Try My EA! Need Group Feedback... | Page 17 | Forex Factory

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...

  • #321
  • Dec 5, 2014 11:20am Dec 5, 2014 11:20am
Quoting cristian23
Hello, First, I want to thank to the creator of this EA. Really impressive! 50% last month with daily charts. {image}

Hi Cristian,

Can you tell me how you tested this EA? Didn't it have open orders when the testing period finished closing which prematurely most likely would result in a loss at the very end right? How did you manage to get over that?

@Pips - Don't worry, I am not trying to backtest. I am trying to identify number of times where we would have been ranged out. Perhaps we out to take very small profits to ensure that we never get swiped other than the worst of the worst scenarios.
  • #322
  • Dec 5, 2014 11:43am Dec 5, 2014 11:43am
Quoting kashTrader
I will simply change code where I see it will be helpful with execution of the idea.
Hi kashTrader,

If I may suggest and if Thepipster is ok with it, would you be able to add a function that will 'cut off' the bot once a set overall equity target is reached? Say, we have a daily target of 5%. A cycle closes => equity is up by min. 5% => the bot shuts itself down => we say 'Thank you, Mr Market'.
mind over matter
  • #323
  • Edited 12:14pm Dec 5, 2014 11:55am | Edited 12:14pm
Quoting kashTrader
{quote} Could you please explain with a difference in case of this EA. I fail to see a difference!
Right, at the end of my blabbering there, I said it doesn't really affect this EA the way I have it setup.

Quote
Perhaps we out to take very small profits to ensure that we never get swiped other than the worst of the worst scenarios.

ah, you're catching on! It is better to take quick, smaller profits, even if it means a little less money at the end of the month...then we can have money at the end of the year, vs. $0. That's why the lower spread of armada is so facinating. It has capability to profit much easier\faster, but requires more margin because more steps\trades are needed to avoid a "hedge box".
I think the margin issue isn't actually a problem, as long as the broker allows the used margin to be cancelled with a hedged order. Either the price will move far enough to close all trades, or opposite trades will be opened, lowering the used margin, freeing usable margin. EXAMPLE: Currently my explorer shows 10 sells and 9 buys, so the margin used is only $44, leaving the rest of the account equity avail for free margin.

~KOSMO...the dark side is slippage, IMO. I have tried a reversed idea of this strategy using only limit orders...would work wonderful right now, but when the market moves, instant margin call!!! It's a lot better to just wait, even if it's a week, on the market. The example I gave on my explorer where it closed trades too slow is the "darkest" thing I've seen. It does happen sometimes, but not in the majority...it gets outweighed by all the good times later.

Quote
If I may suggest and if Thepipster is ok with it, would you be able to add a function that will 'cut off' the bot once a set overall equity target is reached? Say, we have a daily target of 5%. A cycle closes => equity is up by min. 5% => the bot shuts itself down => we say 'Thank you, Mr Market'.

You can do that...I don't think I would use it. I'm stuck with Finfx, pretty much, so I have to play the long waiting game.

I still get weird fuzzies looking at your trade stats, under the trade report section.
Too close for pips, switching to points.
  • #324
  • Dec 5, 2014 12:03pm Dec 5, 2014 12:03pm
@ Pipster, how often in your testing did you get "hedge boxed"?
  • #325
  • Dec 5, 2014 12:17pm Dec 5, 2014 12:17pm
Quoting dambuster
@ Pipster, how often in your testing did you get "hedge boxed"?
Never happened after I used 30 steps per side. I have $30 window, though, because of spread. I had started with $10 "box" and it happened once then, but I increased it after finding that. As long as margin allows, this EA basically will eventually climb out of the hole with additional trades added in steps because of the effect of hedging on used margin, and the inevitability of the gold market to really move ($50-100).

I super recommend holding off of going live!
If kashtrader figures out how to move the opened orders to the broker to be closed via TP\SL, our trading will start looking closer to kosmo's unattainable utopia. I would wait at least until we see results from that.

I have posted this EA as a gift, but for feedback to me on other's accounts (which I have learned a ton from in just a few days THANKS), but also so that, when the idea was established and accepted, we can work on improving execution! We're not there yet, IMO. Unless armada suddenly throws it's doors to all traders worldwide and has no slippage, we have to figure out other ways.

~Kashtrader...
If you want another idea to throw into the complex coding mix, to solve the closing pending orders problem, could the EA be programmed to keep them open, except for ones that are not "valid", and then just have to place new ones in the "gaps"?
Too close for pips, switching to points.
  • #326
  • Dec 5, 2014 12:52pm Dec 5, 2014 12:52pm
Quoting Thepipster
{quote} Never happened after I used 30 steps per side. I have $30 window, though, because of spread. I had started with $10 "box" and it happened once then, but I increased it after finding that. As long as margin allows, this EA basically will eventually climb out of the hole with additional trades added in steps because of the effect of hedging on used margin, and the inevitability of the gold market to really move ($50-100).
So, in a box hedge scenario the EA will open additional trades?
  • #327
  • Dec 5, 2014 12:53pm Dec 5, 2014 12:53pm
Thanks for the EA to check out. I'll be gone a week, but I'm going to start it up from a fresh demo when I get back and can leave it running. In the account here, only the last 2 days or so are valid and purely the result of the EA as other trades are mixed in before.
A good man knows his limitations ~ Clint Eastwood
  • #328
  • Dec 5, 2014 1:02pm Dec 5, 2014 1:02pm
Quoting Thepipster
KOSMO...the dark side is slippage, IMO.... It's a lot better to just wait, even if it's a week, on the market.
I agree... if we only focused on selecting the times when gold is likely to move far and quick enough to outrange the box in one direction, even with some inevitable slippage, I guess the results might be close to mine of today, on live.

Quoting Thepipster
Kashtrader... If you want another idea to throw into the complex coding mix, to solve the closing pending orders problem, could the EA be programmed to keep them open, except for ones that are not "valid", and then just have to place new ones in the "gaps"?
I like the idea. Would definitely speed up restarting the cycle, especially for the 60 steps setting.
mind over matter
  • #329
  • Dec 5, 2014 1:08pm Dec 5, 2014 1:08pm
Attached Image (click to enlarge)
Click to Enlarge Name: armada.png Size: 73 KB

Attached Image (click to enlarge)
Click to Enlarge Name: Fin.png Size: 75 KB


here is screenshots from the spreads during NFP this morning...NFP was at 7:30 on the shots, but you can see the dots, and the spike on FinFX's account. That spike is $1.30, enough to cause a little problem, and different from slippage, but same result. I looked at several other brokers I am familier with, and they all had the same type of spike, some over $2. What gets me is Armada doesn't go over $0.10 the whole time. It's like there's a cap on Gold there, a flat line. This is a little hard to believe because liquidity providers do spike their spreads. If somebody goes live with armada and makes a million, without any problems, than they're great. I somehow just don't think it would turn out that way. I'm guessing they would accuse you of arb or something and boot you out. Looking at their spread history, they do have mild spikes every once in a great while, but NFP should be a given.

Thoughts?
Too close for pips, switching to points.
  • #330
  • Dec 5, 2014 1:43pm Dec 5, 2014 1:43pm
Quoting Thepipster
Thoughts?
Interesting... Armada is not my main broker. I've used them live on and off for over a year as a kind of experimental playground for various bots, BUT never on gold. From my experience, their live execution is good, with occasional slippage in the range of 0.00003-0.00005.
mind over matter
  • #331
  • Dec 5, 2014 1:59pm Dec 5, 2014 1:59pm
Quoting kosmo
{quote} Interesting... Armada is not my main broker. I've used them live on and off for over a year as a kind of experimental playground for various bots, BUT never on gold. From my experience, their live execution is good, with occasional slippage in the range of 0.00003-0.00005.
hmmmm...that sounds pretty reasonable...I'm curious what slippage would have been on gold this morning live, because that spread is just flat.

I'm going to post the copy of the today's trade report from my broker when it comes out after the market closes. Then we can see how long it took those trades to close.

Your results this morning was what my original intention was when I conceived this idea a long time ago...tons of really small orders with small profits, that add up from high volume. How else do you get $400 from 0.01 orders?
But I'll just have to live with physical limitations here in US.
Too close for pips, switching to points.
  • #332
  • Dec 5, 2014 2:24pm Dec 5, 2014 2:24pm
Quote Thepipster If somebody goes live with armada and makes a million, without any problems, than they're great. I somehow just don't think it would turn out that way. I'm guessing they would accuse you of arb or something and boot you out. Looking at their spread history, they do have mild spikes every once in a great while, but NFP should be a given.

Thoughts?
I am relatively new to Forex , so I am a very interested participant/observer of this embryonic EA trading system thread. I also know next to nothing about programs and code. But I would like to throw these thoughts into the ring and watch with interest other more experienced traders responses. If any !
In sports betting with a bookmaker ( for fx trading read broker) as soon as a punter develops a gambling system which becomes profitable, and looks like it might be repeatable, especially automated, the bookmaker either, if your lucky, severely limits the amount of stake allowed, thus limiting the profit potential of the system. Or more likely shuts you down. One way or another a market maker, whether a broker or liquidity provider will do what ever is necessary to protect or enhance their profits, legally or otherwise and worry about the consequences later. The only place where a sports betting system can flourish is on an 'open' exchange such as betfair, where they are only connecting 'handsoff' peer to peer punters and 'should' have little or no interest in the outcome. The temptation for any market maker/ facilitator (bookmaker and even Betfair etc) to profit from covert intervention, at an appropriate moment, in the operation of a profitable system, must be irresistable, human nature, especially 'business' human nature, being what it is, when money is involved going through their hands. So any 'mass' market strategy which can make a s**tload of money is going to be well above the radar for the vultures. unless it works at the core of the transaction process and cannot be kneecapped without compromising the fundamentals of the transaction process, in this case, NDD retail forex trading .
More thoughts ?
  • #333
  • Dec 5, 2014 2:48pm Dec 5, 2014 2:48pm
Pips,

The first thing I am trying to do now is setup the EA so that I can run back tests. Now, these backtests are not to see how profitable the EA is. But, to find out if and how many times we get boxed in. We need to be able to find a solution to the boxed in part somehow to be able to make this system work. Otherwise, One box in could take away all profits of a year.

So, to start off the process, I have modified the EA slightly so that it automatically closes all trades once there are no pending orders remaining. This is probably the worst case scenario given that the EA should close exactly when the boxin happens. Meaning the maximum possible drawdown.

Now firstly, I want to see if we do get boxed in. If we do then we need to figure out ways to get out of this. So, firstly to help with this experiment. Can you please provide me with the best possible settings for Armada at 100,000 equity. Spread on Armada is around 25.

Please specify the values for buffer, grid step and equity gain from the experience you have. I want to run it with that and see if we would have gotten boxed in. Once I have that we can figure out ways to minimise the problem, open additional trades on the side where the last pending order hit, open order in between gaps (although I didn't get that) and so on.... Lets start here and take baby steps to fixing this.
  • #334
  • Dec 5, 2014 2:58pm Dec 5, 2014 2:58pm
By the way the current version of the EA has a lot of bugs. One that I fixed is as follows:

When TP is hit, the EA starts closing all open trades....
It starts closing all pending trades....... However, while closing the pending trades of one of the pending trades hit and turns to an open trades that is no longer managed by the EA.

A worse problem is what we have been noticing. The current version of the EA cycles through all trades 3 times and looks for open trades to close. For some reason or other if trades fail to close (because of connections or broker) the EA gives up and no longer goes back to cleaning these trades up for closing.
  • #335
  • Dec 5, 2014 3:18pm Dec 5, 2014 3:18pm
Quoting kashTrader
Pips, The first thing I am trying to do now is setup the EA so that I can run back tests. Now, these backtests are not to see how profitable the EA is. But, to find out if and how many times we get boxed in. We need to be able to find a solution to the boxed in part somehow to be able to make this system work. Otherwise, One box in could take away all profits of a year. So, to start off the process, I have modified the EA slightly so that it automatically closes all trades once there are no pending orders remaining. This is probably the worst case...
You might be able to see if it "boxes in" backtesting, but it might not be able to really show you because the backtest won't be able to close the trades when they would close in forward trading. It's just the way it is. But, we can get around getting stuck by adding more pending orders on each side, continuing in steps, maybe 10 more? Also, we can figure out a way to close orders that we know are hedged and therefore inconsequential, too...this will speed up things. But we would have to watch out for the target not being affected.

hmmm...looks like we need improvement on making sure trades are closed, which could be helped with the idea of leaving them open for the next time, and just replacing the gaps...to do this would mean more of a fixed grid structure, instead of distances from current price.
Too close for pips, switching to points.
  • #336
  • Dec 5, 2014 3:27pm Dec 5, 2014 3:27pm
Hi Kashtrader.
I think my Armada Demo ES is locked up by the scenario you outlined earlier ( another non closed orphan). The open trades are not equidistant from the current price action on the chart. They are equidistant from a PA approx 4 or 5 hours ago and there is a trade deficit running consuming most of the days profit. Is it worth just switching it off and restarting it Sunday evening.
  • #337
  • Dec 5, 2014 3:34pm Dec 5, 2014 3:34pm
Quoting go4
Hi Kashtrader. I think my Armada Demo ES is locked up by the scenario you outlined earlier ( another non closed orphan). The open trades are not equidistant from the current price action on the chart. They are equidistant from a PA approx 4 or 5 hours ago and there is a trade deficit running consuming most of the days profit. Is it worth just switching it off and restarting it Sunday evening.

I am not sure how the current version deals with left over trades. Pipstar says it deletes them. I have now re-written the code to ensure that trades do close before moving on to the next round. Hopefully some testing and then release
  • #338
  • Dec 5, 2014 3:40pm Dec 5, 2014 3:40pm
Quoting Thepipster
{quote} You might be able to see if it "boxes in" backtesting, but it might not be able to really show you because the backtest won't be able to close the trades when they would close in forward trading.
I know there will be slipups and it wont be the same. But, atleast we will have an idea of the size of boxes that have formed in the past so that we optimise the settings for the forward testing. You said you haven't been hit since using 30 steps but I just got hit. I have now increased the size but am beginning to think that adding more orders would be the only way to go. Yes, closing out orders that cancel each other makes good sense. I will think about adding that to the EA now. Lets see what I can come up with. Won't happen before Monday because I need to be able to test it when I fix these. Have a good weekend. I will be posting results of the back testing soon.
  • #339
  • Dec 5, 2014 3:43pm Dec 5, 2014 3:43pm
Quoting go4
Hi Kashtrader. I think my Armada Demo ES is locked up by the scenario you outlined earlier ( another non closed orphan). The open trades are not equidistant from the current price action on the chart. They are equidistant from a PA approx 4 or 5 hours ago and there is a trade deficit running consuming most of the days profit. Is it worth just switching it off and restarting it Sunday evening.
Is the orphan trade still not able to be closed manually, like the last one?
If so, the platform or broker has some weird problem.

I think a closeable orphan trade could happen, like kashtrader said, if a pending order is opened while the opened orders are being closed, but the EA is supposed to go back through and close them all. It probably is worth posting an updated EA once we're sure of modifications.

~Kashtrader...your FXCM account has four sell trades at 1189.03-same price! that might have something to do with it.
My trades currently have $0.66 between them, so I have a $39 box. I would prefer 0.40 between them, and am contemplating taking the spread out of the math of the buffer and gridlevel.
Too close for pips, switching to points.
  • #340
  • Dec 5, 2014 3:52pm Dec 5, 2014 3:52pm
Quoting kashTrader
By the way the current version of the EA has a lot of bugs. One that I fixed is as follows: When TP is hit, the EA starts closing all open trades.... It starts closing all pending trades....... However, while closing the pending trades of one of the pending trades hit and turns to an open trades that is no longer managed by the EA. A worse problem is what we have been noticing. The current version of the EA cycles through all trades 3 times and looks for open trades to close. For some reason or other if trades fail to close (because of connections...

Currently the closing of all active and pending orders are handled by a single function. Maybe we can split that into two say, CloseAllActiveOrders() and CloseAllPendingOrders(). When TP is hit, we will first close all the active orders and then close all pending orders. Again we call the CloseAllActiveOrders() once more from inside the CloseAllPendingOrders() at the end which will close the active orders if any present.

just my 2 cents.
1 trader viewing now