Chimera - Trending EA - based on Phoenix. - page 5
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Lets try this. What about that? This may work!
It looks like a bunch of chichens being chased by a fox. Every one is heading in different directions without any reasons.
You cant develope a trading system untill you write the plans first. Without one your just going to go around and around ending were you started. Kind like going on a road trip without the map or compass. Lord only knows were you'll end up.
One clear sign when system development is headed no were is when we hear,
Lets try this.
What about that?
This may work!
Etc..
Unless you just like to program and test every indicator you can find, I would seggest developing a game plan first. All those switches and inputs are not what will make a winning system.
By the way what happen to the Phoenix? It was a good system with a solid foundation. So it had a few problems that needed addressed. Why was it bulldosed over? These new systems should not be called Phoenix as they are only by name.
You all should pay more attention to people like davidke20 and all the others that have come and gone. You are not going to get any one that knows trading and system development to give you any help if you go around with your nose in the air and call these people names like you all been doing. I've been following this thread for some time and I think that you all will not be any closer to being done in a year from now.
I do not post much but this needed said. Please have respect for yourselfs and don't start calling me names too.
Mis B. Havin'
Bid in' every one a nice trade.
What next?
Very kind of you to offer your thoughts on the problem. I take full responsibility for the emergence of Chimera, which may or may not be a worthwhile project. I proposed to create a trending system based on Phoenix, and incorporating the signals from EASY. The idea was tolerated and even encouraged. Inevitably it would be criticized as well, something I hope that will prove beneficial.
If you have constructive suggestions, please do put them forward. Should we not use EASY signals? I find some of the Phoenix signals difficult to grasp, and it seemed easier to switch from countertrend to trend if you know what the signals are doing. (Programming the EASY signals raises its own challenges: using analogs to them may enable us to circumvent them.) Many of us are rank amateurs and could certainly benefit from your experience.
Trailing stops were never a central part of Phoenix. Everyone concurred that it would be a positive move to get them working smoothly. That has also been a focus of attention. Other exit techniques are also possible (see Phoenix 6).
Looking forward to hearing from you.
Filter for Chimera
Here is a bit of code I am trying to work up into a signal for Chimera. It is intended to prevent trades when the difference between the upper and lower Bollinger Bands is less than a certain number of pips. Since I do not have the programming skills to write exquisite code, I wonder if PContour or Daraknor would be kind enough to correct my errors. Perhaps this is even the wrong way to go about it. Thanks in advance.
double UpperBand = iBands(NULL,0,BandPeriod,BandStDev,0,BandPrice,MODE_UPPER,BandShift);
double LowerBand = iBands(NULL,0,BandPeriod,BandStDev,0,BandPrice,MODE_LOWER,BandShift);
if(U_UseSig6)
{
if(UpperBand - LowerBand > BandSpread) {BuySignal6 = true;}
if(UpperBand - LowerBand > BandSpread) {SellSignal6 = true;}
}
else
{
SellSignal6 =true;
BuySignal6 =true;
I find davidke's comments constructive and helpful. I wish I had more experience with trading and EAs to be able to answer his list of questions. We need more constructive criticisms such as posted by davidke.
Mate, I've never trusted EA to open buy stop/sell stop or buy limit/sell limit, unless its a trendline trader. There is few things I think you should start thinking off before we go furthur. I have found Metatrader to be very unreliable with pending trades in general. I prefer monitoring the open trades if we're waiting for a trade to close, then open the history file. If we're entering while another trade is still open, then I put the code inside the check managed. I wrote a pending trades system on contract, and I explained to my client how it wasn't very robust to do it this way. He went with it anyway, and there was an order issue about 1 out of 40 trades even after many hours of debugging. Letting the broker know when you enter the next trade also seems like a carrot on a stick. The brokers don't seem to mind cheating to get the carrot by adjusting prices. I do agree that direct entry is better.
FXDD has seemed to be an honest type trader in general. I still see their price adjustments on the history files vs chart files, but they are usually similar and there aren't many spikes, more price flattening. I also noticed the highs and lows slip up and down a few extra pips on the chart data. Compared to IBFX, I couldn't even match market data with chart data most of the time. :/
I was considering Oanda to write code for, but the entry price was a little steep for me. My two other options were MBTrading SDK and TradeBullet. I haven't played with TradeBullet yet, but it is broker independent. I'll reply to your bulleted list of questions separately.
Thanks very much for the fix on the parabolic signal PContour, I'll try to test it ASAP.
Meanwhile, I do believe that the TS setup may be misfiring. It appears that it kicks in as soon as the trade is 1 pip in profit, and immediately moves the TP to TS and sets SL to a new position. It then shifts the new SL-TP pair upwards one pip with each pip gained. Is this the way it is intended to work? And how do you activate the scalable TS?
We desperately need a mnaual for P6. MoveTPonTS is the number of pips the TP will get moved up each time the SL is adjusted. If this number is 0, the TP remains the same. Expanding trailing stops in this fashion tend to have double digits higher average profit, with no extra loss. If the pips are coming in, why quit when you're already covered? If ticks are rare or you trade during news, I recommend a value of 2, otherwise MoveTPonTS=1 is fine.
'scalable' has two methods. The first is a parameter not included in this version (I have it working elsewhere) where the trailing stop is configured as a percentage of profit in the init() section. Since we're recommending optimization with static TP & SL values to start with, the trailing stops can be configured by hand. The second type is the parabolic SAR function, which I haven't had time to test yet.
Answers to David's questions:
1) Whether we go with trend trading
Chimera is a trend follower. It really is a separate EA altogether, but we're using the same trailing stop code, order management code, etc. Chimera was initially a project by a non programmer attempting to implement signals from a manual into an EA. We're working together on making that happen.
2) Anti trend trading
Phoenix is counter trend system.
3) Robust with beginning of the trend(anti trend) to take advantage only on low target and low stop loss. Could take days to open 1 trade(less risk, low return)
Sound advice for a smooth equity curve. Upon first taking on the mantle of "code maintainer" I was disheartened that the system was completely unprofitable on some brokers. We also noticed the trades flipping back and forth with each ripple. The Consecutive Signals code was added by a member of ForexFactory, and I integrated it in with a low signal count with good results. Myself and others found that increasing the consecutive signals made up for other deficiencies in the signal entry, and the EA became more robust and profitable, but still fragile.
4) Trade expiration(even opened trade, not just pending order)
Grace/Force exit code is used.
5) How reliable the indicator is
The Phoenix (not Chimera) indicators appear to be curve fit toward a specific market. When the market type changes, they fall apart. I would rate the signal as not robust at all. On the other hand, it almost won a contest.
6) Will the volatility affect your indicator? Say, a spike could bring you into the trade
Not that I have seen in forward or backward testing. The main trade provocation is small ripples on the way down. A consolidated market is also traded, but dislike that type of entry since the signals didn't predict the break out direction. It may be possible to configure the settings improperly enough to trade spikes, but I haven't seen it.
7) Exit strategy? Target? Stop loss?Based on PIP's define by trader? Or robust by the system itself to judge? Indicators? Price action? Swing trading?
Phoenix has a TP and SL optimized for each currency pair, and each market type. This is extremely frustrating to me. I was considering making TP and SL numbers calculated from the last 3-5 top to bottom cycles.
8) Do we enter all the trades at the same leve? In that case, I will never go split order. Because when we die, we die togather, why not we win togather. That is not diversify the risk, but risking what you should made about.
The "mode 3" like behavior all used the same entry point. It may be possible that the trailing stop and breakeven code completely supercedes this type, but I have also had independent requests for this type of trading. I have nearly no opinion if it is wise at all. I tended to risk management by increasing the consecutive signals, decreasing the size of each trade, and then increasing the maxtrades parameter. This way I might enter later trades, but only if the signal is still valid. This also led to hedging if the TP/SL completely missed the immediate bump.
9) 2 level martingale? 1 trade not success, go for another 1 with double lot size(only if you're confidence with your indicator)
I only recommend doubling the lots if the second trade has a chance at profit on it's own.
10) When is the due date of this project? What is the objective? Who is the objective? Is the market majority using this EA, will it affect the price movement life cycle? If it is, then safe it. No point to carry on.
The objectives were stated in the main phoenix development thread. #1 I wanted to rewrite phoenix with the same functionality, but not restricted by mode. I also disliked the True/False grid for signals. #2 Add a trend filter so we don't fight strong trends anymore. #3 Write new exit strategies and tinker with different signals. #4 Add support and resist level awareness.
Chimera simply uses the same basic code for a new purpose.
We tried very hard to get the same signal when all other variables were identical. Since the broker prices are different, the system trades differently with different brokers. To my knowledge, only trends, fibonacci levels and tight channels are self reinforcing. I have been pondering the "emergent" vs "deterministic" question a lot, and I think Phoenix (and probably Chimera) are limited to measuring emergent behaviors. A side project of mine is working on deterministic behaviors.
11) Refer 10, is that every broker, every platform, every user are going to get the same signal. You buy, I buy, he buy, she buy, everybody buy. Or its based on the broker feed?
Broker feed. If we were trading with large numbers there might be issues if either A) we overload the broker's position, and they feel compelled to enter the market B) we collectively become a market maker by pushing billions of dollars through several brokers and islands. Consider the large number of people on both sides of the fence, I don't think we'll make much of a difference beyond the broker even if we tried. If we do make a difference with an unscrupulous broker, they wouldn't affect the market so much as alter their price feed to hit stops.
Here is a bit of code I am trying to work up into a signal for Chimera. It is intended to prevent trades when the difference between the upper and lower Bollinger Bands is less than a certain number of pips. Since I do not have the programming skills to write exquisite code, I wonder if PContour or Daraknor would be kind enough to correct my errors. Perhaps this is even the wrong way to go about it. Thanks in advance.
double UpperBand = iBands(NULL,0,BandPeriod,BandStDev,0,BandPrice,MODE_UPPER,BandShift);
double LowerBand = iBands(NULL,0,BandPeriod,BandStDev,0,BandPrice,MODE_LOWER,BandShift);
if(U_UseSig6)
{
if(UpperBand - LowerBand > BandSpread) {BuySignal6 = true;}
if(UpperBand - LowerBand > BandSpread) {SellSignal6 = true;}
}
else
{
SellSignal6 =true;
BuySignal6 =true;In Phoenix6 this would be a filter. What I didn't see in this P5 style code is how you would ever get a false value. When I troubleshoot, I try to see if my code is returning the right data, then I try to see if the code is doing the right thing based on that data. I didn't check your data, but I'd recommend setting a false value right after the subtraction in an else.
if(UpperBand - LowerBand > BandSpread)
{BuySignal6 = true;}
else BuySignal6=false;
if(UpperBand - LowerBand > BandSpread)
{SellSignal6 = true;}
else SellSignal6=false;
More on Chimera
Thanks very much for your very detailed response on Signal 6 in Chimera. I hope to be able to finalize the system with only one more signal.
At the moment I am focusing on 5.7.2 because it is what I am familiar with and because I believe it should not be left unfinished. If we have a version which is completely stable (right down to the TS!!) then we can release it for testing for profitability. I would prefer to wait until this is done before shifting my attention to P6Chimera.
You will find a question on TS in another thread, but in the meanwhile I am working up Signal 7, as follows (I don't think it executes properly). If it is possible for you to make it shipshape I would very much appreciate the effort. The idea again is to issue a false signal when the curve is flat so we don't engage trades that don't go anywhere, or else go where we don't want them to go. This is based on Daraknor's model for Signal 6. I suspect that both of them are misbehaving: they seem to reverse good buys and sells (making them bad) in addition to excluding them when the curve is flat.
Thanks.
#
//=====================SIGNAL7=======================
bool BuySignal7=false, SellSignal7=false;
double UpperEnvelope = iEnvelopes(NULL,0,P_EnvPeriod,P_EnvMethod,P_EnvShift,P_EnvPrice,P_EnvPercent,MODE_UPPER,0);
double LowerEnvelope = iEnvelopes(NULL,0,P_EnvPeriod,P_EnvMethod,P_EnvShift,P_EnvPrice,P_EnvPercent,MODE_LOWER,0);
double EnvPriceAv = (Low[0] + High[0] + Close[0])/3;
if(U_UseSig7)
{
if( EnvPriceAv < UpperEnvelope)
{BuySignal7 = true;}
else BuySignal7=false;
if(EnvPriceAv > LowerEnvelope)
{SellSignal7 = true;}
else SellSignal7=false; }
else
{
SellSignal7 =true;
BuySignal7 =true;
}
Chimera status
For the purposes of the new thread here is what's cookin.
1. My focus is ChimeraX which is based on P5.7.2W with signals switched to EASY and a couple of signal filters. (no original Phoenix filters).
2. The goal is to trade once or twice a day max, no losses.
3. This can only be done (I suspect) if TS moves SL to breakeven.
4. I will move to development of P6Chimera when and if ChimeraX is stable and there appears to be potential for improvement by going to P6Chimera, but I can do demo testing with P6Chimera when it is stable. In principle, the ChimeraX signals could be ported to P6Chimera.
At least this seems most appropriate given the current status of both EAs. How does that sound?