How to get 99% tickdata for metatrader backtesting? | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

MetaTrader Backtesting questions 4 replies

MetaTrader Expert Advisor BUILDER for backtesting and auto trading 4 replies

metatrader- bad backtesting or bad broker quotations? 0 replies

Tickdata during visual strategytester modus 0 replies

MetaTrader BackTesting, Reliable or Not?? 17 replies

Attachments: How to get 99% tickdata for metatrader backtesting?
Exit Attachments

How to get 99% tickdata for metatrader backtesting?

  • #21
  • Edited 2:18pm Sep 6, 2013 7:59am | Edited 2:18pm
Guys, many thanks for your suggestions, very helpful.

I am attaching a very small sample of results, from exactly the same period, just 2 days (4H XAUUSD). Forex.com V Armada Markets.

NOTE: The smallest lot size on Forex.com is 0.10, whilst Armada is 0.01 . Hence, the opening balance for Armada was $22,000
this way they both start on 0.10 (since, I am using a Lot Ratio Multiplier system).

ALSO: ''Leverage'', I think Forex.com is on 200 & Armada is on 250?

How many discrepancies are there?

Again, appreciate your views.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Armada Markets XAUUSD H4 (Results).PNG Size: 73 KB
Click to Enlarge Name: Forex.com XAUUSD H4 (Results).PNG Size: 85 KB
  • #22
  • Edited 8:36am Sep 6, 2013 8:16am | Edited 8:36am
OK, here is a screenshot of the graphs both identical periods; from 2013/01/24 to 2013/09/06 (7 months)!!!

Same settings, but just look at the difference!
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Armada Markets (Graph).PNG Size: 60 KB
Click to Enlarge Name: Forex.com (Graph).PNG Size: 49 KB
  • #23
  • Sep 6, 2013 10:42am Sep 6, 2013 10:42am
Right click on Results tab area -> "Save as Report". It would be more convenient to work with.

One more thing, if you have access to source code and know how to do this then try to switch off MM, so tests will be performed with fixed lot.
  • #24
  • Edited 1:43pm Sep 6, 2013 1:20pm | Edited 1:43pm
Thanks again, for helping out, attached are the Saved Reports.

I couldn't switch off MM, so it has to be compensated for.

They also have different Leverage, again I think it's 200 & 250, not sure really!
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Graph, StrategyTester (Forex.com).gif Size: 7 KB
Click to Enlarge Name: Graph, StrategyTester (Armada Markets).gif Size: 7 KB
Attached File(s)
File Type: txt Report, StrategyTester (Forex.com).txt   72 KB | 371 downloads
File Type: txt Report, StrategyTester (Armada Markets).txt   79 KB | 347 downloads
  • #25
  • Sep 6, 2013 3:47pm Sep 6, 2013 3:47pm
Why don't you simply run each one in visual mode in the strategy tester and compare each trade? It should become clear pretty quickly what the differences are.

Are the candles the same?
Are the signals the EA is trading off the same?
Is the spread the same?
And so on.
  • #26
  • Sep 6, 2013 3:53pm Sep 6, 2013 3:53pm
Quoting Xaphod
Why don't you simply run each one in visual mode in the strategy tester and compare each trade? It should become clear pretty quickly
I did, they're not the same, it trades differently on every broker, very strange!
  • #27
  • Sep 6, 2013 4:00pm Sep 6, 2013 4:00pm
Quoting Xaphod
Are the candles the same? Are the signals the EA is trading off the same? Is the spread the same? And so on.

Everything is the same!

Something even more spooky, if I optimize it on Forex.com it will not make a profit on Armada Markets. BUT if I optimize on Armada, it will
make a profit on Forex.com.
  • #28
  • Edited Sep 7, 2013 2:04pm Sep 6, 2013 4:22pm | Edited Sep 7, 2013 2:04pm
Quoting GreatForex
{quote} Everything is the same! Something even more spooky, if I optimize it on Forex.com it will not make a profit on Armada Markets. BUT if I optimize on Armada, it will make a profit on Forex.com.
An indicator like the CycleIdentifier may give different signals if it does not start at the point in time on the charts.

Ie you should see to it that the historical data on both brokers is the same. Same starting time, same data.

It could be the neural network used in the EA that causes to the variation:
"The final cream on the pie is an exclusive Neural Network Propagator designed to optimize the strategy and filter out unexpected losing trades. By the same token the NN Propagator confirms all orders prior to trading; it has the final say in the various stages that lead FCP to open a buy, or sell order."


Edit:
Without information on how the EA works internally, I do not think you will find the answer to your question.
  • #29
  • Edited Sep 7, 2013 2:06pm Sep 6, 2013 4:46pm | Edited Sep 7, 2013 2:06pm
Your posted simulation reports use Open prices. Do you know that the EA functions correctly in that mode?

"Model Open prices only (only for Expert Advisors that explicitly control bar opening) "

Given the description of the EA: 8 indicators, 153 variables, neural networks, complex math, yada yada, I would think it would also need 'all the data' to function correctly, not just opening prices.
  • #30
  • Oct 28, 2013 4:14pm Oct 28, 2013 4:14pm
On the subject of Broker discrepancies, I found something very strange, which is causing havoc with the EA.

When trading GOLD in H4.

Dukascopy has 30 candles to the week.
Forex.com has 31 candles to the week.
Armada Markets has 32 candles to the week.

Could someone please explain, what is going on?
  • #31
  • Last Post: Oct 28, 2013 7:36pm Oct 28, 2013 7:36pm
Quoting GreatForex
On the subject of Broker discrepancies, I found something very strange, which is causing havoc with the EA. When trading GOLD in H4. Dukascopy has 30 candles to the week. Forex.com has 31 candles to the week. Armada Markets has 32 candles to the week. Could someone please explain, what is going on?
Because of the difference in their server time ... but also based on their friday close and sunday/monday open times.
MT4/MT5 EAs/Indicators/Alerts coder
0 traders viewing now