Try My EA! Need Group Feedback... | Page 22 | Crypto Craft

Options

Go to
Permalinks
Print Thread

Related Instruments

Gold/USD

Similar Threads

USD/TRY and EUR/TRY 12 replies

Fibo Group Feedback 23 replies

Need feedback on my trading breakouts system 8 replies

One World Capital... Need feedback 2 replies

Need feedback on FXCM Educational workshops 3 replies

Attachments: Try My EA! Need Group Feedback...
Exit Attachments

Try My EA! Need Group Feedback...

  • #421
  • Dec 9, 2014 1:12pm Dec 9, 2014 1:12pm
http://www.kitco.com/images/live/nyg...07787682455684

Thanks for the suggestions, Kosmo. If you could help me with figuring out the unexpected closure problem by posting a statement when it's available from your broker, I can start working on that.

About the max possible drawdown, look at kashTrader's trade explorer http://www.forexfactory.com/kashtrader/58. It currently has all orders opened, no pending orders left. Now, the spread is smaller than average, but the floating loss is about 2.4% Up to this point, it can't lose anymore because all orders are hedged. It's true that if he were to close all trades, and start EA again it could get hung up, but my point was centered around the worst possible scenario in one instance of time. The 30% figure comes from the spread and gridlevel setting I use, which is average among most brokers, and assumes the user will not change settings. Having less spread is a limited advantage, and the drawdown is far less, but price can range it into a lockup. Gold has the breakout characteristics that make this EA work well with gridlevel gaps of 0.40-0.60 over time, or more numbers of orders with ~0.10 sized spread.

This part below will only be relevant for a few hours.
Here's a couple of explorers that are similar to my broker\settings. Note they have a start entry point close to mine.
http://www.forexfactory.com/mucky/77

http://www.forexfactory.com/spyderman/65

http://www.forexfactory.com/dambuster/20 -he increased his buffer and gridlevel, so gaps are about $0.50
Too close for pips, switching to points.
  • #422
  • Dec 9, 2014 1:34pm Dec 9, 2014 1:34pm
I registered VPS and setup the EA on that. I am newbie about vps i hope it would be fast and work without problem.
  • #423
  • Edited 2:29pm Dec 9, 2014 1:48pm | Edited 2:29pm
Quoting msdogan
I registered VPS and setup the EA on that. I am newbie about vps i hope it would be fast and work without problem.
Yay! good luck.
If you don't mind, make your lots size and profit public.
Too close for pips, switching to points.
  • #424
  • Dec 9, 2014 3:25pm Dec 9, 2014 3:25pm
Quoting Thepipster
{image} Thanks for the suggestions, Kosmo. If you could help me with figuring out the unexpected closure problem by posting a statement when it's available from your broker, I can start working on that.
Thanks for another clear explanation, Pipster. No more questions, for now.

Please find attached screenshots of yesterday's 9th/last cycle (the Ninth Circle of Dante's Inferno comes to mind, lol). I hope the answer is there to be found... Fingers crossed! Note that the last entry on each page is repeated as the first of the next.

Attached Image (click to enlarge)
Click to Enlarge Name: 0812a.png Size: 151 KB

Attached Image (click to enlarge)
Click to Enlarge Name: 0812b.png Size: 160 KB

Attached Image (click to enlarge)
Click to Enlarge Name: 0812c.png Size: 109 KB
mind over matter
  • #425
  • Dec 9, 2014 3:34pm Dec 9, 2014 3:34pm
Quoting Thepipster
{quote} Yay! good luck. If you don't mind, make your lots size and profit public.
It is public already
  • #426
  • Dec 9, 2014 4:12pm Dec 9, 2014 4:12pm
Thanks, kosmo!

wow...21 seconds to close trades 21:04:11-21:04:32
and...$0.43 between when lowest and highest buy close price, and $0.24 between lowest and highest sell close price!
Still lost $67 for the day before subtracting commission!

Buys had $6.49 spanning amount from lowest to highest orders, with 57 buys.
Sells had $3.74 spanning amount from lowest to highest orders, with 36 sells.
I just mention this to show the range of box of orders...

57 buys slipping $0.43 would mean about $24.51 lost,
36 sells slipping $0.24 would mean about $8.64 lost,
....multiplied with the 0.02 lots...
totaling an average of $66, very close to the actual $67!

My conclusion is these trades took too long to close...21 seconds. I'm guessing price flickered up high enough to hit the target, then retraced really fast.
We can see with the math that had all the opened trades been closed at the exact same time, there would have been virtually no loss.

I've had some time to view my own statement, which did very well after the first rough loss of the period. The price changed $0.20 while those trades were closing, moving against me in the 5 seconds it took. Oh, well...i got a positive balance today !

This definitely makes the case for trying to get TP and SL so the broker closes all orders at the same split-second, instead of waiting 20 seconds, since there are so many orders.

Let's keep testing!
Too close for pips, switching to points.
  • #427
  • Dec 10, 2014 9:48am Dec 10, 2014 9:48am
Hey everybody. I've been doing some research on how to implement having the EA place and modify TPs and SLs to hit target equity.

One idea I have would be almost a total re-write of the code. The only things that would be the same is the lot size being calculated according to account balance.

Basically, the EA would start with just 2 buys and 2 sell pending orders with TP already placed to hit a settable target. For this example, we will use 0.1%, so the target has the spread calculated so that we net 0.1% profit. We would want as close of a pip distance as spread allows possible. Lets say for this example the gap between orders and buffer to each side is $0.50. This means in order to profit and close trades, the TP would be set $1.25 from the first level's order's price. Continuing with the example. Lets say the market moves a little up and opens the first buy order. Immediately, the EA would cancel\delete the closest sell pending order, place one more sell pending order at the 3rd level away from start price, and modify the TP of both sell pending orders to $2.75 from the closest sell pending order (remember it used to be the 2nd closest.), and place a SL at the price of the cancelled sell pending order for the opened buy order. It would have to continue calculating TP and SL, cancelling appropriate opposite orders when a pending order is opened, placing new pending orders, etc.

In my opinion, this looks like way too much trouble! So, I have a better idea...

I think since the focus of the goal is to close the opened orders via TP and SL to have instant capture of the target, without such massive latency slippage, the current EA can just be modified to calculate mathematically the appropriate TP and SL and modify all orders, pending and open, as new orders are opened. Simplicity is better. I don't think we have to worry about closing hedged orders early, since they will also have SL\TP. That will simplify calculating, because the EA still has to calculate lot size, and target based on what it can see with floating equity, account balance, etc. While it would be nice to have the EA start with a few pending orders and add new ones to grow the "grid" if needed, it's not completely necessary, as long as the EA closes all pending orders with no orphans before starting the next new cycle.

So, unless someone wants to tackle the first idea, please give consideration to my simplified idea.
Too close for pips, switching to points.
  • #428
  • Dec 10, 2014 10:08am Dec 10, 2014 10:08am
I think we must test the first EA for 1 or 2 months. After that we can say different suggestions. It is so new we must be patient.
  • #429
  • Dec 10, 2014 10:31am Dec 10, 2014 10:31am
Quoting Thepipster
Thanks, kosmo! wow...21 seconds to close trades 21:04:11-21:04:32 and...$0.43 between when lowest and highest buy close price, and $0.24 between lowest and highest sell close price! Still lost $67 for the day before subtracting commission! Buys had $6.49 spanning amount from lowest to highest orders, with 57 buys. Sells had $3.74 spanning amount from lowest to highest orders, with 36 sells. I just mention this to show the range of box of orders... 57 buys slipping $0.43 would mean about $24.51 lost, 36 sells slipping $0.24 would mean about $8.64...
Thanks for the analysis, Pipster. It's hard to argue with the math, and your conclusion makes sense in this particular case. BUT, it happened again today! The terminal was losing connection time after time on my VPS, so I restarted it, and that's when the EA started closing the cycle. There was no logical reason for it; no spikes that could get the price anywhere near the target; the floating P/L was well in the negative. Please see the attached screenshots.

This is getting really annoying. There's no way to get reliable test results with an EA that refuses to work as it should at its very basic level.

Attached Image (click to enlarge)
Click to Enlarge Name: 1012a.png Size: 159 KB

Attached Image (click to enlarge)
Click to Enlarge Name: 1012b.png Size: 157 KB

Attached Image (click to enlarge)
Click to Enlarge Name: 1012c.png Size: 81 KB
mind over matter
  • #430
  • Dec 10, 2014 11:03am Dec 10, 2014 11:03am
Quoting kosmo
{quote} Thanks for the analysis, Pipster. It's hard to argue with the math, and your conclusion makes sense in this particular case. BUT, it happened again today! The terminal was losing connection time after time on my VPS, so I restarted it, and that's when the EA started closing the cycle. There was no logical reason for it; no spikes that could get the price anywhere near the target; the floating P/L was well in the negative. Please see the attached screenshots. This is getting really annoying. There's no way to get reliable test results with...

If you're having connection problems with the VPS, that can cause the EA to lose track of the profit target. Then, when you restart it, it sees the target as 0.2% above current equity, which is actually a loss.

I was watching your explorer a few hours ago, noting it only had sell pending orders left, all buys were open. I could see it had successfully closed a basket of trades earlier today, and didn't see anything unusual up to the point it closed everything. It had a floating loss of 200£ the last time I looked at it before you posted. If your VPS is acting up, it will be a threat to all EAs, because something could happen after it disconnects, making you lose lots of money while you're not able to catch it.
Too close for pips, switching to points.
  • #431
  • Dec 10, 2014 11:17am Dec 10, 2014 11:17am
http://www.kitco.com/images/live/nyg...07787682455684

I edited the first post to include this, but anybody must be aware if MT4 is closed for any reason while EA is active, it will not know the original starting equity, but instead start with whatever the equity shows on MT4 startup, which will most likely net you a loss, since the target will be actually negative. I'm sorry, but this is a MT4 limitation, there's no way around it. Make sure your computer or VPS is reliable.
Too close for pips, switching to points.
  • #432
  • Dec 10, 2014 11:37am Dec 10, 2014 11:37am
Quoting Thepipster
{quote} If you're having connection problems with the VPS, that can cause the EA to lose track of the profit target. Then, when you restart it, it sees the target as 0.2% above current equity, which is actually a loss. I was watching your explorer a few hours ago, noting it only had sell pending orders left, all buys were open. I could see it had successfully closed a basket of trades earlier today, and didn't see anything unusual up to the point it closed everything. It had a floating loss of 200£ the last time I looked at it before you posted....
I see, thanks Pipster. So, it's actually my fault this time. Before I restarted the terminal, I had tried to recover the connection with the 'rescan servers' option a few times (which worked for a few seconds each time), and it didn't affect the EA then. So, I presume it was only the actual restarting of the terminal that spoilt the party. Oh well, we learn something new every day...
mind over matter
  • #433
  • Dec 10, 2014 11:43am Dec 10, 2014 11:43am
Quoting Thepipster
So, I have a better idea... I think since the focus of the goal is to close the opened orders via TP and SL to have instant capture of the target, without such massive latency slippage, the current EA can just be modified to calculate mathematically the appropriate TP and SL and modify all orders, pending and open, as new orders are opened. Simplicity is better. I don't think we have to worry about closing hedged orders early, since they will also have SL\TP. That will simplify calculating, because the EA still has to calculate lot size, and target...
I think we need to be consistent in our initial goal to make the best of the EA as it is now, first. It has a lot of potential, and the instant SL/TP function will surely make it even better. Hopefully, someone can implement it...
mind over matter
  • #434
  • Dec 10, 2014 12:30pm Dec 10, 2014 12:30pm
Quoting kosmo
{quote} I think we need to be consistent in our initial goal to make the best of the EA as it is now, first. It has a lot of potential, and the instant SL/TP function will surely make it even better. Hopefully, someone can implement it...
I agree...I want people to be able to run it for a few weeks on current settings to allow everybody to see the robustness of the EA. We will just have to ignore accidents in judging the results. We do want to know if there are issues, of course. I've learned a lot just by posting this thing and seeing how it's working for others. Give a few months, and this thread will be catching servers on fire.
Too close for pips, switching to points.
  • #435
  • Edited 6:19am Dec 11, 2014 6:07am | Edited 6:19am
just a proposition, although i'm not running the EA.

I was writing EAs and considered the fact that if u have long running EAs and your profit is based on balance, it may hit profit calculation issue when u restart it. or it messes up when u run it for multiple counters.

You can actually do this instead, write the current account balance or equity into a flat file using the functions available, upon the start of a new trading cycle. When the EA starts, in init() it loads the balance/equity written in the file into a variable, rather than what's on the account. If there is no file, it will default take what's on account of course.

This will allow u to run the EA on multiple counters, even if the profit is based on balance/equity. Just specify the OrderSymbol as part of the file name.
  • #436
  • Dec 11, 2014 7:53am Dec 11, 2014 7:53am
We are ok for testing with current method of calculating equity. The only significant change that will benefit us right now is being able to calculate the TP\SL that would correspond with the equity target.

News this morning is all pro-USD, so we'll see if gold makes an extended move down...
Too close for pips, switching to points.
  • #437
  • Dec 11, 2014 8:04am Dec 11, 2014 8:04am
Hi Thepipster,

my first demo test got screwed up when updates were added to my laptop and restarted it. I re-created a new explorer with a new demo account. I don't have a vpn account so this will have to do, and yes I changed my win updates settings to prevent this from happening.

Sorry about that, it's a new beginning for that test.
  • #438
  • Dec 11, 2014 8:07am Dec 11, 2014 8:07am
Quoting jpdlegault
Hi Thepipster, my first demo test got screwed up when updates were added to my laptop and restarted it. I re-created a new explorer with a new demo account. I don't have a vpn account so this will have to do, and yes I changed my win updates settings to prevent this from happening. Sorry about that, it's a new beginning for that test.
Cool beans...(how many people use that saying? I ought to poll it)

I made sure I turned off windows updates when I launched my VPS, and recommend everybody do the same.

YAY! gold went down enough to close my trades.

~msdogan...lookes like your EA is doing good! You do still have your lots and account locked within your explorer. You don't have to show these, but it helps for evaluating results. You can change the settings within explorer to public for those categories.
Too close for pips, switching to points.
  • #439
  • Dec 11, 2014 8:19am Dec 11, 2014 8:19am
@ Pipster, re your recent PM: between 2 and 4 seconds to close 30ish trades, faster if fewer and a bit longer if more (VPS with ~3ms ping)
  • #440
  • Dec 11, 2014 8:20am Dec 11, 2014 8:20am
Quoting Thepipster
{quote} Cool beans...(how many people use that saying? I ought to poll it) I made sure I turned off windows updates when I launched my VPS, and recommend everybody do the same. YAY! gold went down enough to close my trades. ~msdogan...lookes like your EA is doing good! You do still have your lots and account locked within your explorer. You don't have to show these, but it helps for evaluating results. You can change the settings within explorer to public for those categories.
I havent turned lots to closed. Interesting when i look my explorer it says it opened to public
0 traders viewing now