Free Coding to backtest your strategy | Page 3 | 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

  • #41
  • Edited 6:36pm Aug 17, 2019 9:41am | Edited 6:36pm
Quoting abokwaik
{quote} What about this one for the first link : DailyTradingUsingSMA5+BB20-2v1.1.ex4 As for the second link, I can see many coders are active on the thread. I am not sure if an EA was made or not. Please confirm. If not, I will need clear system rules as it would be difficult for me scan >3000 posts.
Ah, actually I did see and test that daily EA a while ago. Probably a year ago now. It appears to take a bunch of trades when the candle has not passed through the SMA. Ignoring this key rule:

 

  1. Long - When previous D1 bull candle's body has clearly crossed SMA and did not touch BB upper band.
  2. Short - When previous D1 bear candle's body has clearly crossed SMA and did not touch BB lower band.


You can see several examples within just a few days on this visual test. None of the trades (well, arguably the first one) follow candles that have crossed through the SMA (blue).

In addition, once the EA is made I would like to do some work on how the trades are exited, to maximize profits and minimize losses there. That EA (even if it correctly followed the strategy), obviously doesn't allow for that so doesn't offer much chance of creating something profitable going forward.

ADDED: I think there is a signals indi for this strategy that does the same. My guess is that the EA has been built simply using this indi, therefore repeating its error.

Attached Image
  • #42
  • Aug 17, 2019 5:28pm Aug 17, 2019 5:28pm
Quoting clemmo17
{quote} Easy, maybe, but not necessarily straightforward as I've found after years of attempting to make systems that take advantage of these tendencies. The devil is in the details. Maybe I'm taking liberties by suggesting more than one system but you might like this idea better since it requires no indicators and takes advantage of the first property mentioned. Simply enter after price has moved more than n% of its ATR (or average pip value) without retracing to 50% of its current bar height but only AFTER 80% of its current bar time has...
I like this very much, and I thank you for the detailed description.
Although description was detailed, but not detailed enough, and that's how good ideas start.

So let's get the devil out.

To start with, these are few points that need more details:

  1. Then enter in the direction opposite to the movement of the tall bar (in the direction of the anticipated retracement) with a TP target of the bar midpoint minus a safety factor (5-10%).

    1. This means if we were lucky by selling at the Top of the tall bar, our maximum TP would be 40-45% of the tall bar range. Since we will not be lucky most of the time, most of profits would be in the 0-20% of the tall bar range. Maybe we could limit our entries to occasions where price retracement at entry is less than a certain % (10%), or we could increase our TP, and trail our stop as price goes in our favor
    2. Since we are entering while the bar is still Alive, the bar range could change, for example if we are using H4, and news came in the last 30 minutes of the H4 bar (which is >80% of H4 bar lifetime). In such a case and if price trends after the news, the 50% TP target would be much less than 50%, and probability of retrace would decrease.

  2. And that bar has NOT retraced to its midpoint... (If the bar is not yet closed this is possibly a moving target.)

    1. It is a moving target. But I assume that our interest in the bar starts after 80% of its life time, we do the range check, if OK, we do the retrace check. Since we are targeting 50% of the bar range (at entry time), the retrace % should be much less than 50% so that we have enough pips for our TP target.

No guts, no glory
  • #43
  • Aug 18, 2019 2:21am Aug 18, 2019 2:21am
Quoting josi
{quote} hi there, I'm looking for someone who could coude this EA: entries (underlined words or numbers to be entered into EA by users/extern): Magic number: i. e. 3245678 trade mode: i. e. long/short/long&short lot size (in pips): i. e. 0.1 use money management: yes/no money management: percent of account: i. e. 1 Stop-Loss (in pips): i. e. 20 use Trailing-Stop: yes/no Trailing-Stop (in pips): i. e. 20 use Take-Profit: yes/no Take-Profit (in pips): i. e. 100 Stochastic: Stochastic_1_Period_K: i. e. 8 Stochastic_1_Period_D: i. e. 3 Stochastic_1_Slowing:...
I hear you josi. I assume this will use MT4 standard Stochastic indicator.
Can you be more specific on (and falling) and (and rising)?
No guts, no glory
  • #44
  • Edited 4:51am Aug 18, 2019 2:34am | Edited 4:51am
Quoting abokwaik
{quote} I hear you josi. I assume this will use MT4 standard Stochastic indicator. Can you be more specific on (and falling) and (and rising)?
Thanks, man, for getting back to me.
If you look at the screenshots you'll see that the Stoch32 is moving upwards or downwards (while at the same time being above or below 50)
So - you wouldn't want to take a sell trade (using the Stoch8-cross-down) while the direction/momentum of the Stoch32 is upwards (while below 50) and
you wouldn't want to take a buy trade (using the Stoch8-cross-up) while the direction/momentum of the Stoch32 is downwards (while above 50)
because higher TF-momentum is against you at that moment.
So, if Stoch32 is above 50 && Stoch32momentum is > = what it was during the last candle: buy
So if Stoch32 is below 50 && Stoch32momentum is < = what it was during the last candle: sell
actually we don't know how many candles, as yet - would it be possible to make number of candles an extern function, so we could integrate it into the backtesting
(i. e. 0 = disregard Stoch32 momentum; 1 = momentum higher/lower than last candle; 2 = momentum higher/lower than last two candles etc)
If this is too difficult - just forget about "upwards" and "downwards" for the time being.
Thanks again!
  • #45
  • Aug 18, 2019 7:19am Aug 18, 2019 7:19am
Quoting abokwaik
{quote} Hey Ty, It's been a long time. I hope you're doing well. Such cases can be handled through PMs, but no promises. The idea here is for people to participate and put some effort in designing and testing mechanical systems. Those who give a hand will eventually get rewarded .
I am doing great AB thanks for asking man.

I think I'll send you a PM (especially when you consider that the strategy involves combining pairs and using an offline chart - a.k.a. Basket trading). As far as I know MT4's strategy trader cannot allow backtests using offline charts just like it can't allow backtests involving Renko charts...?

Anyway, I can guarantee that YOU'LL learn a new strategy (fingers crossed)...and so will others (potentially) because those interested in Basket trading will be able to download whatever comes off of your coding from the relevant PUBLIC threads eventually (promise).

See you on the inside.

Regards,

Ty
My job: I speculate and manage risk.
  • #46
  • Aug 18, 2019 8:25am Aug 18, 2019 8:25am
Quoting northwind
Hi Abokwaik, Thank you for starting this interesting project. If possible, I'd like to see how EA with variations work using this simple strategy. Indicator: MLADEN's TMA_CG_NRP ( mq4 and ex4 attached) User Inputs: ...Same as in indicator; ...Lot Size; ...Select either - TP & SL, and/or Trailing Stop Distance; ...Exit Method. Enter on Signal from Indicator. Exit Choices: ...Price Crosses Center Line; ...Price Crosses Opposite Band; ...Hits Trailing Stop. Let me know if you have questions. Thank you again. All the best, John {file} {file} {image}...
Hi northwind,

Thanks for this addition. Great input from you. Let's hope AB can create magic with this!

Regards,

Ty
My job: I speculate and manage risk.
  • #47
  • Aug 18, 2019 11:46am Aug 18, 2019 11:46am
Quoting TyLamai
{quote} Hi northwind, Thanks for this addition. Great input from you. Let's hope AB can create magic with this! Regards, Ty
Hi Ty,

Thank you for your reply.
I think that repainting is not as big an issue when using higher time frames.
And, using another confirming indicator to time entry and exits.
As Derekw commented in post #24, he has been trading profitably using just Mladen's TMA-CG indicator.
Here is another example with Abokwaik's TMA+CG mladen NRP-AK and Hull Moving Average for confirming filter.
Attached Image (click to enlarge)
Click to Enlarge Name: Image1.png Size: 31 KB
Attached File(s)
File Type: ex4 Hull moving average_mladen.ex4   18 KB | 317 downloads
File Type: mq4 Hull moving average_mladen.mq4   14 KB | 422 downloads
Perfidious Pips
  • #48
  • Aug 18, 2019 12:23pm Aug 18, 2019 12:23pm
Quoting MathTrader7
{quote} I haven't yet seen a nice looking indicator which is not repainting!
Only closed candle doesn't repaint

The rest repaints due to their construction, even popular oscillators use an average approximation over discrete data (closed candles) so there is always some repainting. The worst are the polynomial regression (bands) and they can repaint over 5-20 bars depending on their construction, these are loved by system sellers showing how good the historical signal was but it is impossible to catch it real-time because it didn't exist.
  • #49
  • Aug 18, 2019 12:35pm Aug 18, 2019 12:35pm
Quoting northwind
{quote} Hi Ty, Thank you for your reply. I think that repainting is not as big an issue when using higher time frames. And, using another confirming indicator to time entry and exits. As Derekw commented in post #24, he has been trading profitably using just Mladen's TMA-CG indicator. Here is another example with Abokwaik's TMA+CG mladen NRP-AK and Hull Moving Average for confirming filter. {image} {file} {file}
Yes Northwind you are quite correct about higher timeframes being stable - I can attest to that.

I trade using the H4 chart and most of what you see in hindsight were so at those points while trading live. Those signals that weren't are mostly negligible anyway.

Regards,

Ty
My job: I speculate and manage risk.
  • #50
  • Aug 18, 2019 12:38pm Aug 18, 2019 12:38pm
Quoting abokwaik
I've reproduced 130+ research papers about "predicting the stock market", coded them from scratch and recorded the results.
What is worth reading in this area? I would like to read 10-20 papers to have an independent opinion, which ones would you suggest, please? Preferably last 2-3 years if possible. Last time I tried resigned after quite funny paper about "testing" predictive power of various averages.

Quoting abokwaik
The two positive take-aways that I did find from all of this research are:

  1. Almost every instrument is mean-reverting on short timelines and trending on longer timelines. This has held true across most of the data that I tested. Putting this information into a strategy would be rather easy and straightforward (although you have no guarantee that it'll continue to work in future).
  2. When we were in the depths of the great recession, almost every signal was bearish (seeking alpha contributors, news,

...

Thank you for this input.

1. It would be great to discuss further this strategy, the first question I see here is when we will have a confirmation (the first, the second) of the reversal of the trend, how this confirmation would look like. Which "signals" would likely begin the new trend - knowing that we can pretty easy construct something "safe"- trading further confirmations or signals after them.

2. I saw a guy who has made pretty money trading technical entries into a fundamental situation he spotted. He was a "middle term" fundamental trader before, and then "scalped" using TA into trends he foresaw.
  • #51
  • Edited 2:35pm Aug 18, 2019 2:23pm | Edited 2:35pm
TMA REVERSALS EA (take 1)

Inputs :

Attached Image (click to enlarge)
Click to Enlarge Name: TMAR_Inputs.png Size: 97 KB


I like the "REVERSAL STRATEGY" mode , see back test on GY H1 (March 1st to June 20, 2019) with 9.99 quality. Settings file attached.

Attached Image (click to enlarge)
Click to Enlarge Name: TMAR_bt_GY_2019.png Size: 120 KB



EDIT : almost forgot. It uses the "TMA+CG mladen NRP-AK" custom indicator. (attached)
Attached File(s)
File Type: rar TMAR_GY_H1_2019.rar   < 1 KB | 831 downloads
File Type: ex4 TMA Reversals V1.00.ex4   30 KB | 895 downloads
File Type: ex4 TMA+CG mladen NRP-AK.ex4   31 KB | 818 downloads
File Type: mq4 TMA+CG mladen NRP-AK.mq4   20 KB | 950 downloads
No guts, no glory
  • #52
  • Edited 5:35pm Aug 18, 2019 4:00pm | Edited 5:35pm
Quoting abokwaik
{quote} I like this very much, and I thank you for the detailed description. Although description was detailed, but not detailed enough, and that's how good ideas start. So let's get the devil out.
Agreed.

Quote
To start with, these are few points that need more details: Then enter in the direction opposite to the movement of the tall bar (in the direction of the anticipated retracement) with a TP target of the bar midpoint minus a safety factor (5-10%). This means if we were lucky by selling at the Top of the tall bar, our maximum TP would be 40-45% of the tall bar range.
Correct.
Quote
Since we will not be lucky most of the time, most of profits would be in the 0-20% of the tall bar range. Maybe we could limit our entries to occasions where price retracement at entry is less than a certain % (10%), or we could increase our TP, and trail our stop as price goes in our favor
Potentially, but this is a strategy that depends on plentiful entries. The danger to this strategy is a sudden trend, not earlier-than-expected reversion. If we are only capturing 20% of the range I would be happy because it means the price is still ranging which means our risk is low.

Quote
Since we are entering while the bar is still Alive, the bar range could change, for example if we are using H4, and news came in the last 30 minutes of the H4 bar (which is >80% of H4 bar lifetime). In such a case and if price trends after the news, the 50% TP target would be much less than 50%, and probability of retrace would decrease. And that bar has NOT retraced to its midpoint...
Yes, this is the worst-case scenario. To avoid disaster there is the 3x signal range stop. However if price doesn't reach that point it is surprising, even after strong unexpected news, how often (and quickly) price comes back to 50% of the tall bar. This is often the last time price comes back to that point, before it begins trending, but this 'whiplash' is a feature of sudden price movement, generally.

Quote
(If the bar is not yet closed this is possibly a moving target.) It is a moving target. But I assume that our interest in the bar starts after 80% of its life time, we do the range check, if OK, we do the retrace check. Since we are targeting 50% of the bar range (at entry time), the retrace % should be much less than 50% so that we have enough pips for our TP target.
Correct. If too much retracement has already occurred then the trade is not worth pursuing. There are two (maybe three?) ways to handle this.

  1. Enter immediately after a signal occurs. This means price has not had time to retrace and has exceeded our arbitrary bar height. In this case we must suffer if price continues to move in the impulse direction before the retrace begins in earnest. We could enter again once price has begun to move back towards the midpoint of the tall bar. Then the profit of the second trade will offset the minimal profit or loss of the first trade. However in the event this is the start of a trend we will suffer doubly. Since price is truly ranging more often than it is truly trending this may still be acceptable.
  2. Always wait for 80% time to elapse in the signal bar to capture the majority of the impulse move and avoid premature entries. In practice it seems that most retracements begin around that time, but of course this can vary. The drawback of this approach is that some retracements will start earlier than 80% and may be completely finished before we are prepared to enter. Of these two approaches I prefer this one because it is safer, though possibly less profitable. It would be of interest to test both and see which is better.
  3. If it's possible to program this, I don't know, wait for 'the retrace to begin in earnest'. I guess this would be like a type of reverse trailing stop, a 'trailing entry' if you will, where some self-adjusting invisible level is trailing behind the impulse move waiting for it to get crossed and signalling that the retrace has begun. If this is possible it might be the best of both worlds.

There is one more situation that I neglected to mention. In cases where the signal bar is >> than average - say 2.5x an average signal bar height (a 'mega-bar' which occurs in < 20% of all signals) then the reversion trade is unwise and instead trend-mode should be immediately activated once such a bar is detected. These tend to continue in the direction of the impulse and retrace only after a long time, if at all.

  • #53
  • Aug 18, 2019 7:07pm Aug 18, 2019 7:07pm
Quoting abokwaik
TMA REVERSALS EA (take 1) Inputs : {image} I like the "REVERSAL STRATEGY" mode , see back test on GY H1 (March 1st to June 20, 2019) with 9.99 quality. Settings file attached. {image} EDIT : almost forgot. It uses the "TMA+CG mladen NRP-AK" custom indicator. (attached) {file} {file} {file} {file}
Is it possible to get a template file that has the EA attached so that I can test this in the FXBlue Trade Simulator?
  • #54
  • Aug 18, 2019 7:20pm Aug 18, 2019 7:20pm
Great results!
just do it
  • #55
  • Aug 19, 2019 3:36am Aug 19, 2019 3:36am
Quoting Fred-IV
{quote} these are loved by system sellers showing how good the historical signal was but it is impossible to catch it real-time because it didn't exist.
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.
Trading is the hardest way to make easy money...
  • #56
  • Aug 19, 2019 6:36am Aug 19, 2019 6:36am
TMA Reversal EA

Another backtest on EU. Settings file attached.

Attached Image (click to enlarge)
Click to Enlarge Name: TMAR_bt_EU_2019.png Size: 98 KB



Back testing is a good way to improve a good system by -for example- spotting periods of big draw downs and putting solutions to minimize trades during those period or any other way to minimize the bad effect on equity in such situations.

I don't have enough tick data with 99.9 quality, so I would like to see longer periods tested and results posted from members who have such data.
Attached File(s)
File Type: rar TMAR_EU_H1_2019.rar   < 1 KB | 708 downloads
No guts, no glory
  • #57
  • Edited 6:51am Aug 19, 2019 6:38am | Edited 6:51am
Quoting clemmo17
{quote} Is it possible to get a template file that has the EA attached so that I can test this in the FXBlue Trade Simulator?
I don't quite understand the request. It looks to me that you can do that yourself, can't you? why?

EDIT: Ooh, I know why. I will think of a solution to that.
No guts, no glory
  • #58
  • Aug 19, 2019 6:56am Aug 19, 2019 6:56am
want an ea 10 ema and 300 ema cross and cut on 10 ema and 100 in 1 hr chart
  • #59
  • Edited 7:40am Aug 19, 2019 7:15am | Edited 7:40am
Quoting abokwaik
TMA Reversal EA Another backtest on EU. Settings file attached. {image} Back testing is a good way to improve a good system by -for example- spotting periods of big draw downs and putting solutions to minimize trades during those period or any other way to minimize the bad effect on equity in such situations. I don't have enough tick data with 99.9 quality, so I would like to see longer periods tested and results posted from members who have such data. {file}

Thank you Abokwaik,

I ran back test with Original and basically default settings (H4 time frame and changed lots to 0.2).
For three weeks 02/01/2019 to 02/21/2019
Looking at the results tells me I have to adjust the settings.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Image1.png Size: 17 KB
Click to Enlarge Name: Image2.png Size: 14 KB
Click to Enlarge Name: Image3.png Size: 30 KB
Click to Enlarge Name: Image4.png Size: 26 KB
Perfidious Pips
  • #60
  • Aug 19, 2019 7:34am Aug 19, 2019 7:34am
Quoting abokwaik
TMA Reversal EA Another backtest on EU. Settings file attached. {image} Back testing is a good way to improve a good system by -for example- spotting periods of big draw downs and putting solutions to minimize trades during those period or any other way to minimize the bad effect on equity in such situations. I don't have enough tick data with 99.9 quality, so I would like to see longer periods tested and results posted from members who have such data. {file}
I just ran a backtest with the setfiles you posted in #56 on my favourite mean reverting pair AUDSGD from Aug. 2010 to today (99.0 % quality). The only deviations were:
- TF: 4H
- Trading Mode = Original Strategy
- Use MoneyManagement: false
Attached Image (click to enlarge)
Click to Enlarge Name: Image2.png Size: 46 KB
Attached File(s)
File Type: zip AUDSGD 4H.zip   < 1 KB | 368 downloads
0 traders viewing now