Great EA in backtest! - page 144

New comment
 

I don't know much about this EA but I have been getting these errors

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: ErrorValues:Symbol=EURUSD,Lots=2,Bid=1.3475,Ask=1.3478,SlipPage=3StopLoss=1.346,TakeProfit=40

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: Error on the Long Entry: 130

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: OrderSend error 130

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: StopLevel:1.346

2008.02.07 19:30:17 CyberiaTrader_v1.85g-p142 inputs: ReverseIndex=3.82; MoneyTrainLevel=4; MACDLevel=10; MAXLots=2; ValuesPeriodCount=7; ValuesPeriodCountMax=7; SlipPage=3; Lots=0.01; StopLoss=30; TakeProfit=40; SymbolsCount=2; Risk=5; StopLossIndex=2.5; StaticStopLoss=15; StopLevel=0; TrailingStopFactor=15; GMT=0; MagicNumber=123000;

2008.02.07 19:30:16 CyberiaTrader_v1.85g-p142 EURUSD,H4: loaded successfully

 
Beno:
I don't know much about this EA but I have been getting these errors

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: ErrorValues:Symbol=EURUSD,Lots=2,Bid=1.3475,Ask=1.3478,SlipPage=3StopLoss=1.346,TakeProfit=40

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: Error on the Long Entry: 130

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: OrderSend error 130

2008.02.07 19:30:23 2007.05.29 20:00 CyberiaTrader_v1.85g-p142 EURUSD,H4: StopLevel:1.346

2008.02.07 19:30:17 CyberiaTrader_v1.85g-p142 inputs: ReverseIndex=3.82; MoneyTrainLevel=4; MACDLevel=10; MAXLots=2; ValuesPeriodCount=7; ValuesPeriodCountMax=7; SlipPage=3; Lots=0.01; StopLoss=30; TakeProfit=40; SymbolsCount=2; Risk=5; StopLossIndex=2.5; StaticStopLoss=15; StopLevel=0; TrailingStopFactor=15; GMT=0; MagicNumber=123000;

2008.02.07 19:30:16 CyberiaTrader_v1.85g-p142 EURUSD,H4: loaded successfully

Ha! you found that my fix made things worst - thanks for the post. This time I won't say that I've corrected it, just that I have made a change.

The problem was that the take-profit calcuation was producing a number lower than the Ask Price on a buy.

Would you try it again?

 

Hi everyone:

I use "v1.85g jpy" with other name in my live miniaccount without errors, since a couple of day.

I have a question: I need to do optimization over this EA?

Thanks in advance

 

I been playing with this EA for over a year now and still find it the most intriguing system ever released! For reasons im not going into now but this EA peaked then started to fail around August 2006. For 5 years prior this EA would have made a fortune but the market changed across all major as the dollar sank fast. I now know every scrap of code in this EA i could almost write it with my eyes shut and I used it sparingly on live accounts.

I was fortunate enough to witness 2 months of profits in Q4 2006 but by the time we see Xmas 2006 this thing died with the rallies. ALL majors were effected. I was hoping by January 2007 things would normalize but alas no. However, the backtest over this period also showed this EA was dead so it totally rules out that the backtest are flawed. When the backtest shows money i made money. I know there is much to say about the backtester being flawed but you must get 90% on the tester. Then on a live system you can run this for one month maybe two. Then rebacktest the same period and its VERY VERY close. Almost without fail the trades open and close in the same places except when a stop is hit by 1 pip live will change things a bit. Both for the period it makes money and when it has lost. If you dont see this then your broker is bad. Must have 2 pips euro and 3 spreads other majors fixed permanent no matter what news comes out and live trades must be filled sub 3 seconds.

Throughout 2007 i fired it up sometimes ran all my routines and still no good.

The bottom line is this. No amount of filters can change the market conditions. This is a trend fake system. It waits for a pulse and grabs a few pips on the pullback. Those pullbacks have diminished to almost zero. There is a certain quality of pullback this system requires not just a visual scan of the eye. It may recover one day but then it might also take another 4 or 5 years. The fact the good trend lasted from 1999 to 2006 it wouldnt surprise me to wait another 5 years although i won't hell you can rob a bank and do less time LOL

Any attempt to add ADX, madc, rsi etc basically ANYTHING you throw at it does not change it and that applies to all majors. But you can fake it with filters and throw it in to a pure curve fit mode where is looks good but it fails in the days or weeks ahead trying to use it live due to the curve fit induced. You know when you have achieved this fake profit fit once you have employed 2 or more filters with fixed stops and tp's. When the market is right you don't need any dials or filters at all......none! But there are some coding issues that need sorting including tweaks to the probability logic. Then it works best on 1 hour.

This EA must be carefully tuned to one pair at a time. The market depth, volatility, noise floor everything has to be just right so don't go banging this out on 6 pairs and believe it will make money. Never has, never will, not even if you see the right market again in the future. Right now and for the last 13 months the market is in trend mode and breakout mode not recursive mode. The other many other EA's which have also stopped working through 2007. Going to have to wait to get another solar eclipse or something before this works again.

There is one final explanation. Metaquotes may have possibly issued a patch to all MT4 brokers to safeguard against using this system. Many brokers banned its use when it first came out. Its rather ironic the EA stopped working within a few weeks of public release. The patch may also effect the meta quotes collected backtest data which has been somehow filtered. We know special system filters exist as they were forget to be turned on during the 2007 metaquotes contest allowing a pipster system to make a ton till it was filtered out. One of the common filters is that every tick is passed to you late like 1 second late. The broker therefore has hindsite. Lets say your pipsing for 1 pip up on euro. Broker already knows the next ticks is down before you do so he accepts your offer on the old price then drops down 2 ticks to sync back with the market price. This has nothing to do with spread its the stolen pip trick. Now the other way is market jumps 3 ticks up while your price is in the comparator. Your seeing the delayed price but broker knows market is up. The solution is now to requote you:)

 

Fascinating commentary. I enjoyed this much and learned a little.

ES

bolt:
I been playing with this EA for over a year now and still find it the most intriguing system ever released! For reasons im not going into now but this EA peaked then started to fail around August 2006. For 5 years prior this EA would have made a fortune but the market changed across all major as the dollar sank fast. I now know every scrap of code in this EA i could almost write it with my eyes shut and I used it sparingly on live accounts.

I was fortunate enough to witness 2 months of profits in Q4 2006 but by the time we see Xmas 2006 this thing died with the rallies. ALL majors were effected. I was hoping by January 2007 things would normalize but alas no. However, the backtest over this period also showed this EA was dead so it totally rules out that the backtest are flawed. When the backtest shows money i made money. I know there is much to say about the backtester being flawed but you must get 90% on the tester. Then on a live system you can run this for one month maybe two. Then rebacktest the same period and its VERY VERY close. Almost without fail the trades open and close in the same places except when a stop is hit by 1 pip live will change things a bit. Both for the period it makes money and when it has lost. If you dont see this then your broker is bad. Must have 2 pips euro and 3 spreads other majors fixed permanent no matter what news comes out and live trades must be filled sub 3 seconds.

Throughout 2007 i fired it up sometimes ran all my routines and still no good.

The bottom line is this. No amount of filters can change the market conditions. This is a trend fake system. It waits for a pulse and grabs a few pips on the pullback. Those pullbacks have diminished to almost zero. There is a certain quality of pullback this system requires not just a visual scan of the eye. It may recover one day but then it might also take another 4 or 5 years. The fact the good trend lasted from 1999 to 2006 it wouldnt surprise me to wait another 5 years although i won't hell you can rob a bank and do less time LOL

Any attempt to add ADX, madc, rsi etc basically ANYTHING you throw at it does not change it and that applies to all majors. But you can fake it with filters and throw it in to a pure curve fit mode where is looks good but it fails in the days or weeks ahead trying to use it live due to the curve fit induced. You know when you have achieved this fake profit fit once you have employed 2 or more filters with fixed stops and tp's. When the market is right you don't need any dials or filters at all......none! But there are some coding issues that need sorting including tweaks to the probability logic. Then it works best on 1 hour.

This EA must be carefully tuned to one pair at a time. The market depth, volatility, noise floor everything has to be just right so don't go banging this out on 6 pairs and believe it will make money. Never has, never will, not even if you see the right market again in the future. Right now and for the last 13 months the market is in trend mode and breakout mode not recursive mode. The other many other EA's which have also stopped working through 2007. Going to have to wait to get another solar eclipse or something before this works again.

There is one final explanation. Metaquotes may have possibly issued a patch to all MT4 brokers to safeguard against using this system. Many brokers banned its use when it first came out. Its rather ironic the EA stopped working within a few weeks of public release. The patch may also effect the meta quotes collected backtest data which has been somehow filtered. We know special system filters exist as they were forget to be turned on during the 2007 metaquotes contest allowing a pipster system to make a ton till it was filtered out. One of the common filters is that every tick is passed to you late like 1 second late. The broker therefore has hindsite. Lets say your pipsing for 1 pip up on euro. Broker already knows the next ticks is down before you do so he accepts your offer on the old price then drops down 2 ticks to sync back with the market price. This has nothing to do with spread its the stolen pip trick. Now the other way is market jumps 3 ticks up while your price is in the comparator. Your seeing the delayed price but broker knows market is up. The solution is now to requote you:)
 

thanks bolt for you contribution,

very intereresting

 

Thanks a lot, Bolt

 

maybe we aren't done just yet...

What if it was re-coded for a broker that doesn't use MT? Say, OandA? would they block it? could they? It would be a journey, but if the market needs to be right too, then it might be a fruitless one. What do you think, bolt?

 

Oanda doesn't have free automation. They have an API but its expensive 600 bucks a month or used to be. Plus oanda spreads are wild and variable so wont work anyway.

Its probably more to do with the market then anything else as too many other EA's that rely on price retracement don't work either its not just a problem for CT. If euro drops below 1.40'sh by the summer and goes into a tight range then just might get to some quality pullbacks. But if dollar keeps weak the trend bursts far outstrip the retracements and these systems keep getting premature stop hits for losses.

In the meantime you should be using trend following and breakout systems. Don't touch anything martingale based or 10Point3, Pheonix etc

There is no point trying to force these systems into something they cant cope with. Each month just run an honest backtest using the same settings that worked back in pre 2006 and see if it has made a good profit. If not then don't even bother with it.

 

what about a mt4 to oanda bridge through api or they have a oanda to trade station bridge here...

Shield-systems technology company

1...137138139140141142143144145146147148149150
New comment