LRSF - TMS Expert Advisor | Page 46 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

LRSF Monitor - Assistant for TMS trading 28 replies

LRSF - TMS Expert Advisor - MT5 85 replies

TMS basket: Dashboard for a basket of MA-Filtered RSI Signals 135 replies

PA patterns with TMS system 45 replies

Polapip The Meandering System (TMS) 29 replies

Attachments: LRSF - TMS Expert Advisor
Exit Attachments

LRSF - TMS Expert Advisor

  • #901
  • Sep 5, 2018 2:09am Sep 5, 2018 2:09am
Dear Janb000,
I have the first backtesting with optimazation. The results I send as attachments are perfect yet? Can you guide me how to submit the most detailed result report. Thank you very much.
Attached File(s)
File Type: pdf inputEA1.pdf   111 KB | 221 downloads
File Type: pdf Strategy Tester_ BiM_LRSF_19_opt_H1_EU.pdf   107 KB | 189 downloads
Good luck to all
  • #902
  • Sep 5, 2018 9:52am Sep 5, 2018 9:52am
Quoting forexhanh47
Dear Janb000, I have the first backtesting with optimazation. The results I send as attachments are perfect yet? Can you guide me how to submit the most detailed result report. Thank you very much. {image} {image}
I think you are on a right track.
It's just the purpose of backtesting is to test an ea on a long period of time.
May be a year, or more.
As for optimization - the same story.
The right way to do it, according to the "book", is to optimize it on some time period in the past. For example, from February 5th, 2018 to July 5th, 2018
Then sort the results, best to worse, by clicking on the top of optimization results. Do not close those results. Copy them and save somewhere. I save in Excel.
Then double click, let's say, the best result. Now those settings are automatically loaded in your ea, in your tester.
Now you run backtest, and forward test, by running an ea in your tester without optimization from December 5th, 2017 to September 5th, 2018
See, if you like the results. If not - use some other settings from your optimization table. Until you find what you really like.
  • #903
  • Sep 5, 2018 12:08pm Sep 5, 2018 12:08pm
Quoting forexhanh47
[quote = jansb000; 11440738] {quote} The ATR is an indicator that comes with the installation of MT4. Maybe you accidentally deleted it from the Indicators folder? If the ATR is missing then other pairs should have trouble as well. So you have to fiddle a little bit to find the problem. [/ quote] Dir Jansb000, i just learn forex few months, knowledge is very limited but i will try to improve my backtesting. Hope I will have better backtesting results to send to you. Can you post a link on how to calculate ADR 100 in your EA based on the ATR available...
Your calculation is correct !!!

Inserted Code
   // Average True Range
   // We use this indicator to find key values for setting SL and TP.
   atr.setup(lv_sym_symbol, lv_tframe, 26, 0);
   // Initial value for the Average Daily Range
   double lv_adr100 = atr.value(lv_sym_symbol, PERIOD_D1, 100, 0);
   gADR100 = NormalizeDouble(val2pips(lv_adr100), 0);
   if (gADR100 < gSymbol.get_point()) {
      Comment("ERROR: ADR(100) can not be initialised !!");
      return(INIT_FAILED);      
   }
  • #904
  • Sep 5, 2018 12:16pm Sep 5, 2018 12:16pm
Quoting stef241
Hi guys, I'm a silent reader of this thread since a longer time and decided to join ff for giving some feedback. I really appreciate your work and the nice ideas you came up with. However, some remarks from side. I trade an automated "TMS strategy" with a lot of filters in forward test myself. By seeing the backtests and previous trade reports the huge floating loss/drawdown is a serious issue. The balance/equity curve looks more like an EA a grid system would produce, so I think the MEME component is really dominating and would for 100% kill an...
You would have to read the TMS thread (link in post #1) to learn how to use MEME. Like you say, it should not be needed but in auto-trading it often saves the day. The trick is to decide, when to NOT use it.
The lower the trading timeframe, the more often the big trend will not be recognised, and MEME will destroy your account eventually.
  • #905
  • Sep 5, 2018 12:34pm Sep 5, 2018 12:34pm
Quoting stef241
...Can somebody please backtest without using MEME to see the real "edge" of the EA?...
Of course, since backtesting is only possible since a very short while (due to a bug with the slippage setting), we might find that certain design desicions are not selected well and some "corrective programming" might be in order.
I do not pretend that this a finished and stable product. Contrary to that, I am still developing this EA.
  • #906
  • Sep 6, 2018 3:19am Sep 6, 2018 3:19am
Quoting Zazazaz
{quote} Until you find what you really like.
Dear Zazazaz,
Thank you for the backtesting tutorial.
From the Excel data storage, EA parameters can automatically update to the EA? Or you have to declare each parameter based on this Excel.
Thank you
Good luck to all
  • #907
  • Edited 4:53am Sep 6, 2018 4:25am | Edited 4:53am
Strategy Tester Case

I decided to make an example of a Strategy Tester session. Just to show how it works. I did not download detailed data so the quality of this test is not great (just 45%). I just want to show the process.

First, I had a run of my LRSF EA Version 21. I used EURUSD on H1.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 45 KB

With this (limited) dataset we can see that the EA made profit, but there was also a big fall due to a big retrace of the price. The retrace was so big that the default setting of 6 MEME's wasn't enough to solve this problem.
So, how many MEME's would we need for optimal performance? This is where the Strategy Tester comes in handy...

I am going to vary the maximum number of MEME's to see what the effect on the test results is.
Attached Image

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot3.png Size: 5 KB

You can see that I decided to vary the setting from 0 to 20. At the same time I decided to have the setting for using Dynamic MEME part of the test as well.

The result looks interesting:
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot4.png Size: 65 KB

It appears that a setting of 7 or higher would give better results than with 6 MEME's as is the default.
In the result-list I sort for profit, and with a right click I copy the settings for a final test of just this single setting.
Attached Image

Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot6.png Size: 18 KB

Now we have a result that looks a bit better. Similar tests with other pairs must verify if a MEME setting of 7 will indeed be better. It also shows that using the new Dynamic MEME with RSI gives slightly better results then the MEME with fixed values.

What next:
For this pair, I would now set the MEME to 7 and start testing with the other settings to see if I can optimise the results further. With those results I would have to decide if I am happy with the outcome. Certain settings may look good in theory but are not good in practice.

This simple test of just two parameters took the better part of a whole night (6 hours) on my old PC. So, doing a full optimisation of all parameters can take a long time (specially on an old slow PC).
Certain parameters have nothing to do with the algorithm and hence there is no point making then vary per test. For example the lot-size is a parameter, but changing this one makes no sense. Same goes for username, magicnumber.

Hope this gives you some ideas on how to proceed with Strategy Testing.
Cheers...
  • #908
  • Sep 6, 2018 4:45am Sep 6, 2018 4:45am
Quoting jansb000
{quote} WOW, i asked for backtesting and I am happy to see that I got plenty of results. Thats great. I do notice that running the EA on H4 seems to be safer that running it on H1 !!! That is because on H1 the EA doesn't recognise the correct trend. I mentioned this risc before and it seems to be a real issue. I will work on that a.s.a.p. For now I would advise to only use the EA on H4 or D1. Also I notice that on many of the backtests the modeling quality is very low, from unknown to approx. 65%. This needs to be >90% to be a little bit reliable....
Hello Jans,
For the optimiztion testing, shall I put 'tick' mark in all columns, with the default numbers already there ?
Please let me know.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 217 KB
  • #909
  • Sep 6, 2018 8:49am Sep 6, 2018 8:49am
With a bit of fiddling I got this result:
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 24 KB

This was done by switching trend-filter lvl2 => false

Next i'll investigate the impact of the "Confirmation MA".
This will run for approximately 70 hours !!!
  • #910
  • Sep 6, 2018 8:59am Sep 6, 2018 8:59am
Quoting mavilakandy
{quote} Hello Jans, For the optimiztion testing, shall I put 'tick' mark in all columns, with the default numbers already there ? Please let me know. {image}
No this is not needed. Just leave them all as they are.
These are settings to abort the Strategy Test. When something goes horribly wrong during the test you don't want to wait for days until all the tests are finished only to find out that after 2 simulated orders your account was already dead.
We are now at a stage where we want to see where the EA goes. Without restrictions.
  • #911
  • Edited 9:21am Sep 6, 2018 9:05am | Edited 9:21am
Proposal for strategy testing

Maybe it would be an idea to co-ordinate the tests so that everybody works on a different set of pairs/parameters.
just before you start a test, you might want to post here exactly what you are going to test, so others know that they can focus on other parameters or pairs.

Note: It is best to focus on a small number of pairs. Because for every version of the EA you potentially want to repeat the tests.
Cheers...
  • #912
  • Sep 6, 2018 9:14am Sep 6, 2018 9:14am
Data Quality 90%

I managed to UP my data quality to 90%. This is still too low, but it will give some directions on the impact of some parameters. To get your data quality UP as well follow these steps:

  1. Go to menu Tools
  2. Select History Center (F2)
  3. Find the pair that you want to test
  4. Double click on that pair

    -> you will now see a list of timeframes, possibly grey. If it is grey no data is available. With other colors data might be available but it might still not be up to date.
  5. Select every timeframe (one by one) and click [Download].

The missing data on your computer will now be loaded from the MetaQuotes website. The data from MQ might be a bit different from the data from your broker. That is why it is best to have a separate installation of MT4 specially for testing.

With additional tools (like TickStory) you can download even better quality data, and your test results will be a lot more realistic.

  • #913
  • Edited 11:13pm Sep 6, 2018 7:39pm | Edited 11:13pm
Quoting jansb000
Data Quality 90% I managed to UP my data quality to 90%.
Dear Jansb000,
I understand.
I will start backtesting under the best data that I can have. The parameter ea will be the optimal parameter. I will check all currency pairs on H4 because I see this should run only on H4 and D1. How do I check backtesting results?

I'm running your EA on a real account (0.01 lot - H4). From the beginning of the week to yesterday there are 2 orders. All are profitable. Take a look at the picture. H4 is a good timeframe for your EA
Thank you very much, I'm trying to learn forex every day and very hard. I hope your EA will succeed. I always support you
Attached Image (click to enlarge)
Click to Enlarge Name: 7.9.2018_real3_Exness.jpg Size: 594 KB
Good luck to all
  • #914
  • Sep 7, 2018 1:34am Sep 7, 2018 1:34am
Dear Jansb000,
Optimization MEME + Backtesting EURUSD H4 from 1/1/2018 to 31/8/2018 with MEME optimization input
Attached Image (click to enlarge)
Click to Enlarge Name: MEMEoptimization.jpg Size: 225 KB
Attached File(s)
File Type: xlsx ea19_meme.xlsx   10 KB | 129 downloads
File Type: pdf Strategy Tester_ BiM_LRSF_19_H4_Opt_MEME_8months.pdf   130 KB | 189 downloads
Good luck to all
  • #915
  • Sep 7, 2018 12:46pm Sep 7, 2018 12:46pm
Quoting jansb000
Data Quality 90% I managed to UP my data quality to 90%. This is still too low, but it will give some directions on the impact of some parameters. To get your data quality UP as well follow these steps: Go to menu Tools Select History Center (F2) Find the pair that you want to test Double click on that pair -> you will now see a list of timeframes, possibly grey. If it is grey no data is available. With other colors data might be available but it might still not be up to date. Select every timeframe (one by one) and click [Download]. The missing...
Hello Jans,
I am downloading data from MQ: H4 and D1 for GBPJPY and EURJPY.
Since I am not good in these things, please explain, where do I find the downloaded data and how I can use that data for back testing - optimization.
My MT4 platform is of FXCM.
Thanks
  • #916
  • Sep 7, 2018 1:26pm Sep 7, 2018 1:26pm
Hello Jans, well done with your EA, i can see a lot of progress has been made since i last posted here, even though i've been monitoring. Firstly, do you know why the EA does not take trades with my broker; instaforex (i have shared this concern before) where i have a real account. However, i have been testing version 19 on H4 with another broker and it rocks. Over 860 pips made just in a little over a week! This version is very active and profitable and i wish we could stick to it. The only settings i turned off was the 'soft breakeven'. Also, do you still plan to use MM to define percentage risks in lot sizes? That would be a great idea
Attached Image (click to enlarge)
Click to Enlarge Name: Global Prime - MetaTrader 4.png Size: 130 KB
  • #917
  • Sep 7, 2018 1:32pm Sep 7, 2018 1:32pm
Quoting mavilakandy
{quote} Hello Jans, I am downloading data from MQ: H4 and D1 for GBPJPY and EURJPY. Since I am not good in these things, please explain, where do I find the downloaded data and how I can use that data for back testing - optimization. My MT4 platform is of FXCM. Thanks
https://www.babypips.com/trading/for...t-mt4-20140328

https://www.metatrader4.com/en/tradi...ion_parameters
  • #918
  • Sep 7, 2018 7:28pm Sep 7, 2018 7:28pm
Quoting forexhanh47
Dear Jansb000, Optimization MEME + Backtesting EURUSD H4 from 1/1/2018 to 31/8/2018 with MEME optimization input {file} {image} {image}
First read and implement the procedure of post #912. Get your data quality up to 90%.
  • #919
  • Sep 7, 2018 7:31pm Sep 7, 2018 7:31pm
Quoting mavilakandy
{quote} Hello Jans, I am downloading data from MQ: H4 and D1 for GBPJPY and EURJPY. Since I am not good in these things, please explain, where do I find the downloaded data and how I can use that data for back testing - optimization. My MT4 platform is of FXCM. Thanks
You must download data for ALL timeframes. The EA does checks on several timeframes.
First step is to get your data quality up to 90%. See post #912 for procedure.
  • #920
  • Sep 7, 2018 7:37pm Sep 7, 2018 7:37pm
Quoting timidave
...Firstly, do you know why the EA does not take trades with my broker; instaforex...
I really can not say why. Maybe you can post a screenshot of the trader (specially if you see a smiley on the right and the messages on the left).
Set the Slippage to 500 for a test. See if that works.

Money Management (MM) as in automatic lotsize calculation is not on my shortlist. There are more important things that I have to solve before I can do that.
Sorry timidave!
1 trader viewing now