Vegas Algorithm auto-trader by Steve and Squalou | Page 12 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Algorithm for Belkhayate Timing and Barycenter 0 replies

Anyone wants to program this profitable algorithm into an EA? 7 replies

what type of an algorithm is this? 0 replies

Question for VEGAS and others using Vegas system..... 2 replies

Attachments: Vegas Algorithm auto-trader by Steve and Squalou
Exit Attachments

Vegas Algorithm auto-trader by Steve and Squalou

  • #221
  • Dec 7, 2010 2:03pm Dec 7, 2010 2:03pm
Quoting SteveHopwood
Ok, I have added required the code - you can see how simple it is to do so. Post 1.

You might like to consider this: as coded, the bot can only look at the previous candle to find a trade, at the start of a new candle. There are a couple of disadvantages to this approach:
[list=1][*]Once a candle after the crossing candle has closed, the possibility of a trade has disappeared.[*]With the new only trade above/below the crossing candle's hi-lo, you could easily miss the optimum time to enter the trade because the bot can only trade at the start...
As an honest-to-goodnes non-coder, this is very confusing. Since the crossing candle's h/l is established prior to the next candle opening, can't that level just become the threshold for entry as soon as price moves above/below it?
Geez, I must sound like one of the children you teach piano to.
Mike
  • #222
  • Dec 7, 2010 2:11pm Dec 7, 2010 2:11pm
Quoting Pinhead
As an honest-to-goodnes non-coder, this is very confusing. Since the crossing candle's h/l is established prior to the next candle opening, can't that level just become the threshold for entry as soon as price moves above/below it?
Geez, I must sound like one of the children you teach piano to.
Mike
It can, but you still have to decide for how long the hi/lo remains relevant. An hour? 2 hours? 4 hours? A day? A week? A month? Indefinitely? You are asking for code to emulate things that you do manually and naturally.

There is nothing natural about computer code. A robot runs along the lines it is coded to run along. That calls for decisions on the part of the system developer - you. There is nothing I can do in the absence of such decisions.

  • #223
  • Dec 7, 2010 2:16pm Dec 7, 2010 2:16pm
Quoting SteveHopwood
It can, but you still have to decide for how long the hi/lo remains relevant. An hour? 2 hours? 4 hours? A day? A week? A month? Indefinitely? You are asking for code to emulate things that you do manually and naturally.

There is nothing natural about computer code. A robot runs along the lines it is coded to run along. That calls for decisions on the part of the system developer - you. There is nothing I can do in the absence of such decisions.

OK. I would say the level remains in effect unless a subsequent candle closes on the other side of Alf. Would that work?
  • #224
  • Dec 7, 2010 4:39pm Dec 7, 2010 4:39pm
Quoting Pinhead
OK. I would say the level remains in effect unless a subsequent candle closes on the other side of Alf. Would that work?
Yes and no.

Yes, it is easy to set the hi-lo variables that represent those of the crossing candle, but these get lost as soon as the platform is closed down for any reason.

It is possible to save these variables in files stored on the hard drive, to be read whenever the bot is run following a platform restart. The trouble here is this: if you have a file that is read by the bot on startup, then the bot will read the file on startup regardless of how long ago this file was written. So, you shut the bot down for a fortnight to go on holiday, or a day to avoid some nasty new events. On restart, the bot could be working with ridiculously out of date information.

Within reason, you can have what you want. Be careful about what you ask for, in case what you get is not quite what you really want.

Bearing in mind that I do not trade, and have no intention of trading this strategy, have not kept up with developments here and do not fully understand what you are trying to do, here is my advice. Either 1) Accept that trades not triggered during the candle following the crossing candle are lost or; 2) have a look-back-x-candles for a cross input.

You would still need to decide how far back in time to look.

Also, if you want to go the exceed-the-crossing-candle-hi/lo before a trade route, you have to jettison the only-trade-at-the-start-of-the-new-candle principle, or you are going to miss most of your trades.

  • #225
  • Dec 7, 2010 4:40pm Dec 7, 2010 4:40pm
Quoting SteveHopwood

So, you might care to consider these two options:

 

  1. Add a look-back-x-candles to find the crossing candle.
  2. Abandon the only-trade-at-the-start-of-a-new-candle requirement.

Steve/Mike, I just caught up with this thread again and would think that option 2 would work along with Mike's suggestion that the hi/lo of the candle crossing and closing above/below the ALF would remain in effect indefinitely until another candle crosses and closes back to the other side of ALF. I am not a coder, but would this work? The EA could set a pending buy/sell stop above/below the hi/lo, plus a pip buffer (user input), of the candle that crossed ALF, either to close an open trade going against you, and then open a new trade in that direction if RSI, or another indi, supports the new trade. I think you would want to add the spread to the pip buffer as well when setting pending long trade entries. What I don't know is whether or not this strategy would fail because of the min pip distances brokers have for setting SL's or opening trades, eg invalid stops. Hope this helps. http://cdn.forexfactory.com/images/icons/icon5.gif

Pete
  • #226
  • Dec 7, 2010 5:46pm Dec 7, 2010 5:46pm
Quoting SteveHopwood
Yes and no.

Yes, it is easy to set the hi-lo variables that represent those of the crossing candle, but these get lost as soon as the platform is closed down for any reason.

It is possible to save these variables in files stored on the hard drive, to be read whenever the bot is run following a platform restart. The trouble here is this: if you have a file that is read by the bot on startup, then the bot will read the file on startup regardless of how long ago this file was written. So, you shut the bot down for a fortnight to go on holiday,...
Losing the variable after a shutdown is not a worry. If the EA is trading round-the-clock, yes, but it's not really meant to be used that way. It's a session-specific system, so starts and stops are normal.

My goal with this is to avoid a head fake across Alf. I'm looking for confirmation by a successive candle that price just may continue in the same direction. I've seen all too often that a candle will test the waters on the other side of Alf only to watch the next candle bump up against the new hi/lo and then retreat. With my money.

One of two things will always happen: a subsequent candle will either exceed the hi/lo and trigger a trade or one will retreat to Alf's opposite side and close. It may take one candle or twelve, but it normally happens within 3 or 4, especially during high-volume hours. So . . . . it sounds like the best option is a look-back-x-candles, especially if it was a user-adjustable input.


If my logic is sound, that would be my request.

Thanks
Mike
  • #227
  • Dec 7, 2010 6:45pm Dec 7, 2010 6:45pm
Quoting Pinhead
Losing the variable after a shutdown is not a worry. If the EA is trading round-the-clock, yes, but it's not really meant to be used that way. It's a session-specific system, so starts and stops are normal.

My goal with this is to avoid a head fake across Alf. I'm looking for confirmation by a successive candle that price just may continue in the same direction. I've seen all too often that a candle will test the waters on the other side of Alf only to watch the next candle bump up against the new hi/lo and then retreat. With my money....
Done. Post 1 as usual. The input is LookBackForCross and defaults to 3.

I have just run 3 'Open price only' EU backtests in strategy tester, over the last year (Alpari UK) H1 charts trading 7.00am to 5.00 pm here in the UK. 0.1 lots on a £10,000 account

LookBackForCross = 1: profit just over 2%
LookBackForCross = 3: profit just over 7%

Then I ran an optimisations of the tp, sl and lookback inputs. The optimum setting for LookBackForCross came out at 19, with tp 80 and sl of 70.

Profit: just a tad over 30%

Ok Mike, you have my attention now.

  • #228
  • Dec 7, 2010 6:57pm Dec 7, 2010 6:57pm
Quoting SteveHopwood
Done. Post 1 as usual. The input is LookBackForCross and defaults to 3.

I have just run 3 'Open price only' EU backtests in strategy tester, over the last year (Alpari UK) H1 charts trading 7.00am to 5.00 pm here in the UK. 0.1 lots on a £10,000 account

LookBackForCross = 1: profit just over 2%
LookBackForCross = 3: profit just over 7%

Then I ran an optimisations of the tp, sl and lookback inputs. The optimum setting for LookBackForCrosscame out at 19, with tp 80 and sl of 70.

Profit: just a tad over 30%

Ok Mike, you have...
19?

Steve, I'm not good at backtesting. In fact, I'm not too good at any of this.
But, my eyeball tests say that 15 or 30M TFs are better, with much tighter stops. Have you tried shorter TFs? And, are you running it OOTB?
  • #229
  • Dec 8, 2010 5:56pm Dec 8, 2010 5:56pm
OK guys. Feeling strong? You need to be. Here is a pm to me from Mike earlier, and my reply. You need to read it properly to understand what this post is about:

Quote:
Originally Posted by Pinhead
Hi Steve.
Although strategy tester is still a bit of a mystery to me, I spent the better part of yesterday running various PH EA setups on it. Haven't got a clue how to use optimization, so I just kept plugging in different variables and running tests. Testing period was 1/7/10 thru today. With some settings I think you might find strange, I was able to get a return of over $58,000 on a $10,000 deposit. I didn't want to post anything in the thread about it until you could give me your take. Since I can't attach my set file, here's what I used:

Lots: 2
TP: 200
SL: 200
Alf cross: OOTB
RSI inputs: 240, 1 period, other settings OOTB
Use MA: false
Trading hours: 8:00 to 18:00 GMT
BE setting: BE, true; BE pips, 15; BE profit, 30
All other settings OOTB

Mike

My reply to Mike:

I have only just discovered optimisation, so I am not sure I am doing it correctly. It is great fun to play with and a bit of an eye-opener.

So, to use optimisation, call up strategy tester as usual. Then:

  1. tick the Use Date box and select a sensible time period for backtesting and optimisation - the last year seems sane to me. Even if we can backtest 30 years, it is really only recent times that are of interest to us.
  2. tick the optimisation box then Expert Properties, and select the Inputs tab
  3. each input has a tick box to the left of it. Optimising your ea, I selected TakeProfit, StopLoss and LookBackForCross
  4. the next thing to do is select the range you want optimising. To do this, you put values in the Start, Step and Stop columns for the inputs you have selected for optimising.
  5. in the TP and SL cols, I put 50, 10, 200 in the respective Start, Step and Stop cols. In the LookBackForCross input cols, I put 1, 1, 10

Click the OK button and you are ready for the fun bit. Make sure you have 'Open prices only' and click Start.

A couple of extra tabs appear - Optimisation Results and Optimisation Graph. It is the first one that is of interest to us.

Once optimisation is complete (not long for our robot), select the Optimisation Results tab. Now, here is the really clever bit:

  1. click the Profit heading in the Profit column to call up the maximum profit made by the most profitable optimisation pass. You might have to do this twice, and watch the slider bar at the right to make sure you really do have the most profitable at the top of the column
  2. now double click the parameters in the Inputs col - the one that begins with TakeProfit=
  3. go to the Settings tab and call up the Expert Properties and you will see the optimised parameters inserted into the inputs

Now run a normal backtest (without Optimisation selected) and drool over the profit increase compared with pre-optimisation.

Rather as you have done but with the process cut much shorter. Now you will understand why you have my attention.

So, to run the pinhead bot on a number of pairs, set up the demo and do an optimised backtest on each pair to be traded. Save the settings in set files for loading into each instance of the bot.

Even better, develop your coding skills so you can enter the optimised values into the robot's code. Some time this week,I will post a version where this code is included so you can adapt it for yourself.

I hope all this helps. Just remember, I have only just found optimisation and am not sure I am doing it correctly.

Cheers
_____

The latest post 1 update includes a new input - UseOptimisedValues. Only set this to true if you have done your own optimisation and can edit the code to include your optimised values. If you are both incapable of doing this and unwilling to learn how (and it ain't hard to learn - I am a professional pianist not a professional coder) then forget this input and leave the bot alone.

To edit the optimised settings in code, load the code into the editor and go to int init() (just scroll down a few lines to find it). You will see the first line of optimised code and can edit the values in light of your optimisation.

To add new pairs, simply copy/paste an existing one and alter the pair and associated values to those of the optimised backtest and press the F5 key to recompile the source code. You do not need to worry about pair suffixes as the code takes care of this; ditto x digit criminals.

I suggest repeating this optimisation at the start of each month.

Those of you unable to do source code editing but are willing to learn, follow the link in my sig to the mptm page and download the 'How to edit a script code' pdf. The lessons you learn there are adaptable to any available mql4 source code.

Have fun.

  • #230
  • Dec 8, 2010 6:12pm Dec 8, 2010 6:12pm
Not quite sure what the benefit is to edit the source code and put your optimized settings into the ea directly.
Seems much easier just to save the set file after your backtest run, then drop your ea on your forward test charts and use this set file.
You can save a set file via the backtester just the same as you do normally (only difference is it saves your optimization start/step/end in there as well, which is ignored if you aren't backtesting).
  • #231
  • Dec 8, 2010 6:25pm Dec 8, 2010 6:25pm
Quoting tazmaan
Not quite sure what the benefit is to edit the source code and put your optimized settings into the ea directly.
Seems much easier just to save the set file after your backtest run, then drop your ea on your forward test charts and use this set file.
You can save a set file via the backtester just the same as you do normally (only difference is it saves your optimization start/step/end in there as well, which is ignored if you aren't backtesting).
It comes in handy when you want to do little things like increase the lot size as the account grows. You can do this once then not worry about updating umpteen individual charts.

I suppose the same thing applies to the umpteen individual charts when you simply update the code in one bot then recompile.

It is up to individuals to decide which method is most productive for them. I know which it is, but then I am a coder.

  • #232
  • Dec 8, 2010 6:27pm Dec 8, 2010 6:27pm
Yeah I guess I missing the difference, which is fine
If you are optimizing each pair, and coming up with different settings for each, it's just a matter of loading the set file for each one.
Ah well, I'll butt out
  • #233
  • Dec 8, 2010 8:39pm Dec 8, 2010 8:39pm
Optimization is really just the computer trading by hindsight. It can run every possible combination of variables until it uncovers the most profitable setup, BUT, it only applies to the test period. So does it really help us with future price movement? It seems to me like it wouldn't.

The IBFX optimizer gave me a setup that works like a charm as long as I run it on the IBFX demo. If I run it on MIG, the 4th trade causes a margin call. On Alpari, it happens on the 6th trade. If the slight variations between brokers can cause such havoc, can we expect any better luck with entirely new data that the next 12 months will produce?
  • #234
  • Dec 9, 2010 5:14pm Dec 9, 2010 5:14pm
Quoting Pinhead
Losing the variable after a shutdown is not a worry. If the EA is trading round-the-clock, yes, but it's not really meant to be used that way. It's a session-specific system, so starts and stops are normal.

My goal with this is to avoid a head fake across Alf. I'm looking for confirmation by a successive candle that price just may continue in the same direction. I've seen all too often that a candle will test the waters on the other side of Alf only to watch the next candle bump up against the new hi/lo and then retreat. With my money....
Mike, Steve: Geez, I am no coder at all and feel somewhat stupid to offer this suggestion, http://cdn.forexfactory.com/images/icons/icon11.gifbut the look back option seems complicated compared to this: If a candle crosses AND closes above/below the ALF, have the EA at the beginning of the next candle immediately set a pending buy/sell stop trade above/below the hi/lo(plus some added pips for a buffer as a user input) of that crossing candle. This pending trade remains in effect until it is either triggered, or is canceled by the EA if price changes direction and closes on the other side of the ALF, or if the EA is shut down for any reason.

Does that make any sense, or am I all washed up? http://cdn.forexfactory.com/images/icons/icon9.gifAn answer here would help me understand the required logic for coding.

Trying to contribute.

Pete
  • #235
  • Dec 9, 2010 6:11pm Dec 9, 2010 6:11pm
Quoting ForexPete
Mike, Steve: Geez, I am no coder at all and feel somewhat stupid to offer this suggestion, http://cdn.forexfactory.com/images/icons/icon11.gifbut the look back option seems complicated compared to this: If a candle crosses AND closes above/below the ALF, have the EA at the beginning of the next candle immediately set a pending buy/sell stop trade above/below the hi/lo(plus some added pips for a buffer as a user input) of that crossing candle. This pending trade remains in effect until it is either triggered, or is canceled by the EA if...

Hey, Pete--Steve'll have to take that one.
I'm not a coder either, and for me, comprehending some of that stuff ain't easy. My only consolation is that Steve was in the same boat at one time.

FWIW, I think your solution is very good.

Mike
  • #236
  • Dec 10, 2010 10:10pm Dec 10, 2010 10:10pm
Quoting ForexPete
Mike, Steve: Geez, I am no coder at all and feel somewhat stupid to offer this suggestion, http://cdn.forexfactory.com/images/icons/icon11.gifbut the look back option seems complicated compared to this: If a candle crosses AND closes above/below the ALF, have the EA at the beginning of the next candle immediately set a pending buy/sell stop trade above/below the hi/lo(plus some added pips for a buffer as a user input) of that crossing candle....
Steve: Would this option work? Especially if it was done internally by the EA until one of actions occur that would either trigger, or cancel the trade?

Pete
  • #237
  • Dec 15, 2010 4:23am Dec 15, 2010 4:23am
This thread's not completely dead, although it is starting to smell a little. Actually, I've been trying to sort out trending issues. RSI wasn't really cutting it, imo. A little too vague at the moment of truth--the actual cross of the 50 level. Looks great after the fact, but I think something more concrete is needed in real time.

Here's an alternative: Multi-timeframe Alfs--Red is 1440 min, White is 240 min & Blue is current (0) Alf. You could argue that it's just 3 moving averages, but Alfs act a lot different and don't spend nearly as much time tangled up like MAs like to do--I think. I'm not convinced all 3 are necessary, but I haven't figured out which one to get rid of yet. The arrows are a MTF RSI indi, set to 240 (2). I'm not crazy about how they sometimes flicker on and off while making up their minds, but they give a heads up on change of direction every now and then, so I left 'em in.

Manually trading this would be pretty easy, but not sure how to string together the 'ifs and thens' for an EA yet.

Just thought I'd put this out there and maybe get some feedback from anyone who might still be checking in here. Plus, I'm only about 5500 posts behind Hopwood, and I'm trying to catch him by the end of the year.

Mike
Attached Image (click to enlarge)
Click to Enlarge Name: mtf alf.gif Size: 21 KB
Attached File(s)
File Type: mq4 RSI-8(50)Arrows.mq4   7 KB | 371 downloads
File Type: ex4 ALF-MTF.ex4   3 KB | 308 downloads
File Type: tpl triple alf.tpl   66 KB | 320 downloads
File Type: mq4 Sessions.mq4   8 KB | 330 downloads
  • #238
  • Dec 15, 2010 4:43am Dec 15, 2010 4:43am
[quote=Pinhead;4244712]This thread's not completely dead, although it is starting to smell a little. Actually, I've been trying to sort out trending issues. RSI wasn't really cutting it, imo.

Pinhead what trending issues are you trying to solve?

rgds

Peter
  • #239
  • Dec 15, 2010 4:53am Dec 15, 2010 4:53am
[quote=peterke;4244759]
Quoting Pinhead
This thread's not completely dead, although it is starting to smell a little. Actually, I've been trying to sort out trending issues. RSI wasn't really cutting it, imo.

Pinhead what trending issues are you trying to solve?

rgds

Peter
I really didn't like using an EMA to define the trend, what with it going north while Alf's going south, and vice versa so much of the time, so that left me with just RSI. If you watch RSI live, it crosses and retreats from the 50 level sometimes several times, which can't be a very good thing. I wanted something definite and stable.

Mike
  • #240
  • Dec 15, 2010 4:57am Dec 15, 2010 4:57am
hey bros,

does someone have a ''close basket'' or close ''all positions'' ea?

would be awesome

ty!
Hero calls followed by margin calls...
0 traders viewing now