Thank you Mcclaine, I will check it out.
Thank you Mcclaine, I will check it out.
podberry
Well-known member
- Messages
- 395
- Likes
- 1
Hi All, where is a good place (website etc.) to learn how to do accurate MT4 back-testing, as most of you know I am a converted Oandian and have never traded MT4 prior to FMT. Would be nice to do my own testing in the future. Your help would be appreciated. 🙂
pipe_kasuko
Newbie
- Messages
- 1
- Likes
- 0
Hi all,
I purchased and installed the EA just yesterday, and have had my first profitable trade already this morning!! 🙂
Look forward to reading all 1000 pages - should take me a couple of months ..
I purchased and installed the EA just yesterday, and have had my first profitable trade already this morning!! 🙂
Look forward to reading all 1000 pages - should take me a couple of months ..
We're on Daylight savings time. The manual says 8:30 for FXDD, but that's Standard time.
Looks like to me you need to set your EA back an hour to 07:15 and or 07:30 until the time changes later this year. Depending upon where you're located, the correct trading time here on the East coast of the U.S. would actually be 1:15, and 1:30 a.m. Though remember, the broker and London's time changes are not necessarily changed forward or back on the same weekend, so you need to be careful during those time changes each Spring and Fall.
To check the correct times you need go to the World Clock website.
Yep, £ rocketed through the longer term resistence levels around 1.6400'ish, a day when a trailing stop would have worked perfectly, rate rise potential will be helping the £ also, but the $ is in a dire state, with the US devalueing to try and drive up exports and inflate with QE, have a look at the $ Index - however, all this could quickly reverse if the US Gov buckles under public pressure and starts raising interest rates - either way - not looking good in the US - I'm also currently short the S&P500 which is looking ever more fragile - have a nice Easter!🙂
Have a look at the pages from mid March when we had lots of postings in relation to London BST time changes and broker times.
Hi Pod,
Post 6348 Feb 20th from Chris Heath was good reply, although I've still not had time to do it myself yet - post content copied below.
Re: ForexMorningTrade journal
--------------------------------------------------------------------------------
Quote:
Originally Posted by Remraf
A lot of posts seem to revolve around the ability to do backtesting looking for the Holy Grail - is this a dark art only performed by the chosen few - or can anyone have a go at it? - I'd be pleased to be enlightened, as long as it doesn't involve a) Being exceptionally clever, or b) too much time on my hands!
Time for bed
Its basically quite straightforward. This may appear daunting, but its really not difficult.
Once you have it setup, you can quickly test an EA with varying parameters.
Make sure you EA is loaded into MT4. On the chart go into 'View' , at the top, then 'Strategy Tester'. A window opens up below the chart(which you can enlarge, if you want).
Now you can choose your EA, the symbol, timeframe, model accuracy, (Control points for quick, inaccurate test, Every tick for slower, more accurate test).
'Use date' allows you to choose the date range for backtesting. Visual mode allows you to see the trades as they occur on the unfolding chart, but its SLOW.
'Expert properties' shows you the options for the EA, 'Testing' shows initial deposit, currency, and whether you want to test Long only, Short only, or both. (Optimization, you can play with later). ' Inputs' shows the options for the EA , under 'Value', eg Start Time,Stop Loss, Target etc. Click OK to exit - maybe twice.
Once you've set up your EA parameters, pressing 'Start' will magically run the test. The tabs at the bottom of the Testing window change, and you can see a 'Graph' develop, look at the trade 'Results', and , when test is finished, view the 'Report'. (You can save the 'Report' by right clicking on the Report, and clicking on umm.. 'Save as Report.)
You can view the chart with the trades that occurred by clocking on 'Open Chart'. There's an option to modify the EA, only if you have the source code. (Not for FMT, or TMT.)
Optimizing goes a step further. You can get the Strategy tester to run through a series of numbers, eg Stop Loss from 20pips to say 50 pips in steps of 1, and generate a Report.
To do this, click on 'Expert Properties', 'Inputs'. Go to the parameter you're trying to optimize. If the little box, at the far left is white, you can optimize that parameter. Click on the box, and a tick will appear, and set your optimizing setup by setting 'Start' value, eg 20 for Stop, 'Step' for an increase of 1 (ie pip) each time, and 'Stop' value eg 40 . I would advise only optimizing one parameter to start with.
Click OK to exit.
If you look underneath 'Period' on the main Tester screen, you'll see 'Optimization'. Click on that, a tick appears. Click on 'Start'.
An optimization graph tab appears, and an Optimization Results tabs, which are what they say. (You can 'Save as Report', as before.)
The results generated are different from a non-optimised test, and can indicate the best parameter for the EA.
To learn to use the Strategy Tester, it may be best to run a few non optimized tests, with a small date range ( maybe 2 months) - to get you're feet wet with all the available options, and to get familiar with the results, and what they mean.
You can, of course try the different settings you've seen in the Forum.
If you run lomger tests (say, a year or more), you'll see the length of time it takes.
When you optimize, you'll multiply that time by the number of steps in your optimization. It can be hours for say 3-4 year backtesting, covering a lot of steps. (Unless you have a spare Tianhe-1A computer.)
That is a very brief overview, but should get you going. Then you'll discover the nuances of backtesting on MT4. And quite possibly drive yourself mad. But don't concern yourself with that just yet!
Good luck! Chris
This is a revised posting of my manual look-back at FMT trades using the new settings announced this week. I thank Mcclaine for pointing out an apparent error. It turns out that I didn't have my head completely out of my butt when I uploaded a spreadsheet summary.
The error does not change my basic work, which I will show you in the attached full spreadsheet, but the way that I did the extensions with the spreadsheet functions. Somewhere along the line, the summary dropped out some monthly column results off to the left of the summary when I tried to convert to an older version of Excel . Let's leave it at that.
For the last three weeks, the results are actually from observed trades while following Hammy's results and double checking on my own platform.
The 06:15 results are quite a bit better than the previous default 06:30 setting, so I'm going with it. (I know Mark has said this, but I only trust my own work). Thanks to Hammy for the great idea. The 2010 results give me a warm and fuzzy because they show good consistency.
Truth is, a lot of settings work around this 06:30 timeframe, which is another good indicator of how robust the system is.
The results of each trade in each monthly column was checked using tick data (with a fixed 2.5 pip spread). In addition for 2011, I backchecked this tick data against my own broker's actual data (Go Markets), and with Alpari UK, whose trades are sometimes different on Mondays due to a different start times compared to my baseline data.
I am very confident in the daily results that you see. In some cases where the fill was too close to call, I called it a loser or a BE trade as applicable, so any testing errors are on the conservative side.
The spreadsheet is in Open Office format, but should open properly in any newer Excel version.
Sorry for the previous false alarm. (Spreadsheet is in next post)
Al
The error does not change my basic work, which I will show you in the attached full spreadsheet, but the way that I did the extensions with the spreadsheet functions. Somewhere along the line, the summary dropped out some monthly column results off to the left of the summary when I tried to convert to an older version of Excel . Let's leave it at that.
For the last three weeks, the results are actually from observed trades while following Hammy's results and double checking on my own platform.
The 06:15 results are quite a bit better than the previous default 06:30 setting, so I'm going with it. (I know Mark has said this, but I only trust my own work). Thanks to Hammy for the great idea. The 2010 results give me a warm and fuzzy because they show good consistency.
Truth is, a lot of settings work around this 06:30 timeframe, which is another good indicator of how robust the system is.
The results of each trade in each monthly column was checked using tick data (with a fixed 2.5 pip spread). In addition for 2011, I backchecked this tick data against my own broker's actual data (Go Markets), and with Alpari UK, whose trades are sometimes different on Mondays due to a different start times compared to my baseline data.
I am very confident in the daily results that you see. In some cases where the fill was too close to call, I called it a loser or a BE trade as applicable, so any testing errors are on the conservative side.
The spreadsheet is in Open Office format, but should open properly in any newer Excel version.
Sorry for the previous false alarm. (Spreadsheet is in next post)
Al
Last edited:
deserteagle
Active member
- Messages
- 205
- Likes
- 1
Hey, I will add something about backtesting in MT4.
Spread: In the historical data the spread values are not saved, so which is the spread value used in the backtest? the current one at the moment of backtesting, therefore if you run the same test many times you can get different values due to the differences of the spread, this makes our test not valid. Also, for brokers with variable spread, the same test run on a weekday in the morning or during the weekend will give us results totally differents, mainly if we are testing a scalper system.
How to solve this issue? There are some ways, but one of them is really simple, install a spread indicator, so you can see the spread value at anytime, open the login window and write user 5555, then when you see the spread you want just click to login with the invalid user 5555 (it could be any number), this action will disconnect our MT4 client from the server, so the selected spread value will be frozen for your tests. Remember to run all your tests using the same spread value, also consider this value should be realistic, not too high, no too low.
Metaquotes data: When downloading data from MT4 (Menu -> Tools -> History Center) you should know the data quality of the broker. For example, Alpari UK have same gaps (many days) in the data, so the quality is not 100%, I think is not easy to find data for MT4 at 100% of quality. But after trying hard, I found Alpari NZ data has high quality, therefore I use this broker for my tests.
Sorry for my off topic.
Cheers.
DesertEagle aka SandBird 😆
Spread: In the historical data the spread values are not saved, so which is the spread value used in the backtest? the current one at the moment of backtesting, therefore if you run the same test many times you can get different values due to the differences of the spread, this makes our test not valid. Also, for brokers with variable spread, the same test run on a weekday in the morning or during the weekend will give us results totally differents, mainly if we are testing a scalper system.
How to solve this issue? There are some ways, but one of them is really simple, install a spread indicator, so you can see the spread value at anytime, open the login window and write user 5555, then when you see the spread you want just click to login with the invalid user 5555 (it could be any number), this action will disconnect our MT4 client from the server, so the selected spread value will be frozen for your tests. Remember to run all your tests using the same spread value, also consider this value should be realistic, not too high, no too low.
Metaquotes data: When downloading data from MT4 (Menu -> Tools -> History Center) you should know the data quality of the broker. For example, Alpari UK have same gaps (many days) in the data, so the quality is not 100%, I think is not easy to find data for MT4 at 100% of quality. But after trying hard, I found Alpari NZ data has high quality, therefore I use this broker for my tests.
Sorry for my off topic.
Cheers.
DesertEagle aka SandBird 😆
I will not be trading Friday or Monday 4/22 and 4/25 due to my broker spreads during this time period. I have monitored them this evening and they are double than normal. With AUD NZD GBP EUR all with bank holidays this is the reason for the spread issue for my broker. I too am treating this just like the december period. The normal market dynamics will not be in play therefore the odds will not be stacked in my favor. The spread issue is my major decision not to trade.
:smart:
wiseambitions
Established member
- Messages
- 764
- Likes
- 7
I think it would be futile to trade on a day like this, there was no action in the Asian session after US ended, and it would only be a major event that woke up any of the big dogs to trade in the Europe and London session
It was only because the £ came off the top by the end of the US session that the indicators would now say sell. But volume is so negigible that it wouldn't take a lot to tip the market the wrong way if it moved at all today. That's my pennyworth!
Have a great weekend
It was only because the £ came off the top by the end of the US session that the indicators would now say sell. But volume is so negigible that it wouldn't take a lot to tip the market the wrong way if it moved at all today. That's my pennyworth!
Have a great weekend
RedGuerrilla
Active member
- Messages
- 218
- Likes
- 1
Thank you . . . I've been staying with it since Nov 8, 2010 . . .
RedGuerrilla
Active member
- Messages
- 218
- Likes
- 1
Stats for 04-21-2011
Live Accounts:
FMT 4.3, EA standard settings: 45 SL, 25 BE, +5 BE, 35 TP
6:15 UK Time, FXSolutions..., Long/Profit...., +35.0, Stats since 11-08-2010: Total= +172.0
6:15 UK Time, IBFX.........., Long/Profit...., +34.2, Stats since 01-04-2011: Total= +366.3
6:15 UK Time, PFG Best......, Long/Profit...., +35.7, Stats since 03-01-2011: Total= +147.2
Demo Accounts:
FMT 4.3, EA standard settings: 45 SL, 25 BE, +5 BE, 35 TP
6:15 UK Time, PFG Best......, Long/Profit...., +35.2, Stats since 02-01-2011: Total= +466.1
6:15 UK Time, FXCM.........., Long/Profit...., +34.9, Stats since 02-01-2011: Total= +494.8
6:15 UK Time, Alpari(US)...., Long/Profit...., +35.1, Stats since 02-01-2011: Total= +456.8
6:15 UK Time, FXDD.........., Long/Profit...., +36.6, Stats since 02-01-2011: Total= +383.2
Notes:
All pip totals include slippage
FXSolutions is the only dealing desk broker
Live Accounts:
FMT 4.3, EA standard settings: 45 SL, 25 BE, +5 BE, 35 TP
6:15 UK Time, FXSolutions..., Long/Profit...., +35.0, Stats since 11-08-2010: Total= +172.0
6:15 UK Time, IBFX.........., Long/Profit...., +34.2, Stats since 01-04-2011: Total= +366.3
6:15 UK Time, PFG Best......, Long/Profit...., +35.7, Stats since 03-01-2011: Total= +147.2
Demo Accounts:
FMT 4.3, EA standard settings: 45 SL, 25 BE, +5 BE, 35 TP
6:15 UK Time, PFG Best......, Long/Profit...., +35.2, Stats since 02-01-2011: Total= +466.1
6:15 UK Time, FXCM.........., Long/Profit...., +34.9, Stats since 02-01-2011: Total= +494.8
6:15 UK Time, Alpari(US)...., Long/Profit...., +35.1, Stats since 02-01-2011: Total= +456.8
6:15 UK Time, FXDD.........., Long/Profit...., +36.6, Stats since 02-01-2011: Total= +383.2
Notes:
All pip totals include slippage
FXSolutions is the only dealing desk broker
shoot2scoot
Active member
- Messages
- 150
- Likes
- 1
I'm in for low £pp because of the technical setup, sandwiched between the weekly R2 and above the daily Pivot, a sudden break might generate a rally south. Needless to say I will manage manually!
