Swing momentum trading - the robots are taking over | Page 16 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Swing high / Swing low indictator+alert? 20 replies

Trading Swing by Swing 39 replies

Bad Robots - The Swing n' Miss approach to learning Forex 5 replies

Swing highs and swing lows 3 replies

swing high/swing low indicator and volume tic history 13 replies

Attachments: Swing momentum trading - the robots are taking over
Exit Attachments

Swing momentum trading - the robots are taking over

  • #301
  • Jun 9, 2010 9:44am Jun 9, 2010 9:44am
Quoting squalou
What's funny is that i just spent the past hour adding that to my EA...
oh well... i will take yours if it has more features than mine!

I see Hiredwhip is using MPTM all the time, both for manual and EA trading, so i decided it would be good to add it.
It's probably better to have it builtin rather than sticking to the hard SL/TP.

I also switched to my "reliable" versions of the OrderXxxx() functions -- i do that for all EAs now... it gives much less errors.

We can combine both into a single nice SS EA for sure!

Email me your EA,...
Whichever version you two decide (with Steve's blessings) to post to the group would be appreciated.

Thank you for your contributions!
  • #302
  • Jun 9, 2010 10:51am Jun 9, 2010 10:51am
Quoting octomanoak
Whip,

I see you have added trailing stop and adjusted your hedge settings from the previous set you posted. Was there a particular observation you had on your prior set that triggered this change? Or are you just testing different settings?
Hi Octo,
The reason I did that is I gave both SL and TP an increase from the prior set. I added a little more "safe space"......

whip

Octo, my prime target is to get to the Global Profit setting as fast as possible and then go to the next batch.....when I'm scalping!
  • #303
  • Jun 10, 2010 10:34am Jun 10, 2010 10:34am
Been running the SS bot for about a whole day. Seems to be relaly selective as to entering any trades which is good. One question though to you guys. I've been getting strong sell/buy signal alerts without market enters and I was wondering if that should be the norm. I've read and reread the pdf file but there just isn't much info there to understand.

On all my charts it says "No SS signal" followed on the next line by the trend direction with sometimes an additional "untradable". I guess what I'm confused about is since I'm getting all these alerts for strong buy/sells, why does it not enter? I'm assuming the RSI direction has to be the same as the SS signal but sometimes even if it is the same, no enters.
  • #304
  • Jun 10, 2010 11:43am Jun 10, 2010 11:43am
Hi all,

here is a new version of the SS bot.

I reorganized the inputs so that most important inputs come first.

I added BE and jumpingStop stuff, with the contribution from peejaywk also (thanks!)

These are inspired (borrowed i should say...)from MPTM EA, of course!

Here are the new inputs:

BreakEvenPips=0; // when >0, will move SL to BE+BreakEvenProfit when price has reached BE+BreakEvenPips
BreakEvenProfit=5; // lock-in this amount of pips when BE is triggered;

TrailingStopPips = 0; // Pips to trail the Stop; if 0: no trailing stop
TrailingStopStep = 1; // trailing Stop jumping step in pips

//-------------------------------------------

The "trend" detection is performed using H1 RSI, but i found that using smaller TF would give better results in backtesting.

So i added an input for the RSI TF itself:

RsiTF=1440; // RSI TF: 0=current chart TF, or 1,5,15,60(H1),240(H4),1440(D1),etc
RsiPeriod=21; // when >0, RSI is used as trend filter

RSI trend filter can be DISABLED by setting the RsiPeriod to 0 (or any negative value, this way you can disable the RSI filter and still keep track of the value you were using before... it just saves a "true/false" extra input).

//-------------------------------------------

"SuperSignal" indicator inputs:
i renamed them for clarity.
I removed the "dist1" input as it wasn't used anymore.
It is just commented-out, so you can easily bring it back in;
The code that originally used Dist1 is still there too, but inactive.

SuperSignalShift=1; // this is the original "CandleShift" input;
SuperSignalDist2 = 21;

//-------------------------------------------

Added a call to an external indicator that will draw a 50pips RN grid
I finfd it helpful also in backtesting windows, instead of the crappy useless MT4 default grid (disable this crap with ctrl-G !)

draw_RN_grid = true;

The RN-grid indi is also attached here.
Put it in your "indicators" sub-dir.

//-------------------------------------------

Now, here are my latest findings on EU backtesting:

I found that H1 RSI filter (RsiTF=60) is finally the best compromise, and using RsiPeriod=2 gives the most reactive trades.
Using a slower RSI (higher RsiPeriod values) will revent
Along with SuperSignalShift=3 or 4 will give best results.
Also shorter values of SuperSignalDist2 will work better: typically between 6 and 10, not more.
I use M15 charts only. (Haven't tried any other TF yet.)

The reason is most probably because the H1 RSI will move from the buy to the sell zone (or opposite) quickly enough to catch a swing signal indicated by SuperSignal.
Using SuperSignalShift values >1 will make the signal stronger, avoiding too early swing detections from SuperSignal, which is giving too many early (meaning false) swing signals.
With SuperSignalShift =3 (or 4), on M15 charts, it ensures that the swing signal has not changed in the past 3(4) bars, so the swing is stronger, not just a very short retrace.

Not using the trend filter gives bad results, it seems.
(try the same settings with and without the RSI(H1,2) and you'll see what i mean...)

Now you'll have to play with SL/TP settings...

Results are very sensitive to the TP/SL, and also the TrailingStop value -- somtimes no trailing is better, sometimes trailing set at TP level is better...

Having played quite a lot with optimizing on EU, i can see that optimized values for a given period of time will not work on the long run...

I am testing a "1-week-optimized sliding window" method:
i optimize settings over the past week only (mon-fri), and use the found settings for the coming week.
That way i use the most recent "price pattern".
It seems to give good results.
Very few weeks in loosing land with this method.

I am even thinking of using a "5-days-optimized sliding window" mode, where i optimize over the past 5 days for use for the next day, to give even closer pattern change reaction.
With the recent higher volatility, it is probably the best we can do.


I would be absolutely fantastic if the EA could run the optimizer by itself !!!
It would be just like those Neural Networks that automatically "train" themselves at regular times...

If anyone knows how to trigger the optimizer from an EA, than please share this with us!


Good testing everyone!


Sq
Attached File(s)
File Type: mq4 Steve Hopwood's Super_Signal auto trading robot SQ.mq4   23 KB | 1,193 downloads
File Type: mq4 RN-grid.mq4   4 KB | 970 downloads
  • #305
  • Jun 10, 2010 11:46am Jun 10, 2010 11:46am
Quoting obnoxious2
Been running the SS bot for about a whole day. Seems to be relaly selective as to entering any trades which is good. One question though to you guys. I've been getting strong sell/buy signal alerts without market enters and I was wondering if that should be the norm. I've read and reread the pdf file but there just isn't much info there to understand.

On all my charts it says "No SS signal" followed on the next line by the trend direction with sometimes an additional "untradable". I guess what I'm confused about is since I'm getting all these...

"untradable" means that the RSI is sort-of "retracing", so the EA will not trade in this condition.

If it says "up" or "down" only, then it should enter trade as soon as it gets a "strong signal" from the SuperSignal,
but it will wait for it to stabilize during SuperSignalShift bars before entering the trade.
That's probably why you don't see it trading immediately.

And this is also the reason for using shorter TF charts (M15 for me) rather than H1 charts: it is more reactive to the swings.


Sq
  • #306
  • Jun 10, 2010 11:47am Jun 10, 2010 11:47am
Quoting obnoxious2
Been running the SS bot for about a whole day. Seems to be relaly selective as to entering any trades which is good. One question though to you guys. I've been getting strong sell/buy signal alerts without market enters and I was wondering if that should be the norm. I've read and reread the pdf file but there just isn't much info there to understand.

On all my charts it says "No SS signal" followed on the next line by the trend direction with sometimes an additional "untradable". I guess what I'm confused about is since I'm getting all these...
The bot does not take a trade unless at the time of the signal it is consistent with the RSI.

Most signals are not consistent with RSI, hence no trades taken in those instances.

Don't worry about the comments on the screen. The bot is working.

The bot is very selective. If you were trading only the indy without the RSI filter, you would have many trades, but a good portion of them would go into significant -ve UPL and possibly not recover.
  • #307
  • Jun 10, 2010 11:49am Jun 10, 2010 11:49am
One of my forward demos is on M5. It was losing most trades until I changed the RSI to 3. Then it turned around...at least so far only winning trades since then.

This may only apply to the M5 timeframe. The M15 demo still is using the RSI 21 setting and while it takes few trades, they have all been winners.

I should add that I am using MPTM to close the trades on both M5 and M15.
  • #308
  • Jun 10, 2010 11:50am Jun 10, 2010 11:50am
Quoting squalou
"untradable" means that the RSI is sort-of "retracing", so the EA will not trade in this condition.

If it says "up" or "down" only, then it should enter trade as soon as it gets a "strong signal" from the SuperSignal.


Sq
Thanks for clarifying that up sq! On my EURJPY chart now, it is saying uptrend with storng sell. No position taken though. This is correct right?
And much thanks for the updated SS version. I shall test it out tonight.
  • #309
  • Jun 10, 2010 12:00pm Jun 10, 2010 12:00pm
Quoting squalou
Hi all,

here is a new version of the SS bot.
Excellent! Thank you Squalou.
  • #310
  • Jun 10, 2010 5:09pm Jun 10, 2010 5:09pm
Quoting squalou
Hi all,

here is a new version of the SS bot.

I reorganized the inputs so that most important inputs come first.

I added BE and jumpingStop stuff, with the contribution from peejaywk also (thanks!)

These are inspired (borrowed i should say...)from MPTM EA, of course!

Here are the new inputs:

BreakEvenPips=0; // when >0, will move SL to BE+BreakEvenProfit when price has reached BE+BreakEvenPips
BreakEvenProfit=5;...
sq, this is a simply fantastic contribution. Many, many thanks.

I cannot adopt your version as the one to go into the post 1 zip because:

 

  1. I have to understand what you have done. There is so much new code in your version that I do not.
  2. I have to be able to update the user guide. I cannot do this, because of point 1.

What I have done is include a link to your robot in post 1, along with the strongest endorsement I can make - if anyone thinks I can make this stronger still, then sing out.

Cheers sq. You are a star. I hope I have mentioned this before.

  • #311
  • Jun 10, 2010 6:19pm Jun 10, 2010 6:19pm
Come-on Steve...
I learned so much by studying your EAs, and this is how i got here!
I should be the one rewarding you!
Can i vouch for my voucher ????
(yes! it seems i could do it! )

Sq
  • #312
  • Jun 10, 2010 6:34pm Jun 10, 2010 6:34pm
Quoting squalou
Come-on Steve...
I learned so much by studying your EAs, and this is how i got here!
I should be the one rewarding you!
Can i vouch for my voucher ????
(yes! it seems i could do it! )

Sq
Hehe. I will stroke.......................

Re mql4, we all started from the same point. You have taken your skills to a much higher level than I have mine, and I applaud you for doing so.

Better stop now, before this thread drowns in a tide of reader-vomit.

  • #313
  • Jun 10, 2010 6:38pm Jun 10, 2010 6:38pm
while playing around with the #MTF_RSI looking for better settings on the RSI filter for the SS bot (which lead me to RSI(H1,period=2) as the best so far), i found some interesting new (at least for me) concept applied to RSI: why not apply the good(?) old MA crossover method to the RSI ?

And here it goes: using a crossover between RSI(H1,2) and RSI(H1,6) seems to give better (earlier) trend reversal signals...
and this was confirmed with backtesting with the SS EA...

i see this as if the slower RSI (6) playing the role of an "adaptive" reversal limit line, rather than the fixed 45~55 hard limits used in the EA's RSI trend detector.

Try the following template to get a feeling...

what do you think ?

Sq
Attached File(s)
File Type: tpl ss ea1.tpl   23 KB | 515 downloads
  • #314
  • Jun 11, 2010 1:12am Jun 11, 2010 1:12am
Quoting octomanoak
One of my forward demos is on M5. It was losing most trades until I changed the RSI to 3. Then it turned around...at least so far only winning trades since then.

This may only apply to the M5 timeframe. The M15 demo still is using the RSI 21 setting and while it takes few trades, they have all been winners.

I should add that I am using MPTM to close the trades on both M5 and M15.
Hi octomanoak,

so you only changed Rsi to 3 (M5 tf)? The other settings like dist1, dist2 and arrow distance default?

Best regards!
  • #315
  • Jun 11, 2010 8:54am Jun 11, 2010 8:54am
Quoting Meister81
Hi octomanoak,

so you only changed Rsi to 3 (M5 tf)? The other settings like dist1, dist2 and arrow distance default?

Best regards!
That's right. I'm using MPTM to close trades though using BE and Jumping stops.
  • #316
  • Jun 11, 2010 10:04am Jun 11, 2010 10:04am
Quoting octomanoak
One of my forward demos is on M5. It was losing most trades until I changed the RSI to 3. Then it turned around...at least so far only winning trades since then.

This may only apply to the M5 timeframe. The M15 demo still is using the RSI 21 setting and while it takes few trades, they have all been winners.

I should add that I am using MPTM to close the trades on both M5 and M15.
It is interesting as Forex.com it takes very few trades and almost all winners. IBFX it takes a ton of trades and isn't performing profitably...

Interesting how the feeds are so different that it changes the profitability so greatly w/ this bot.
We are our own best indicator.
  • #317
  • Jun 11, 2010 10:19am Jun 11, 2010 10:19am
Quoting bluemele
It is interesting as Forex.com it takes very few trades and almost all winners. IBFX it takes a ton of trades and isn't performing profitably...

Interesting how the feeds are so different that it changes the profitability so greatly w/ this bot.
Blue,
Which version and settings do you use? My prime broker is forex.com too.
  • #318
  • Jun 11, 2010 10:25am Jun 11, 2010 10:25am
Very interesting! I am also using forex.com. Are others finding the bot profitable on brokers other than forex.com or ibfx?
  • #319
  • Jun 12, 2010 2:26am Jun 12, 2010 2:26am
Quoting octomanoak
Very interesting! I am also using forex.com. Are others finding the bot profitable on brokers other than forex.com or ibfx?
I'm running it on IBFX however it has not taken many trades. The trades it has taken have not been good. But I am still messing around trying to find an optimal RSI filter to use depending on TF and the period #. Another factor that is hampering it is the # of candles to look back on.
  • #320
  • Jun 12, 2010 3:45am Jun 12, 2010 3:45am
we all have the same issue here: finding the correct settings for RsiTF, RsiPeriod, SuperSignalShift and SuperSignalDist2.

Backtesting and optimizer show that these settings are definitely dynamic, no do-it-all set is possible.

And they will also be different for different pairs...

My findings so far are that the most consistent results are achieved with a "5-days-optimizer-sliding-window", which requires to re-otimize every day based over the last 5 days, but this is by far giving the most consistent performance on the long run.

Backtesting with this kind of "dynamic optimization" is really a pain!

I don't think it is possible to run the optimizer from an EA directly.

Perhaps we could batch the optimizing task:
run the optimizer in a separate "terminal" (MT4 platform), save the optimized settings into a file,
and the EA actively trading on another MT4 terminal would just import the settings on a regular basis, to get the most recent optimized inputs and use that for the day.

If anyone knows how to do that...
Steve ? Scoobs ? RR ? 7bit ? ...
you are welcome to show up and tell us what you think... thanks guys!


Sq.
1 trader viewing now