Lucky1.1 - page 22

 
nittany1:
I'm only running it on the 15m eur/usd right now. I'm running version 1.2 and my .set file is attached to this post. It has made more money this past week than PipBoxer has made in the past 7 months, and I paid for PipBoxer, Lucky is free.

Does the set file go into the preset folder (I assume so, but would like your conformation please).

Dave

 

There is NO stops on this EA so dont use on real account unless it has been modified. The only reason the trades turns to winners most of the time is because its Lucky that the price goes away and comes back later to be filled

However luck dont come into it when it real cash on the line so change the code to add stops. Only those running by the clock hours will be forced closed however IMO this is BAD. If the stop (if there was one) is not hit then sometimes often overnight the trade will turn to profit next morning. Scalping occurs all hours i see no benefit of time trading it should be 24 hours.

The limit order of 18 etc should never be reached in real life as the in profit code should have already closed once 1 or 2 pips ages ago. Second so many ask what time frame this runs on? IT DONT MATTER it works by the ticks and therefore works on any time frame 5 mins or 1 week its all the same. However for visual aid to see open orders on the screen 15 mins is nice.

 
felipeds:
which the configurations?

It is in the code, so you just play with seting of NLMA with false autoset. It is making much less trades, so will see, if safer, but for now i am in loose with Lucky 1.3a, hope is only cause i had bad net conection.

Marko

 
bolt:
There is NO stops on this EA so dont use on real account unless it has been modified. The only reason the trades turns to winners most of the time is because its Lucky that the price goes away and comes back later to be filled .....

Not true. 'Limit' is the SL which is checked continuously and if reached, the trade is closed by the EA - there are no hard-stops used.

 

omelette your right i stand corrected after looking at the code again i see limit is not profit limit but soft stop. Maybe this should be renamed SoftStop but actually a hard stop is much safer. If the markets suddenly bangs 100 pips in one second then you have zero chance of closing plus if you lose internet connection or windoze suddenly dies on you then your a bit stuffed.

SoftStop is good when you are trying to get closer then default 10 pips set by many brokers of course you cant set tiny stops with these brokers but at 12 to 20 pips away from then hard stops should be used. The other way to code this is you set a hard stop at 1/3 more then softstop so you have protection plus anonymous closing from shark brokers that like to go stop hunting. There is much more to be added including timer routines to check how long a trade is pending in the system once called and held by the dealer and there is no get last error on send orders or close orders so if you get a requote the EA does not know how to handle it.

 
bolt:
omelette your right i stand corrected after looking at the code again i see limit is not profit limit but soft stop. Maybe this should be renamed SoftStop but actually a hard stop is much safer. If the markets suddenly bangs 100 pips in one second then you have zero chance of closing plus if you lose internet connection or windoze suddenly dies on you then your a bit stuffed. SoftStop is good when you are trying to get closer then default 10 pips set by many brokers of course you cant set tiny stops with these brokers but at 12 to 20 pips away from then hard stops should be used. The other way to code this is you set a hard stop at 1/3 more then softstop so you have protection plus anonymous closing from shark brokers that like to go stop hunting. There is much more to be added including timer routines to check how long a trade is pending in the system once called and held by the dealer and there is no get last error on send orders or close orders so if you get a requote the EA does not know how to handle it.

Yes, I agree, the 'terminology' used is 'suspect'

I also think that for scalpers, a soft-stop is the only way to go - as you suggest, you are a stop-hunting target otherwise.

 
saleh derakhshan:
4H test is better with safe risk (5%) 2007

What broker are you using??

Everybody, please take note below!

P.S. For all others, it would be nice to have Version, Settings, Time Frames, Currency Pair, and what Broker you all are using when you report your test findings! This way we can all be on the same page.

Dave

 
zomg:
Not really sure but Im thinking Takiii is right. I looked at the trades in journal and lots of them are being closed the second they are opened. Its really too bad because I found this EA really profitable on demo. Far into too good to be true category as a matter of fact.

However, i also noticed that lots of trades don't get closed and are floating for sometimes alot more than few mins. They get closed in profit almost always down the road. Im a total nub when it comes to EAs and coding, but the way my inexperienced mind sees it is that even if immediate close fails right iff the bat, the EA still picked direction that is very likely to close in profit anyways. Not sure if the trading times 18-7 gtm have something to do with it. Even with 6 positions open floating draw down is quite low most of the time.

Im not sure why do the longer positions get closed in profit anyways but maybe however the direction of the trade is determined, it can be somehow modified to be usable on live account without closing the trade immediately.

Sry if what im saying is somewhat confusing but I myself am pretty confused about this whole thing. Im also kinda sad about Takiii's findings.

BTW does any1 know if any broker will accept this kind of EA despite what Takiii said?

Im running it on E/C and E/G pairs only, 15 min, 5% risk, version 1.3.

Lets see if i can get this attachment figured out, its quite a challenge by computer noob like my slef lol

Edit: success!!! Btw this EA will be no good on this broker because despite having good reputation for feeds, data and execution supposedlty ECN quality, it has very unfriendly scalping rules, 5 mins minimun or 7 pips.

What Broker Are You Presently Using For This Test??

Dave

 

Mmm i getting to quite like this version 1.3 multi posi system. I noticed that the system works as most contra trading systems extremely well in short term range bound markets but gets killed on trends. I also noticed that as per my years experience of backtesting it came no surprise that peaking settings for say October only while giving a fantastic result would be killed on November. I tried for 3 months time frame .....still not good then 6 and finally a years worth of data is required to compress into future positions. So now i tested from 1 oct 2006 till 1st oct 2007 then and only then can forward walk mean anything and produced a healthy out for sample gain for November 2007. This is important if you backtest you MUST forward into out of sample period as if you were going forwards live from now through December.

However i needed to do something about the error management or lack of it so i added OrderSendReliable coding with GetLast Error becuse this ensures all trades are correctly formed and makes full use of error codes returned from the server under live condition. Next i changed lot size parameters to allow trading a real account on the smallest possible account for live testing. I found i can get away with 250 dollar account with Alpari UK which allows 0.01 lots and 2 pips spreads. I also have a live NF account and can tell you they both will place live trades in about 2 to 4 seconds and are about equal with reqoutes which do occur sometimes during news spikes are technical fringes where many are fighting for position. Despite many suggesting scalping systems can not work live compared to demo thats utter BS. I can and do take live 1 and 2 pips all the time and they are filled 95% of the time to the price i want.

Next i needed to do something about trends killing the system so i added a neural network filter based on CCI and use on one hour charts. The CCI is positive trend compliant because the system itself is already contra trading the CCI over 0 is trend is long and below 0 is trend is short there is no advantage in contra a contra system. However the NN produces pattern recognition sampling of CCI and not linear above or below 0 this does not work! The NN does however inspect recognition of local minima which adds a very nice increase to Profit factor and cuts out many wasteful trades. Below is a backtest 1 hour euro/gbp 1 year and successfully walks into out of sample profit for November. The account is geared for a live test soon with 250 starting account with Alpari as shown on backtest. I am not ready to release code as much work to be done and may not even get live now till January as December can be a thin and extremely crazy month anything can happen! We could easily see euro at 155 or 140 during the lead into Xmas hols. PS i have removed parameters as they are meaningless to anyone with this tweaked code with built in NN and will only add to confusion if anyone tries to copy it.

Symbol EURGBP (Euro vs Great Britain Pound )

Period 1 Hour (H1) 2006.10.02 00:00 - 2007.10.31 23:00 (2006.10.01 - 2007.11.01)

Model Every tick (the most precise method based on all available least timeframes)

Bars in test 7680 Ticks modelled 722758 Modelling quality 90.00%

Mismatched charts errors 0

Initial deposit 250.00

Total net profit 1846901.27 Gross profit 3154389.64 Gross loss -1307488.38

Profit factor 2.41 Expected payoff 295.69

Absolute drawdown 19.30 Maximal drawdown 173680.00 (10.44%) Relative drawdown 33.44% (1553.47)

Total trades 6246 Short positions (won %) 2580 (94.88%) Long positions (won %) 3666 (95.66%)

Profit trades (% of total) 5955 (95.34%) Loss trades (% of total) 291 (4.66%)

Largest profit trade 9575.00 loss trade -16920.00

Average profit trade 529.70 loss trade -4493.09

Maximum consecutive wins (profit in money) 259 (79905.54) consecutive losses (loss in money) 13 (-516.00)

Maximal consecutive profit (count of wins) 418515.00 (170) consecutive loss (count of losses) -154170.00 (12)

Average consecutive wins 40 consecutive losses 2

Graph

edit so say despite unbelievable gains this in the real will never be achieved. Not because the pips or technique is flawed but because as soon as the account grows to about 50k or 1 lot trading it will then be targeted by broker and put on manual or delayed MUCH more then the usual 2 to 4 seconds. However, up to 50k should be possible where one should take a fat profit and start again.

 
bolt:
.....edit so say despite unbelievable gains this in the real will never be achieved. Not because the pips or technique is flawed but because as soon as the account grows to about 50k or 1 lot trading it will then be targeted by broker and put on manual or delayed MUCH more then the usual 2 to 4 seconds. However, up to 50k should be possible where one should take a fat profit and start again.

Wow, if only that were so - I mean, being 'allowed' to scalp to 50k with an MT broker un-impeded. Truth is, there is not one MT broker that is happy with their clients scalping, despite the rhetoric - I thought I had found an exception, but this is not the case...

The really fustrating thing is that 'real' ECN's really do not have a problem with it, just the bucketshops, becuase the are taking the other side of the trade, and hooking up an automated system to them is proving non-trivial...