PacMan (my first hybrid) - page 3
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
Come on, please read first everything about Goblin Bipolar, 10 point 3 etc. to learn how Martingale type EAs work: After a small move against them, they increase risk and success probability such that a smaller move in the right direction gets the desired profit. They risk, however, that the small adverse moves add up to a big one. Depending on your configuration, you either use s/l and eat the loss, or you go until the limits of margin etc. and risk account busting.
Did you fully read my post before blaming be that I don't know how these EAs work? Please read it again. I want to enter the market in a different currency instead of buying the same currency again and again to reduce the risk of a sudden "killer move" ... any suggestions on that idea? :-)
...
What I'm talking about is a simple daisy-chain:
step 1: buy 1 lot of eur/usd
step 2: buy/sell 2 lots of eur/chf
step 3: buy/sell 4 lots of eur/gbp
etc...
using this method, chances, that a strong killer-trend on our main-currency is going on which will wipe us, are reduced alot. Of course the lot-sizes above need to be ajusted so that the losses of trade n-1 are fully compensated after TP.
...
Did you fully read my post before blaming be that I don't know how these EAs work? Please read it again. I want to enter the market in a different currency instead of buying the same currency again and again to reduce the risk of a sudden "killer move" ... any suggestions on that idea? :-)Sorry for the offense, it simply seemed heavily off-topic ...
Ok, taking the question seriously now:
I don't see how switching currencies will give you an advantage, you can have the same problem again just at a different place.
One may treat it as a form of hedging or a Martingale at the trade instead of the order level: After a loser you try to regain the loss more quickly by increasing your stakes, this time with a different currency. This only works if you know that the chances of your next trade(s) after a loser are better and that you are increasing chances more than risk. This is called a (gambling) progression.
I tried to apply it in Goblin Bipolar, it works to some extent, but since risk is actually increased, the equity curve fluctuates too much.
Note that the Martingale approach can only work (over the long run), if at the same time that you increase your risk, you also increase your probability of succeeding, and if your entry signal is worth more than a random entry (you have an edge). In all other circumstances you fall back to well-known results from gambling.
Please elaborate more about your idea of switching currencies for the next order.
Yea,thats how it should read.
Volume can be very usefull. It indicates market activity is picking up. If you set volume too high it won't trade. So some one can select there desired level of market activity befor a trade is entered. This will keep it from placing trades on false signals. As a trend comes to an end the markets volume will decrease(I have to test what setting is best) But when it decreases at the end of a trend, it should help keep it from placing that final order that gets left behind when the trend reverses.
I had added the ATR thing. But in back tests were super slow with ATR D1 initialized. I'll do that one manually for now.
DaveFor me, light volume is not necessarily a disadvantage. Note that Goblin Bipolar works very well with random walks.
Backtesting your latest version, the old weakness of the strategy when s/l is not used shows up again. For 2006 I get several heavy drawdowns because it sits out a large adverse move and ends the year with a loss.
I also noticed that the first version closed most of the trades early, whereas the second version now always t/ps, there must be a change in settings too.
What are your ideas regarding reducing drawdowns?
Alassio, yes I got your point but of course I'm not talking about "randomly" entering a different currency in step n+1. Instead I plan to wait for the next "good" moment and enter it then.
The advantage I think of this method is the fact that - after observing the backtest-results from Goblin and Predator - most of these progressing systems die after a strong trend in the wrong direction. The EA will pump more and more cash into the trade in the hope of a counter-move which won't come and ruin us. This is in fact the worst case scenario on all those martingale systems. In my theory this risk can be reduced by distributing the trades to a few currencies. Even if one of those has a strong trend against us chances are high that the other currencies don't behave the same way at this moment.
David, what do you think about this approach? Thanks
Great Job
Thanks for starting the PacMan, Goblin Bipolar Hybrid, Dave.
Actually, I just found this thread today. Been off from forum recently :P
I would like to have a look at the code and start testing it.
As in my attachment, Goblin and Bipolar give pretty good result so far.
Keep up good work mate, no need to listen to "always negative" post
Cheers,
And you have nothing at all to add but your foolish noobie mouth.
Too bad I live on the east coast. May be I'll take a vacation
"""Anyway you hardly change anything of original BiPolar so a new name not worthy. Ripping off bluto again."""\
I changed the signals, added filters, took out signals, added multiple lot size settings. And am working on more.
Mabe some one needs to tell this kid what the definition of "Hybrid" is....
And some one needs to tell him all the goblins are 10 points 3 hybrids.
The reason most of the original code is there is because its the martingale system and order send functions. No need to make changes or to re-write it.
For a guy that suppose to be in college and in c++ classes, you sure are dinse.(that means lacking of common sense)Yo man yeah you do need good vacation. Away from posting ur junk here and from fx trading. lol. And yes I am a great C++ programmer and big difference between hack like you and me is I know proper development, testing and debug cycles for code before putting it to users so they dont trip all over it. Amatueur like you put ur junk here and want everybody to fix it. Did you even get it to work before you post it here. No I dont think so. The code doesnt even make sense what ur trying to do. Dumb ea dumb dumb dumb. You just try to be a big shot again I know.
Alassio, yes I got your point but of course I'm not talking about "randomly" entering a different currency in step n+1. Instead I plan to wait for the next "good" moment and enter it then.
What are you doing with your order n in the other pair?
The advantage I think of this method is the fact that - after observing the backtest-results from Goblin and Predator - most of these progressing systems die after a strong trend in the wrong direction. ...
You only die if you never admit you're wrong and try to sit out every move. This is a good method to generate good looking backtest results, just select the right period and the right settings ...
I don't know why everbody tries running without stops, at some point it just doesn't make sense trying to sit out a position you can't afford. My recipe is:
- Use stops, in Martingale approach use higher orders to get out with decent loss
- Optimize entries to get a high enough winner rate (note that Goblin Bipolar can have 95% winners), the entry gives you the edge
- Optimize lot sizing among orders to maximize expected profit per trade
- Have a relatively high trade rate (a few trades per day is fine)
- Scale lot sizes with account equity for geometric growth
I am following this thread because I am curious what ideas other people are following, especially concerning entries/filters. A lot of this stuff can be found in the original Goblin Bipolar thread.
What are you doing with your order n in the other pair?
Simple: I keep my other order open of course. The method is simple: The risk that a currency-pair is on a strong trend against us is split by using multiple pairs with completely other charts... take a look at some Backtest-results from Goblin, Predator, 10point3 and whatever... they lose exactly in those situations. But if the cash is split on 3 currencies chances that such a situation occurs are just lower.
You only die if you never admit you're wrong and try to sit out every move.
I fully agree on that one. Most systems however have a "too-good-to-be-true" winning percentage because they sit out ALOT. Its really hard to select appropiate settings here. I guess the sharpe ratio of the system should be used to optimize settings here... however "optimizing" is nothing else than storing future knowledge in those settings to produce nice equity curves... so we have to be careful with that one too.
Optimize entries to get a high enough winner rate (note that Goblin Bipolar can have 95% winners)
You're right but those 5% are that bad that they are going to kill you in the wrong market situation. Thats why I was suggesting the above
- Optimize lot sizing among orders to maximize expected profit per trade
Your're right. The real question is how much!
- Scale lot sizes with account equity for geometric growth
Problem: 1 "really bad" trade kills all profit if no tight SL is used
This EA looks really promising so far... I hope it can be improved further.
My recipe is:
- Use stops, in Martingale approach use higher orders to get out with decent loss
- Optimize entries to get a high enough winner rate (note that Goblin Bipolar can have 95% winners), the entry gives you the edge
- Optimize lot sizing among orders to maximize expected profit per trade
- Have a relatively high trade rate (a few trades per day is fine)
- Scale lot sizes with account equity for geometric growth
I am following this thread because I am curious what ideas other people are following, especially concerning entries/filters. A lot of this stuff can be found in the original Goblin Bipolar thread.Correct summarising. Theoretically I agree with all of them. In practice (mainly testing) I have not find a setting that would have worked profitably with stops (probably it is my problem because I was not able to find the right setting).
I would add two points to the list though:
- Make possible to run the trade when it is in a favourable trend
- Hedge the trade to breakeven (or near) then cut it out if it is in a trend against you
Simple: I keep my other order open of course. ...
Bad idea, you still have unlimited risk and don't increase your win probability by following regular Goblin strategy.
...But if the cash is split on 3 currencies chances that such a situation occurs are just lower.
This is a good idea, but you're talking about diversification here, not about improving individual trades. Choose the right pairs and trade them only when the ATR is compatible to your tests.
... ...I guess the sharpe ratio of the system should be used to optimize settings here...
This would be a fair optimization criterion. I use drawdown and loser percentage as approximation, given similar trade rate and profit. But avoid overfitting ...
...
Problem: 1 "really bad" trade kills all profit if no tight SL is used
...No, use SL then you know how much you will lose. You simply need more winners until the next big loser.
Note that Goblin uses always very lose stops (unless maxtrades is small):
E.g. for maxtrades=10, pips=5,takeprofit=10,stoploss=10 you need a 10-14 pips move ANYWHERE in a range of 65 pips until SL hits you. By the way, the goal is not necessarily to wait until SL, the last few orders may be there to reduce loss only, they are not worth forcing them into winners.