Art Phantom EA | Page 57 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Phantom 6 System 1,540 replies

Phantom 6 - Manual System 8 replies

Phantom Entry on ODL Japan. 2 replies

Tradeview CAD/JPY "Phantom" spike 0 replies

Phantom high, phantom low 0 replies

Attachments: Art Phantom EA
Exit Attachments

Art Phantom EA

  • #1,121
  • Dec 9, 2010 9:46am Dec 9, 2010 9:46am
Quoting roundrock
on this, i have one idea. right now biggest issue is the DD. sometimes when the price goes in a slowly expanding channel, then it keeps adding orders in both direction, but unable to move enough in any one direction to meet our PP

Can we limit the number of trades ? lets say we dont add beyond 10, then we will not add new trade, we can contain the DD. but it will reach PP only when it moves enough in one direction.

i am trying to see how we can avoid that death trade..
I was also thinking about this and also posted this idea before. but I was concern about the time to recover from dd. It may take few days... weeks or dont know.
  • #1,122
  • Dec 9, 2010 9:46am Dec 9, 2010 9:46am
roundrock, cailiou:

I have taken the ones at post 998 and testing. Thanks. Will publish the results once i have.
Intraday only.
  • #1,123
  • Dec 9, 2010 9:49am Dec 9, 2010 9:49am
Quoting CyberNerd
I've attached single_pair_phantom_pipstep-ts-be3-qqe.mq4. This has the following filters added; stoploss, trailingstop, break-even, and qqe.
Love to see the perforamance with qqe. i'm not testing this yet though since I'm testing the early version of SinglePair.

Since the EA can close 2-3 basket in 1 or 2 days, it is also good to test if we can close the basket at breakeven (or even at a small lost) after xxx hours of trading with assumption EA can make up the loss with 2-3 addition new basket close.
  • #1,124
  • Dec 9, 2010 9:57am Dec 9, 2010 9:57am
Quoting navk
roundrock,calliou;

Can you advise me which is last version of single pair EA? Thanks.
I am simultaneously testing both version (998 started today & 886 started from Monday), but surprising thing is that both are putting trades in opposite directions.

I am testing post 886 ea from last Monday on 15m, 30, 1hr tf.. results are fine for 30m tf and very good for 1hr tf but problem for 1hr tf is that sometime it stop trading or closing basket before hitting TP target. I restarted that ea 2 times since then.
I not able to find where is the problem..in ea or in my mt4
somebody please help me solve this............................
  • #1,125
  • Dec 9, 2010 10:00am Dec 9, 2010 10:00am
Well everybody here talking about single pair phantom... I am still optimistic about art phantom light.
Basic EA is not doing good with original pairs but its working absolutely fine with EUC.
  • #1,126
  • Dec 9, 2010 10:18am Dec 9, 2010 10:18am
2 more closures on M30 and M15, not sure about the best TF........
  • #1,127
  • Dec 9, 2010 11:20am Dec 9, 2010 11:20am
Quoting 24hrs
I am simultaneously testing both version (998 started today & 886 started from Monday), but surprising thing is that both are putting trades in opposite directions.

I am testing post 886 ea from last Monday on 15m, 30, 1hr tf.. results are fine for 30m tf and very good for 1hr tf but problem for 1hr tf is that sometime it stop trading or closing basket before hitting TP target....

dont test 886. always take latest 998
  • #1,128
  • Dec 9, 2010 11:27am Dec 9, 2010 11:27am
Quoting 24hrs
I was also thinking about this and also posted this idea before. but I was concern about the time to recover from dd. It may take few days... weeks or dont know.
problem with that approach is we may end in DEADLOCK. lets say we get 5 buys and 5 sells, then no more trades, this will result in DEADLOCK. It cant reach PP.

we need to get out with BE or slight loss. wait for few bars before we start new cycle

in short,

1. no more trades after total 12 trades
2. when we reach this state, change PP to slight loss and try to get out
3. after bail out, no trade till few bars (to wait for new range or breakout)
  • #1,129
  • Dec 9, 2010 11:44am Dec 9, 2010 11:44am
Quoting roundrock
problem with that approach is we may end in DEADLOCK. lets say we get 5 buys and 5 sells, then no more trades, this will result in DEADLOCK. It cant reach PP.

we need to get out with BE or slight loss. wait for few bars before we start new cycle

in short,

1. no more trades after total 12 trades
2. when we reach this state, change PP to slight loss and try to get out
3. after bail out, no trade till few bars (to wait for new range or breakout)
Absolutely agree, the problem with this strategy is not the possibility of blowing your account (of course, it could happen but you will have time to react before that), but to reach a situation with same buys and sells and not move for a loooooong time......

So Roundrock could you code your suggestion ? Maybe the "max total trades" before changing PP as an option ?

Again thank you very much
  • #1,130
  • Dec 9, 2010 12:10pm Dec 9, 2010 12:10pm
I've been following this thread and the two by metacoder. Congrats to everyone who contributed on developing this system. I would like to share the results of my testing.

I'm using the version of the single phantom ea I posted yesterday. All my backtests have been performed on metaquotes 90% data. Alpari UK is my broker. Unfortunately, thats the best I can offer. I've run all my backtests from Jan 1, 2008-to date. This takes forever, but fortunately I have a fast gaming system. if you are just testing a couple months, you don't see the whole picture. There are alot of death trades out there waiting.

In order to complete a 3 year back test, I had to start with a 10K balance and reduce the settings to 10% of what most of you are running; MMTU: 0.04 PP: 0.10. (starts trading with .02 lot size. Minimum account size for this would be 5000 trading .01) Also it needs a TF:1hr, I was unable to complete with the lower timeframes. With this setting it completes the three year run. I'm not really concerned about the profits at this point (3635.39). I want to see the danger areas and reduce the DD. this gives us 50% max DD and a PF of 1.05.

I then added the QQE filter. I believe this is hands down the best filter/improvement we have seen and should be added to all the versions of the ea's. I left the QQE default settings the same, but changed the timeframe to 240(4HR). After testing (30min/1hr/4hr/1day), this seems to be the optimal timeframe. Net Profit:4569.48, MaxDD: 29.77%, PF: 1.14

Next I included QQE and a stop loss. Optimal SL for my testing is 2250 (225 pips). Without QQE, optimal is 2500. Results are; NetProfit:6391.54, MaxDD:20.27%, PF: 1.25

One negative thing I've noticed about adding the SL. With the test using just QQE there was one big DD in 2008, then 2009 and 2010 were fairly smooth without any big DD's. After adding the SL, the 2008 DD was more than halved, but new variance/chopiness was added. This appears to be due to reversals after we cut losing trades that would have come back. There is probably a better solution than SL yet to be found.

I'm not completely finished testing with the other filters, but so far it appears that none of them add value that I can see. As has been previously reported, pipstep reduces the DD slightly, but greatly reduces the profit. I haven't tested any of the time filters yet.

Increasing the MMTU and PP, always results in eventual death trades. I know some of you are seeing great profits running MMTU:.4 and PP:1 (so am I), but history shows us these setting will eventually lose all. Please don't take this as negativity on my part. I am very positive about this EA, I just don't want anyone to jump in prematurely and lose real money with this. We need to find a way to dig our way out of the DD or another better filter to prevent it. This is getting very close.
Attached File(s)
File Type: zip 2008-TD-nofilters.zip   84 KB | 194 downloads
File Type: zip 2008-TD-QQE.zip   75 KB | 177 downloads
File Type: zip 2008-TD-QQE+SL.zip   142 KB | 201 downloads
  • #1,131
  • Dec 9, 2010 12:18pm Dec 9, 2010 12:18pm
Quoting CyberNerd
I've been following this thread and the two by metacoder....
Thank you very much CyberNerd,

The more I think about it, the more convinced I am that Roundrockīs mod could avoid this big DDs.

I think too we are close to reduce this situation.....
  • #1,132
  • Dec 9, 2010 12:21pm Dec 9, 2010 12:21pm
EUR/USD, M15, 10.000.-$, 500:1, MMTU: 0.4, PP: 0.5, started test 5 hours ago, max DD: -0.63%

If we could backtest with lower TF than 1 hour, is it possible ?
Attached Image (click to enlarge)
Click to Enlarge Name: M15.jpg Size: 69 KB
  • #1,133
  • Dec 9, 2010 12:37pm Dec 9, 2010 12:37pm
point me to the exact ea (too many versions running around) and the timeframe you want tested, and i'll run it for u

really shouldn't get overly excited about results after 5 hrs tho
even results after a week or two, really don't tell you much
  • #1,134
  • Dec 9, 2010 1:05pm Dec 9, 2010 1:05pm
Quoting tazmaan
point me to the exact ea (too many versions running around) and the timeframe you want tested, and i'll run it for u

really shouldn't get overly excited about results after 5 hrs tho
even results after a week or two, really don't tell you much
I know, I know but if I donīt get excited now, maybe I will not get later, so I like to live the present, just in case the future is worse.

Single_Pair_Phantom from post 998 (Pipstep 0 and the rest of filters false), EUR/USD, M15 (and M30 if is not too much), MMTU: 0.4, PP:0.5, 10.000.-$, 500:1

Thank you very much Tazmaan
  • #1,135
  • Dec 9, 2010 1:16pm Dec 9, 2010 1:16pm
Without filters, I wasn't able to complete a 3 year run on 15 or 30min timeframes. I just tried again with QQE+SL filters. 15min still dies at the end of year1, 30min makes it all the way. Net Profit:8047.45, MaxDD: 27.07%, PF: 1.25. Profit was a litle more with same PF, but DD is higher.
Attached File(s)
File Type: zip 2008-TD-QQE+SL-15min.zip   112 KB | 189 downloads
File Type: zip 2008-TD-QQE+SL-30min.zip   176 KB | 191 downloads
  • #1,136
  • Dec 9, 2010 1:30pm Dec 9, 2010 1:30pm
CyberNerd,

Thanks a million times for your extensive backtesting session. Great job!

Interestingly bigger timeframes work better, so H4 is the optimum according to your backtests. I think using the H4 for determine the trend and using a M15 for entry (that has to match the trend) should improve results. Unfortunately you can't backtest it any longer (not MTF support in MT4...).

Since it really looks like QQE is a solution to get a "smoother ride", I'll try to do some more enhancements...

Cheers
Andy
  • #1,137
  • Dec 9, 2010 1:53pm Dec 9, 2010 1:53pm
that's not a problem in MT4.

H4 is 60 minute x 4 = 240 minute.

if you use a EMA(5) for example on H4 timeframe, it is same thing for you to use EMA(80) on M15 TF.

Quoting hopfi2k
I think using the H4 for determine the trend and using a M15 for entry (that has to match the trend) should improve results. Unfortunately you can't backtest it any longer (not MTF support in MT4...).
  • #1,138
  • Dec 9, 2010 2:01pm Dec 9, 2010 2:01pm
Great analysis, and contributions.

Backtests are always painful to run and summarize.

there might be one alternative to SL in pips, if you are looking at some other possibilities.

maybe, SL in hours might be a way.

if a basket has been opened for more than 72 hours, exit it!

or Roundrock (or who) suggests a wonderful idea: after 12(or you name it) trades, exit it.

when we live in the "PRESENT", we will not be able to run away from "DEATH TRADES" if we do not want to take a LOSS.

Quoting CyberNerd

One negative thing I've noticed about adding the SL. With the test using just QQE there was one big DD in 2008, then 2009 and 2010 were fairly smooth without any big DD's. After adding the SL, the 2008 DD was more than halved, but new variance/chopiness was added. This appears to be due to reversals after we cut losing trades that would have come back. There is probably a better solution than SL yet to be found.
  • #1,139
  • Dec 9, 2010 2:04pm Dec 9, 2010 2:04pm
Quoting charvo
Great analysis, and contributions.

Backtests are always painful to run and summarize.

there might be one alternative to SL in pips, if you are looking at some other possibilities.

maybe, SL in hours might be a way.

if a basket has been opened for more than 72 hours, exit it!

or Roundrock (or who) suggests a wonderful idea: after 12(or you name it) trades, exit it.

when we live in the "PRESENT", we will not be able to run away from "DEATH TRADES" if we do not want to take a LOSS.
  • #1,140
  • Dec 9, 2010 2:11pm Dec 9, 2010 2:11pm
Untill now, art_phantom_light has closed four baskets, the max DD was only - 1.54 %.
Intraday only.
1 trader viewing now