Free Coding to backtest your strategy | Page 6 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Free Coding: your systems 24 replies

I Will Learn Coding in 1 Month - Give me the 80/20 of MT4 Coding 14 replies

Free Expert Advisor with positive backtest results 5 replies

Visual backtest result different from non-visual backtest 0 replies

MT4 EA - strategy tester - backtest results inaccuracies 15 replies

Attachments: Free Coding to backtest your strategy
Exit Attachments

Free Coding to backtest your strategy

  • #101
  • Aug 22, 2019 7:05am Aug 22, 2019 7:05am
Quoting ots931010
{quote} Just read through all the threads about your amazing system. Could you possibly suggest which type of currency pair best for the strategy? Much appreciated!!
Hi OTS,
Try it on any/all pairs.
Watch how behaves in ranging and trending markets at different times during Asia, Europe and USA trading hours.
Adjust the TMA half length and deviation to suit your trading style.
All the best.
An example with Abokwaik's TMA on EU on H1.
Another GJ M15.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Image1.png Size: 27 KB
Click to Enlarge Name: Image2.png Size: 26 KB
Perfidious Pips
  • #102
  • Aug 22, 2019 3:41pm Aug 22, 2019 3:41pm
Quoting itsDavid
{quote} I'll take a look at the first EA right now, thank you! I must have missed that post. For the second link. As far as I am aware there has been much talk of an EA but no one has actually made one. The entire rule-set is in post one which has been kept updated with any changes, so no need to scan though the entire thread. It certainly appears to be profitable from manual back and forward testing. It could be an excellent EA due to the simplicity of the rules. It works on all major pairs. There are refinements which could be made of course....
Trading Made Simple(r) @ RobinHood

TMSR EA (take 1)

Attached Image (click to enlarge)
Click to Enlarge Name: tmsr_inputs.png Size: 91 KB


Sample test on EU/H4 Jan-April/2019 (settings file attached)

Attached Image (click to enlarge)
Click to Enlarge Name: tmsr_bt_EU_H4.png Size: 114 KB
Attached File(s)
File Type: ex4 TMSR EA v1.00.ex4   35 KB | 1,025 downloads
File Type: rar tmsr_EU_H4.rar   < 1 KB | 979 downloads
No guts, no glory
  • #103
  • Aug 22, 2019 6:29pm Aug 22, 2019 6:29pm
Two graphs for Stoch2, default settings.

USDJPY D1 2011.11.15- present
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 38 KB


USDJPY H1 2017.06.03 - present
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot2.png Size: 26 KB


I hope these are not discouraging. I don't put ANY faith in the MT4 tester except for the simplest of strategies. Of course I am too lazy to use tickstory and my data is 'bad' but I don't believe it makes any difference. What is really bad is the MT4 strategy tester which cannot take into account things like the spread change over the weekend.

The only way to really test something is live forward; maybe demo is OK.
  • #104
  • Aug 22, 2019 7:20pm Aug 22, 2019 7:20pm
Quoting abokwaik
{quote} Trading Made Simple(r) @ RobinHood TMSR EA (take 1) {image} Sample test on EU/H4 Jan-April/2019 (settings file attached) {image} {file} {file}
Have you the HA and HMA indicators needed for this strategy? I know I have similar ones but I don't want to muddy the waters with variations.
  • #105
  • Aug 22, 2019 10:45pm Aug 22, 2019 10:45pm
Abokwaik,

Thank you for your work to produce and share the testing EA.
TMA Reversals EA test results EU H1.
Looks like even better results are possible with optimizing EA settings.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Image1.png Size: 36 KB
Click to Enlarge Name: Image2.png Size: 33 KB
Attached Image
Perfidious Pips
  • #106
  • Aug 23, 2019 2:07am Aug 23, 2019 2:07am
Quoting northwind
Abokwaik, Thank you for your work to produce and share the testing EA. TMA Reversals EA test results EU H1. Looks like even better results are possible with optimizing EA settings. {image} {image} {image}
You need to test over 5 to 10 years minimum to understand long expectancies.
Only than an EA cab be judged
  • #107
  • Aug 23, 2019 2:11am Aug 23, 2019 2:11am
Quoting northwind
Abokwaik, Thank you for your work to produce and share the testing EA. TMA Reversals EA test results EU H1. Looks like even better results are possible with optimizing EA settings. {image} {image} {image}
10 trades - wow!
and results after optimisation - wow! (window dressing - is this still a phrase?)
that will really rock the boat - Wall Street will come down, once you release this.
  • #108
  • Aug 23, 2019 2:20am Aug 23, 2019 2:20am
Quoting MathTrader7
{quote} That's all true! Plus the fact that using a high degree of freedom in the input settings, a heavily optimized EA of using such repainting indicators would result in a nice looking balance curve (due to overfit problem) for the selected period of optimization, but the same settings would fail dramatically on other periods.
Optimization is almost "the root of all evil" I don't know where from this came to trading, maybe from software developers community, but this tool is really harmful to everyone who believes in it. Some, maybe "almost all" approaches that guys from IT have in trading are destined to fail imho, that's why 95% of robots fail and will fail in the future. I think that different angle is needed, even several different angles to create something stable (I mean with stable profits).

As for repainting - your approach with Renko is I believe far more robust, Renko doesn't repaint and if a signal is present it is for good, a good point of start I think, in comparison with candle approach the signal is sooner which is a huge plus too.
  • #109
  • Aug 23, 2019 4:34am Aug 23, 2019 4:34am
Quoting abokwaik
{quote} Trading Made Simple(r) @ RobinHood TMSR EA (take 1) {image} Sample test on EU/H4 Jan-April/2019 (settings file attached) {image} {file} {file}
Wow, thank you! I'll test this now.
  • #110
  • Aug 23, 2019 4:58am Aug 23, 2019 4:58am
Quoting clemmo17
{quote} Have you the HA and HMA indicators needed for this strategy? I know I have similar ones but I don't want to muddy the waters with variations.
This is the correct HMA one. You can also download a zip of everything in the first post, here: https://www.forexfactory.com/showthread.php?t=917569
Attached File(s)
File Type: ex4 HMA.ex4   11 KB | 395 downloads
  • #111
  • Aug 23, 2019 5:44am Aug 23, 2019 5:44am
Quoting abokwaik
{quote} Trading Made Simple(r) @ RobinHood TMSR EA (take 1) {image} Sample test on EU/H4 Jan-April/2019 (settings file attached) {image} {file} {file}
My apologies, I forgot to mention about the indicators used. All indicators/templates/rules are on post#1 of the Trading Made Simple(r) thread.

I compared the custom indicators with their MT4 standard versions, and they were the same except for few cosmetics which are not needed for the EA.
The only custom indicator that does not have a standard version is HMA (Hull Moving Average), which can be found on the link mentioned above.

So I used MT4 standard versions in TMSR EA whenever possible.
No guts, no glory
  • #112
  • Aug 23, 2019 6:12am Aug 23, 2019 6:12am
Quoting Fred-IV
{quote} Optimization is almost "the root of all evil" I don't know where from this came to trading, maybe from software developers community, but this tool is really harmful to everyone who believes in it. Some, maybe "almost all" approaches that guys from IT have in trading are destined to fail imho, that's why 95% of robots fail and will fail in the future. I think that different angle is needed, even several different angles to create something stable (I mean with stable profits). As for repainting - your approach with Renko is I believe...
Hi Fred-IV,

Thanks for your comment.
I agree with you, "optimization" will kill an EA using this approach.
Strategy tester doesn't give results that represent live trading or even forward testing.
Rather, it gives an idea whether an EA strategy is more or less working.
But, from my simple back tests, manual testing, and from the tests done by others, an EA using this strategy offers additional, major pips without overfitting.
Perfidious Pips
  • #113
  • Aug 23, 2019 6:25am Aug 23, 2019 6:25am
Quoting Merka
{quote} You need to test over 5 to 10 years minimum to understand long expectancies. Only than an EA cab be judged
Hi Merka,

Thanks for your comment. I agree.
Strategy Tester results with 10 trades does not mean much in the context of successful live EA trading.
What 10 does show is whether the EA is taking trades as it is supposed to. Which is what I had been trying to find out.
Much longer test runs would be ideal to show if this strategy could be used for live trading..
Others have done longer test runs (of weeks, not years), others have done manual forward testing for years.
I am in the process of downloading historical data from Tickstory (as suggested by Abokwaik) and plan to run longer back tests.
This all takes time.
Perhaps you'd like to pick a pair and a time frame and run the EA in strategy tester and post the results.
Everyone would benefit.
Perfidious Pips
  • #114
  • Aug 23, 2019 7:16am Aug 23, 2019 7:16am
Quoting northwind
{quote} Hi Fred-IV, Thanks for your comment. I agree with you, "optimization" will kill an EA using this approach. Strategy tester doesn't give results that represent live trading or even forward testing. Rather, it gives an idea of whether an EA strategy is more or less working. But, from my simple back tests, manual testing, and from the tests done by others, an EA using this strategy offers additional, major pips without overfitting.
Thank you for your comment, I think I understand more or less your approach, roughly you want to know if there is potential in a robot or strategy. Certainly, visual testing can help in finding classes of trader that can be eliminated or focused on, doing this manually would take ages.

Generally, I think that backtesting or forward testing is deceptive on the verge of useless in EA's assessment (with exceptions of few certain classes of systems), that is one of the reasons I suppose most of the robots don't and won't work.
  • #115
  • Aug 23, 2019 8:34am Aug 23, 2019 8:34am
EU H4 TMSR 2010.01.15 to 2014.03.14 (stoppped-out); using 3000 USD test balance.
Attached Image (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 22 KB


I would be curious to know if someone else would test EU H4 during the same period using Khalil's set file and see if they get the same result? You can see how bad the data is (44.48%!) and maybe that's skewing my results.

Maybe I should open a new tickmill demo account as they seem to generate 'accurate' results.
  • #116
  • Aug 23, 2019 8:34am Aug 23, 2019 8:34am
Quoting abokwaik
{quote} Trading Made Simple(r) @ RobinHood TMSR EA (take 1) {image} Sample test on EU/H4 Jan-April/2019 (settings file attached) {image} {file} {file}
Hi

Firstly, thanks for the work you have put into this. It is definitely worth looking at as an EA since the strategy itself is so mechanical. And I know from both backward and forward testing that it works, and makes good money.

I have been testing the EA today and it appears to be taking trades several bars too late in many instances. I have posted one below. It isn't a special example, or an unusual one. Just the one that happened to be on the screen at the time. As you can see the EA waits for 3 candles more than it should have.

Additionally, I know you are still working on the continuation trade part of the EA. This is the area which will be more stable, with much less chance of reversal and so the lowest drawdown. So it is very much worth finishing.

I'm not sure why the EA waits here (and in many other examples), all the mechanics need for a trade are in place three candles sooner. Is the EA maybe waiting for the stochs to change colour?

In live trading this strategy it is a good idea to check the M5 and D1 charts before taking the trade to see if they agree with the trend on the H4. This could be something to add to the EA strategy later on.

Thanks again

David
Attached Image (click to enlarge)
Click to Enlarge Name: incorrect.jpg Size: 708 KB
  • #117
  • Aug 23, 2019 9:30am Aug 23, 2019 9:30am
Quoting itsDavid
{quote} Hi Firstly, thanks for the work you have put into this. It is definitely worth looking at as an EA since the strategy itself is so mechanical. And I know from both backward and forward testing that it works, and makes good money. I have been testing the EA today and it appears to be taking trades several bars too late in many instances. I have posted one below. It isn't a special example, or an unusual one. Just the one that happened to be on the screen at the time. As you can see the EA waits for 3 candles more than it should have. Additionally,...
I appreciate your notes, and will address them in following updates.
As for the chart posted, if you are using the setting file I posted with the EA, that entry was not taken earlier because of Trading Time Settings, as trading hours were set from 10:00 to 18:00.

Attached Image
No guts, no glory
  • #118
  • Aug 23, 2019 11:00am Aug 23, 2019 11:00am
Quoting abokwaik
{quote} Stock2 (take 1) {image} Kindly backtest and let me know of any issues. ----------------------- EDIT : re-uploaded the EA file. (Only Long/Only Short/Long&Short) option was not enabled. {file}
Sorry, man,
the EA doesn't seem to work according to the rules.
It was supposed to check Stoch32
a) above (bullish) or below 50 (bearish)
b) moving upwards (bullish, i. e. from 0 to 100) or moving downwards (bearish, i. e. from 100 to 0)
This was supposed to be a filter for trades.
The entry signal came from the Stoch8
Stoch-line crossing Signal-Line: buy (moving up)
Signal-Line crossing Stoch-line: sell (moving down)

A test shows that the EA takes trades at - for me - random places (see chart)
Maybe you can figure out what's not working or whether I made some mistake here.
Attached Image (click to enlarge)
Click to Enlarge Name: Stoch.png Size: 33 KB
  • #119
  • Edited 12:00pm Aug 23, 2019 11:36am | Edited 12:00pm
Quoting josi
{quote} Sorry, man, the EA doesn't seem to work according to the rules. It was supposed to check Stoch32 a) above (bullish) or below 50 (bearish) b) moving upwards (bullish, i. e. from 0 to 100) or moving downwards (bearish, i. e. from 100 to 0) This was supposed to be a filter for trades. The entry signal came from the Stoch8 Stoch-line crossing Signal-Line: buy (moving up) Signal-Line crossing Stoch-line: sell (moving down) A test shows that the EA takes trades at - for me - random places (see chart) Maybe you can figure out what's not working...
It does work according to the rules.
As you probably know, signal is checked on the Closed bar, if it has a valid signal, Trade is take at the new bar Open.

Your screen shot with my comments. As you can see, the slope of both stoch's was down for all short trades taken.

Attached Image (click to enlarge)
Click to Enlarge Name: Stoch2.png Size: 53 KB


Apparently trend was down, and trading it on the Daily with "suitable" stop loss/trailing stop, would have generated good return.

EDIT 2: This is the same test with SL 200pips (=2000points), no TP, Trailing @ 100 pips triggered after 200pips profit.
Attached Image (click to enlarge)
Click to Enlarge Name: GBPJPYDaily.png Size: 34 KB


--------------------
EDIT 1: Note also the Momentum default is set to 1 Candle. If you want to calculate the slope on more than 1 candle, increase this parameter for both Stoch32 and Stoch8 as required (2, 3, 4 ... ) candles.
No guts, no glory
  • #120
  • Aug 23, 2019 11:50am Aug 23, 2019 11:50am
Quoting abokwaik
{quote} I appreciate your notes, and will address them in following updates. As for the chart posted, if you are using the setting file I posted with the EA, that entry was not taken earlier because of Trading Time Settings, as trading hours were set from 10:00 to 18:00. {image}
Thanks for the explanation. Yes, I used your settings file. I am running the test again now without the time restrictions, to look for improper trades. Going forward we definitely don't want trades being taken too late because of restrictions to the trading time, or for any reason really
0 traders viewing now