My "Ultron" EA for GBPUSD H1 timeframe | Page 22 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Broker Challenge "NDD/STP" vs. True "ECN" and "Mkt Mkr" broker talk 46 replies

Dealing with "Invalid Price", "Requote", "Server Busy" in MT4 3 replies

Timeframe inside current timeframe - is that possible? 2 replies

Seeing higher timeframe indicators on lower timeframe 4 replies

Finding the best "personal" timeframe! 6 replies

Attachments: My "Ultron" EA for GBPUSD H1 timeframe
Exit Attachments

My "Ultron" EA for GBPUSD H1 timeframe

  • #421
  • Nov 20, 2018 1:22pm Nov 20, 2018 1:22pm
I've put a manual trailing stop in. I ibelieve this will hit to but the risk manager in me like having insurance
  • #422
  • Nov 20, 2018 2:05pm Nov 20, 2018 2:05pm
Quoting robots4me
{quote} The reason why it does best in 2018 is because the settings you are using were calibrated using historical data for 2018. Come on -- this issue has already been covered in recent posts.
Listen I've followed some of your posts in this thread and you've made some good points, but don't get arrogant or cocky, I was giving my observations of the 5 year back test, and we can all see the results. Now chill and have a beer mate.
Truth is like poetry. And most people f*cking hate poetry.
  • #423
  • Edited 3:25pm Nov 20, 2018 2:59pm | Edited 3:25pm
Quoting mrdfx
Now chill and have a beer mate.
Okay -- fair enough. But your post sounded like many others here (and elsewhere) who are surprised the EA doesn't perform well back to the beginning of time.

Your words -- "I'm no coder, but if we can try to optimize some of the parameters and improve the accuracy rate I'm sure it will increase the profitability of the EA making it even better than it already is."

So -- it sounded like you hadn't yet grasped that the EA is already calibrated for 2018 -- explaining why your back test shows the EA performs best in 2018. This has nothing to do with coding -- the code is fine. It's a simple matter of calibrating the settings using the most recent historical data.

If you get it then, fine, -- I apologize. It just didn't sound like it to me.

As for the beer -- GREAT idea. It's only noon, but I'm headed to the refrigerator now...
  • #424
  • Nov 20, 2018 3:25pm Nov 20, 2018 3:25pm
Quoting mrdfx
{quote} Listen I've followed some of your posts in this thread and you've made some good points, but don't get arrogant or cocky, I was giving my observations of the 5 year back test, and we can all see the results. Now chill and have a beer mate.
Fine..you talk me into it....Getting a beer also..been into Belgium lately..and here is the one for mrdfx from Sao Paulo..enjoy..

Going to see this guy on Friday

Inserted Video
  • #425
  • Nov 20, 2018 5:42pm Nov 20, 2018 5:42pm
Hello guys, sounds a litle bit strange , but thats we are here , to try to discovery to test to explore.
my point is , have you tried to run this on different pars ???
the altor did recomend only on gbpusd but .....
  • #426
  • Nov 20, 2018 10:13pm Nov 20, 2018 10:13pm
thank you very much Pawel and TradeForexFx.
great EA, great Indi.
I'll stay here for a while
Attached Image(s) (click to enlarge)
Click to Enlarge Name: MetaTrader 4 Admiral Markets.png Size: 129 KB
Click to Enlarge Name: MetaTrader 4 Admiral Markets2.png Size: 129 KB
  • #427
  • Nov 20, 2018 11:14pm Nov 20, 2018 11:14pm
Hey, to everybody I'm not at home and my EA is just working there so I don't know if mine has opened any trade. Also you seriously don't have to be so excited about every trade. Single trade doesn't matter at all, profitable or not.

And about break even - I didn't tested that solution, can't tell you if it works. Remember that this EA is not made to feel you good "oh I cannot lose now, it's break even at worst". This EA is made to make money, so if I will ever test this solution and it will not be worth it I will not add it.

There was a good idea I have once heard - if you are excited about every single trade, then you are using too much leverage

I don't have that much time to stick to the chart so I just leaved it so I can gather some data. Let this EA work for a month or two and then we can talk about improvements. Initial work to make something is nothing when you compare that to maintaining and optimisation.

Have a great day and cheers! I have a lot of work at my college, so I can focus on anything here maybe after 20th December, because I will have a few days off.
  • #428
  • Edited 1:38am Nov 21, 2018 1:23am | Edited 1:38am
Quoting mrdfx
I did a 5 year back test with $1k start and 0.1 lots, I took a closer look at the trades and time line, and I noticed a few things. During the period of 10/2013 to 04/2015 (1.5 years) there was not much profit and the balance fluctuated between $800 and $1300, and from 05/2015-01/2016 the balance fluctuated between $1200-$2000, and after 04-05/2016 did the balance start to grow over $2000. I know that its 100% profit, but this is already 3 years of running the EA. The profit only really started to kick in in 04-05/2015, and 2018 was the most profitable...

A couple of thoughts about it. First of all...the trend. Probably a MTF indicator would fit better than a simply MA working in the same timeframe. Example: the loss last monday was clearly against the main trend (down trend VS long). I prefer to trend with the main trend... I made some manual test in MTF and you can save a couple of losses every month. Sound like not much, but it is not in the long run, if you have 56% wins as start.
Now about the test in 2013. It is ok to check back everything, but as said many times on this thread this EA (as all) should be calibrated. Look GBPUSD in 2013 was trading at 1,60 - 1,70....now at 1,28 (trading around 30% more value in that time). There was a lot of tight losses during that timeframe, this also because the SL was tighter compared to the present.

Attached an example from 2014 (middle apr to beginning of July) 8 losses against the trend VS 3 wins against the trend.
Attached Image (click to enlarge)
Click to Enlarge Name: Unbenannt.png Size: 62 KB
  • #429
  • Nov 21, 2018 1:42am Nov 21, 2018 1:42am
example of July-August 2014 with later strong downtrend: 0 losses instead of 3...

Without to code too much (even if add a MTF indy would not take much time I think)...a simple option buy-sell only would be enough to check manually the trend and just go with it. Just for those interested, you can´t backtest on MT4 with MTF indis.
Attached Image (click to enlarge)
Click to Enlarge Name: Unbenannt.png Size: 51 KB
  • #430
  • Nov 21, 2018 2:55am Nov 21, 2018 2:55am
Quoting ea0Setup
I prefer to trend with the main trend.
Yes -- I also like to check for that. However, it depends on how you determine the main trend. If you ask 10 traders you'll get 11 different responses.

The Ultron EA uses a slow 50 MA. This can serve a couple of purposes: (a) as a "proxy" support / resistance, and (b) main trend indicator.

In the Ultron code a SELL signal can only be generated if the other MAs are below the slow 50 MA, and a BUY signal can only be generated if the other MAs are above the slow 50 MA. I compared it with the Ichimoku Kumo Cloud indicator and the 50 MA tracks closely with the Kumo Cloud. So -- though the slow 50 MA may not be the most sophisticated way to monitor the main trend, it's probably good enough. And many other EAs use something similar.
  • #431
  • Nov 21, 2018 3:24am Nov 21, 2018 3:24am
Quoting robots4me
{quote} Yes -- I also like to check for that. However, it depends on how you determine the main trend. If you ask 10 traders you'll get 11 different responses. The Ultron EA uses a slow 50 MA. This can serve a couple of purposes: (a) as a "proxy" support / resistance, and (b) main trend indicator. In the Ultron code a SELL signal can only be generated if the other MAs are below the slow 50 MA, and a BUY signal can only be generated if the other MAs are above the slow 50 MA. I compared it with the Ichimoku Kumo Cloud indicator and the 50 MA tracks...

sure LOL question of taste but I doubt that in this big picture a trader can get any doubt. I am not looking for long right now.
Attached Image (click to enlarge)
Click to Enlarge Name: Unbenannt.png Size: 45 KB
  • #432
  • Nov 21, 2018 3:40am Nov 21, 2018 3:40am
Quoting M.Propper
thank you very much Pawel and TradeForexFx. great EA, great Indi. I'll stay here for a while {image} {image}
What version do you use ?
  • #433
  • Nov 21, 2018 4:03am Nov 21, 2018 4:03am
First of all, thanks for the EA. I´m testing it on a live account. Can someone add some options to the EA like money management and the magic number to enable various pairs?
Thanks
Attached Image
  • #434
  • Nov 21, 2018 4:06am Nov 21, 2018 4:06am
Quoting ea0Setup
I am not looking for long right now.
I agree. And I suspect if you drew a 50 MA on your chart you'd find that Ultron's signals would all be SELLs. You can verify this using MT4's Strategy Tester.

A few pages back TradeForexFx shared his "Helper Ultron" indicator (post #116). It works pretty well and gives you a rough idea how the Ultron algorithm might generate signals based on the current bar data displayed in your chart. I think it would confirm that if Ultron were to open any new trades they would be SELLs.
  • #435
  • Nov 21, 2018 4:21am Nov 21, 2018 4:21am
Quoting robots4me
{quote} I agree. And I suspect if you drew a 50 MA on your chart you'd find that Ultron's signals would all be SELLs. You can verify this using MT4's Strategy Tester. A few pages back TradeForexFx shared his "Helper Ultron" indicator (post #116). It works pretty well and gives you a rough idea how the Ultron algorithm might generate signals based on the current bar data displayed in your chart. I think it would confirm that if Ultron were to open any new trades they would be SELLs.
Maybe right now, but not last monday, otherwise the EA would skip that (long) loss ;-)
  • #436
  • Nov 21, 2018 5:21am Nov 21, 2018 5:21am
Quoting ea0Setup
...but not last monday...
I just now took a closer look at Monday's trade and I've attached a chart below. Using the Kumo Cloud indicator (which is very good for monitoring trends) it shows that Ultron opened a LONG trade above the Cloud. Ultron's built-in slow 50 MA pretty much tracks with the Kumo Cloud -- so, I'd say that Ultron's algorithm did the right thing.

What killed the trade was a large, 60-pip drop about 2 hours later. In other words, Ultron's BUY signal was correct, but without a crystal ball there was no way to anticipate a sudden, unexpected 60-pip drop.

With the benefit of hindsight, it looks like in the long run it still would have been a losing trade.

By the way -- ignore the other markers -- those are coming from a different EA that is running in the same account.
Attached Image (click to enlarge)
Click to Enlarge Name: ultron-trade-001.png Size: 50 KB
  • #437
  • Nov 21, 2018 5:41am Nov 21, 2018 5:41am
Quoting robots4me
{quote} I just now took a closer look at Monday's trade and I've attached a chart below. Using the Kumo Cloud indicator (which is very good for monitoring trends) it shows that Ultron opened a LONG trade above the Cloud. Ultron's built-in slow 50 MA pretty much tracks with the Kumo Cloud -- so, I'd say that Ultron's algorithm did the right thing. What killed the trade was a large, 60-pip drop about 2 hours later. In other words, Ultron's BUY signal was correct, but without a crystal ball there was no way to anticipate a sudden, unexpected 60-pip...
I understand, but I am talking to replace that MA with a longer term MTF indicator. This is the picture right now in 1H, the larger in daily I posted above. So there was no need to open long there. As I pointed out in the examples above there will be also winning trades versus the trend, but in average there will be less losses in general, if we trade just with the larger trend. Maybe I am going off topic, if so sorry.
Attached Image (click to enlarge)
Click to Enlarge Name: Unbenannt.png Size: 51 KB
  • #438
  • Nov 21, 2018 6:00am Nov 21, 2018 6:00am
Quoting ea0Setup
Maybe I am going off topic, if so sorry.
This is good stuff -- you're not off topic.

I see what you mean by the LTF (Long Time Frame) indicator -- but for me that is much too long. Most of my trades last only a few hours, so a D1 trend that is 24x longer than my H1 EA just doesn't work for me. Using a D1 time frame it looks like for the next month all trades would have to be SELLs. For an EA that already trades only once or twice a week -- forcing it to obey a D1 trend monitor means it probably will only trade a few times during the entire year. If you were to suggest an H4 trend monitor that could be more reasonable -- but I still wouldn't use it. The built-in 50 MA serves as your LTF trend monitor. I think anything more would hurt more than it helps.

It's easy enough to test, though -- simply change the 50 MA to a 1200 MA (i.e. 24 x 50) and run a back test in MT4's Strategy Tester. See how many trade signals are generated during the course of a year.
  • #439
  • Nov 21, 2018 6:23am Nov 21, 2018 6:23am
Quoting robots4me
{quote} This is good stuff -- you're not off topic. I see what you mean by the LTF (Long Time Frame) indicator -- but for me that is much too long. Most of my trades last only a few hours, so a D1 trend that is 24x longer than my H1 EA just doesn't work for me. Using a D1 time frame it looks like for the next month all trades would have to be SELLs. For an EA that already trades only once or twice a week -- forcing it to obey a D1 trend monitor means it probably will only trade a few times during the entire year. If you were to suggest an H4 trend...
Ok.

No, MTF= multi time frame, but it is the same matter. I use the 1D filter even to trade the DAX and DOW on 15M and I am happy with it. The larger view above shows daily chart with weekly filter and monthly RSI. It is ok for me too.
The entries wouldn´t change. I was expressing it badly above I see now. I would keep the EA exactly as it is, but use the indicator just as additional filter. Right now would be just short allowed. We would have less trades? Sure, but also less losses.
  • #440
  • Nov 21, 2018 6:26am Nov 21, 2018 6:26am
I do backtest this EA already I think this EA is work good. now I just try to forward test on my VPS. EA not trade every day may be take time to forward test.

if possible try to develop other version that trade every day may be very interesting.
0 traders viewing now