that Demo testing informs you of bugs -- but 3 months or longer? If you agree that broker data is important then you also understand how important it is to periodically recalibrate with recent broker data -- right? But if you are testing in a Demo account for long periods of time then when you get ready to move to a Real account the indicator settings are stale and need to be recalibrated to take into account changes in the data patterns. How about testing in a Demo account for, say, a week -- to flush out any bugs or crashes....
@robots4me

I did try to work on Real *nanoaccount(cent)* after 3 weeks on demo, but the result was quite different. I think because the spread and slippage for the nanoaccount was too big! So the drawdown % too high, so I close all the trades. and back to demo again.

However, to check the EAs working in vital status did help!

Thanks for the opinion.
  • #17
  • Edited 11:21pm Dec 29, 2018 11:11pm | Edited 11:21pm
Quoting robots4me
{quote} @RickM -- I agree that Demo testing informs you of bugs -- but 3 months or longer? If you agree that broker data is important then you also understand how important it is to periodically recalibrate with recent broker data -- right? But if you are testing in a Demo account for long periods of time then when you get ready to move to a Real account the indicator settings are stale and need to be recalibrated to take into account changes in the data patterns. How about testing in a Demo account for, say, a week -- to flush out any bugs or crashes....
Hi Robots4me

Three months would me the minimum period of a testing program because you would need to test it through some Red News events like US interest rises or other major world events. Also this would allow for reasonable sample of that years cycle of a trending market type to a consolidation market type.
Then if I was happy with the outcome, I would perform a Beta testing period on a small scale on Live charts for another 3 months, then go full scale thereafter.
In practice you would likely discover the EA needed fixing or it was just a bad model so its binned. Either way the correct period from the first design to complete trading would be close to a year for me.
You did raise a issue about optimizing the EA to current data. This is where EA traders start fighting amount themselves, do you under or over optimize a EA.
It does need to be robust and personally speaking I want all my models to be traded on at least 20+ charts per account. I happily sit on the over optimized side opinion where I will Optimize all my EA's on each chart I trade on a regular period every 3 / 6 / 12 months depending on the model.
But never in the testing period.
I talk to professional and semi professional Algo traders on a weekly basics so I get good understanding on current trends.

The current trend is to under Optimize EA's but that's not my style.
Trading thin liquidity at the boundary of the charts
  • #18
  • Dec 30, 2018 12:30am Dec 30, 2018 12:30am
Quoting RickM
You did raise a issue about optimizing the EA to current data. This is where EA traders start fighting amount themselves, do you under or over optimize a EA.
Hi RickM -- thanks for your thoughtful response. Yes -- "over-optimization", "curve-fitting", "over-curve-fitting" -- does bring up some interesting and diverse perspectives.

Surely -- there is such a thing as "over-curve-fitting". However, as far as I know, no one yet has come up with a way to quantify the level of curve-fitting -- from appropriately-fitted to over-fitted. What seems to be the norm around here is whenever a strategy doesn't perform well then someone starts waving their hands and says it's because it is overly curve-fitted.

In my experience it is very, very hard to overly curve-fit an algebraic formula when you are using a large enough sample size. From a statistical point of view the probability (of over curve-fitting) is very low. So, that is why I keep raising the issue of periodic recalibration. How frequent is "periodic" -- I couldn't say. It depends on the data.

Each symbol has a changing data pattern -- where the data pattern is made up of trend up, trend down, ranging and consolidation. The frequencies of these 4 patterns vary from symbol-to-symbol and with time. A strategy -- being a simple algebraic formula -- is only as good as the chunk of data it was trained against. For example -- suppose your strategy was calibrated against a chunk of data that was 40% trends and 60% consolidation -- i.e. it had never seen ranging data. But, then, by the time you add it to your Real account, the current data is 30% trends, 30% consolidation and 40% ranging. Will the strategy continue to perform well? Probably not. But that isn't because it is overly curve-fitted. Rather, it's because it was originally calibrated against a different data pattern. Just like you wouldn't use GBPJPY data to calibrate a strategy you intend to trade on EURUSD, why would you trade a strategy next week that was calibrated from last year's data? It's not fair to the strategy to expect it to perform well under conditions it was never trained for. Real, hardware instruments require periodic calibration -- and this calibration occurs under conditions similar to how the instrument will be used. This makes sense, right? The same goes for algebraic formulas.

@RickM -- my intent is not to criticize your approach -- obviously it works. Since you mentioned this long testing period in Demo accounts, then I used it as an opportunity to give my little speech about periodic recalibration -- I hope you don't mind... And just to be clear -- there are many details we need to look after in order to trade profitably and I would never claim that recalibration is the key. Rather -- since recalibration is simple and doesn't require much brain power then it seems a shame for it to be overlooked.
  • #19
  • Dec 30, 2018 6:44am Dec 30, 2018 6:44am
I think that people has a tendency to over-complicate things when at the beginning things are simple.

The market has only 2 states, Trending or Ranging. Ranging is a flat market while when it trends, it can go up or down.

So, from this finding, you have 3 possible scenarios:

  1. The market is flat.
  2. The market is in an up trend.
  3. The market is in a down trend.

From there, if you can write an EA capable of performing the best it can in all those scenarios and you have the test data where all those scenarios are present, you will be well and you will also discover that over curve-fitting will become very difficult because what you do for 2 or 3 will fail in 1 and vice-versa.

Simplicity is the Ultimate Sophistication.
  • #20
  • Dec 30, 2018 7:03am Dec 30, 2018 7:03am
Quoting broketrader
The market is flat. The market is in an up trend. The market is in a down trend. From there,
a seven years old kid can notice that after maybe 10 minutes looking at charts !

the question is, where/when to enter and to close?, if its a continuation or a reversal ? nobody know ofc and we just gamble in most of the times

a trader should never concentrate on signals but concentrating on a math algo that make you close in profit always even when you lose 3 on 4 trades or even when you lose for a whole month etc...

a perfect math algo to trade the markets with all the restrictions of brokers is possible to find ? IDK

just saying,,, cheers
0 traders viewing now