Yeoeleven,

Are you manually interfering with your combo much? I read earlier you close trades and turn EA's off during major news announcements. Was that for this combo or something else you were running? Thx

Is the Step value of 18 only used for Backtesting in the 10p4 EA as well? if not what does it do?
Thanks

Hi Matrixebiz,

The step value or pipstep is the distance in pips from 1 order to the other in the martingale progression,if its 18 then the price will have to move 18 pips aganist the order until another order is opened, up to maximum trades.It is used in backtesting and live trading.


Hope this helps
tools

OK guys, i already have a list of people around. I'm closing the recruitment now. Late comer[s] will not be entertain, hope this time we're gonna kick some butt. A quick reminder to the testers, backtest for this system aint really help, here is what happened to the optimized backtest pin point from 5th February ~ 1st March, 2007. Make a comparison by yourself with the forward testing results. Trades been taken is totally different and the closing of trade is sucks on backtest. No code will be distribute, if not this little freak will go on sale on ebay with a different name again. Why I sound so mean?! I've seen people use my result to sell sucky EA too. Hope to see some action by next week.

Regards,

David

Below attached a set of backtest result according to forward testing period, and a set of forward testing result. Both uses same setup and same balance amount to start testing. You can judge by your own, how reliable 90% modelling quality is. Check over the screenshot on backtest and live demo too. Its totally different! Good luck

Attached Images

	TesterGraph.gif (29.0 KB, 230 views)
	livegraph.gif (38.3 KB, 239 views)

Attached Files

	10point3_Backtest.zip (9.0 KB, 73 views)
	10point3_DynamicStop.zip (11.1 KB, 149 views)

Well it worked on my demo account using both EAs trading 1 pair each starting at .1 units with an account of $1870 but that was too risky for live trading.

I would use a broker who accepts .01 trades and trade one EA with one pair under test with a starting balance of $500 and hope that profit has been accumulated prior to the first MaxLot loss.

Of course settings will pay a huge part in determining the success of the venture.

Mod1e does not require any additional indicators it uses CCI and MACD both of which are in the indicators folder by default. GoblinFibo uses the indicators attached below.

During the running of these EAs I have manually operated only once to close off both whilst in profit prior to the news events as previously posted. I intend also to close them for the Non Farm Payroll announcements as well as a couple of other volatile announcements. I do not close down for the weekends so there is minimal interferance, of course that would affect any back testing which carried on through the events.

John

Attached Files

	Turbo_JRSX.mq4 (2.7 KB, 67 views)
	Turbo_JVEL.mq4 (11.7 KB, 63 views)
	Turbo_JMA.mq4 (10.8 KB, 71 views)

Yes, but not that different. Yeoeleven got good results, yet the backtest completely washes out. There has to be another reason.

The reason is, new V12 is not closing order based on TP most of the time. Look carefully on the screenshot, at backtest it wont bail the position whilst in profit, but forward tester did bailed. Live account bail less often then demo from my observation. This is the purpose we called for group forward testing.

Regards,

David

Thanks to David for V12. I shall be keen to get that up and running tommorrow night.

I really wish I could get to the bottom of the backtesting problem. I laid awake last night thinking what could cause it. I hit on the idea of differences between brokers. My Mod1e backtest was with Interbank, and I noticed that John's forward testing was with Neuimex. I've just repeated the exercise with Neuimex, fully expecting miraculous results, backing up John's findings. Unfortunately, whilst it did hang on a little longer, the end result was the same .... washout !

I set V12 running on InterbankFX at the same time, on another machine (just out of interest .. as we obviously can't trust it), and it didn't get very far at all. It washed out in 2 days!

Does anybody have a clue why this is happening? What is the point of having a backtesting facility, if the results are absolute rubbish?

It's very worrying, because if you watch the testing, it all seems very credible. Who's to say these washouts can't happen in the real (or forward testing) environment?

Am I missing something very obvious?

Ray

Apologies, I saw V12 (which I hadn't got at that stage) and thought you were referring to something else.

I'm currently dissecting the backtest and John's forward test .. and looking at how the closures differ.

Cheers,

Ray

I came across a thread that mrtools are having an arguement with 1 of the member trying to sell 10point3 mod. And the martha farker claims that mr.tools is testing for too long period, and wanting to make a backtest without MM to prove who's version is better. I ran through a backtest with V12 on conservative mode and tight stop loss, tight pip step, this is the out come without MM. Look carefully on the starting capital and the balance after a year without MM. I think that is the purpose of backtest. No doubt, for me its still rubbish, but atleast I have some important information in my mind, I'll know what is the next step of development. If I modify my EA, what should take into consideration. I personally have 4 different folder on my desktop to collect EAs from different author and I also bought a few commercial EA(thought it would work )

Category A
A best EA did tremendous result on at least 90% and above MQ backtest and getting about half of the performance on live trading compare to backtest.

Category B
A good EA did extremely good result on at least 90% and above MQ backtest and getting less than 20% of the performance of live trading compare to backtest(this type of EA would be over optimize, and pattern recognization. NFP and FOMC will kill them) 10point3 fall in this category

Category C
A bad EA did good result on at least 90% MQ backtest because without stop loss, use this kind of trading robot on live account is a betting game!

Category OMFG (*Oh my farking god)
A worst EA did good result on Control Point only on backtest and wipe out account on the second day after you fund in your live account!

Always remember, 1 year 90% MQ backtest will not give you your proper stop loss or tp, it gives you the idea of whether your EA are coded correctly, whether the stop loss are in place, whether your money management algorithm is working fine. And would like to say that, mr.tools did a great job on 10point3 volatility mod. That is the craziest so far I've ever seen. I hope it works.

Regards,

David

Attached Images

original test file too big.GIF (18.7 KB, 515 views)