IBFX 50:1 demo running untouched since 12 Dec. Phantom6_01.mq4 from the 1st post in the thread. Full statement and settings file attached.
Ignored
great post !!! excellent feedback. we have discussed so many pages and so many months figuring out the right strategy to arrive at right risk/reward whereas you have stuck to the original post and achieved great results. sometimes we overcomplicate the things. the key is to have low lotsize and high account size. in this case 0.01 lot for 3k account.
........ the key is to have low lotsize and high account size. in this case 0.01 lot for 3k account.
Ignored
Up until I started my test I would have probably agreed with you, now I am not so sure with this particular ea
The large leverage account I am using was not intentional The large MMTU was the result of following Nicastro’ s method of testing extremes first I had expected the ea to blow the account away very quickly with this configuration Yet it continues to do the opposite
There have been a few baskets that have reached the 250% Margin Level and I have closed trades manually exactly as per my test method, which has worked very well This is where my equity curve is not as smooth as TheMaxx example But it is still not a bad curve, and the growth is impressive As of course might be the fall if it happens
I would never have thought to consider such a configuration with my ‘ Live ‘ hat on ( and I still do not ... yet at least ) We have it preached to us that large leverage accounts with small account balances can be killers if misused in this way
But it now has me asking is this very profitable test ( so far ) a result of just pure luck, in that the ea has not had to deal with an extended DD Or Is the configuration actually ideally suited to this ea and it is benefiting from the ( relatively ) large lot sizing, requiring only a small movement in price action to reach the PP target which is resulting in baskets closing quickly and efficiently, and therefore ‘ avoiding ‘ that extended DD altogether ?
Joined Dec 2006
|
Status: Backtest is meaningless (to me)
|2,175 Posts
it is pure luck, and let's discuss why it is:
it is very simple probability:
when the EA enters the market, there will be a few opened positions,
assuming that it is RIGHT NOW, what will happen next?
two outcomes:
1. reach x% profit, and a basket is closed (of coz, small x% means fast closure and vice versa).
2. reach death basket that consumes all margin, and you either have to close biggest loser early or close all finally.
now, for 1st outcome, it is relatively easier to happen, since the conditions are easier to meet----just x% profit. HOWEVER, the UPSIDE is limited, at most, you get x% profit (although you may get many times).
for 2nd outcome, the probability to a death basket eating all margin is always existing. and if the probability exists, it'll happen. If it happen, it will cause a loss of, perhaps, 20% capital, or 60% capital, or everything. The higher lotsize you use, the more you will lose when death basket comes. Achab's tests present a few examples.
This EA seems to allow its user to earn enough outcome 1 before running into outcome 2. we should remind of ourselves not to run the opposite way by allowing too much "risk" ( in other words, outcome 2 probability).
you have seasoned sanity NOT use the setting for LIVE. this is great, and it is always my belief that demoing should follow the way you trade LIVE, instead of starting from so called aggressive "let's see how terribly great this EA can be" testing.
Charvo is right. With lots of leverage and trades and small profit percent it is easy to make something which seemingly looks better than it really is.
Always think: what are the hidden risks? We need to make it safer and maintain the profitability.
(this applies to all methods, not just this one, but is particularly important for one like this where the number of open trades is not easily bounded).
Suggestion:
What we need is a day-by-day or hour-by-hour measure of running instantaneous equity. This could be a separate EA writing to a file (and flushing the buffers).
This would more accurately show the risk involved in getting any sequence of clsoed trades, as the closed trades only (until a blowup) can be misleading.
We want a positive return, but more importantly never have too bad an open equity drawdown.
two outcomes:
1. reach x% profit, and a basket is closed (of coz, small x% means fast closure and vice versa).
2. reach death basket that consumes all margin, and you either have to close biggest loser early or close all finally.
Ignored
In point no 2 you are of course assuming the account runs out of margin
The very test I am doing is to try and prevent that very senario .... it may fail to work of course like most things
In point no 2 you are of course assuming the account runs out of margin
The very test I am doing is to try and prevent that very senario .... it may fail to work of course like most things
Ignored
I agree - the only reason achab's tests blew up is he ran out of margin. If you keep your MMTU at an acceptable level to your account size, while you may have to close trades in your basket (some, or all) you won't blow up.
When I've gotten close to a margin call the drawdown in the account was only 1 to 5%, sometimes it was slightly above flat. That meant if I'd closed the basket I wouldn't have suffered much of a loss.
I agree it would not blow up ; but it certainly can take a long time to make a return. You can always add equity for the margin.... I run probably 5-6 versions live and I would rather cut a winner at a certain point when a large number of trades 20+ trades are open --and the winner is close to the biggest loser. The market kind of ranges at times at the end of day so the ea just sits there not making money. Close trades after afternoon or late in day when us closes and then let the market pick up again moving the trades in your direction when it starts to move. Just my viewpoint. I think it can be a good stable return if trades are managed right...however, everyone has a different opinion on how to do that.
My demo closed again Friday and had 2 recovery trades. One was cut when it reached CLM x -PP. The other was opened during the equity pop closure and closed at a loss which took 3/4 of the profit.
So far I am totally unimpressed with the recovery logic. Every recovery trade in my demo has either been cut as a big loser or has opened and closed during the equity pop and caused a loss.
IMO the success of this EA is the pairs that are traded and the Cut Loser option. It has nothing to do with the recovery strategy.
i fixed the use trading hours issue. you can specify the phantom start time, trading start time , end time. at session close if no trades then it goes to sleep mode till next day session. at session close if there are trades, you have a smaller pp, ea tries to close the existing basket at this smaller PP and after the basket is closed, goes to sleep
times are based on yr broker, pl change. mine is GMT-5
Ignored
Hi Roundrock.
Thanks for this. Your generosity knows no bounds!
What is happening to coding your idea as set out your post 767. I’m looking forward to testing it.
Of late I have been thinking of ways to emulate JCS’s performance but with a reduced MMTU while keeping Margin at a respectable level of at least 300%.
I think this can be partially achieved by adding a third synthetically hedged sub-basket. Sub-basket C consisting of Buy EURCAD and SELL USDCAD and SELL EURUSD. But as we already have a sell E/U in Sub-basket A, Sub-basket C will not be placing a second E/U sell trade when EU is a winning phantom (and existing EU trades are in profit).
I’d love to be able to test this variant and compare and contrast it with the existing tests to see how much quicker the baskets close and how margin level is affected. I think with this variant we will be able to reduce MMTU to 1% and still have a respectable number of basket pops per week without reducing Margin to anything below 300%.
We can but only test. On this, Roundrock, I ask for your help once more in coding the addition of Sub-Basket C into the EA which I believe will benefit all of us at FF.
Because of the market closure, during the weekend I use to think instead of other stuff, here.
I thought about something you wrote here, and I have read it this morning.
First of all, a special thank to "charvo" because his syntetic, complete and clear point of view is the starting point of my post, today.
I agree completely with him because this/these EA(s) will blow up, in future with any setup. The problem is to understand when it will happen: before or after that specific EA has gained 100%?
I make a step behind. Maybe someone remembers when I wrote (talking about Phantom6_01xR) something like "MMTU 2.1 isn't too much, even if for a test"?
Ok, I started such kind of tests.
Some blows up, some not. With less MMTU we would have more that don't blow up. In few days...!
But after 3, 4, 5, 7, 10 months? Probably yes.
I liked the Pete's test (TheMaxx) not only because of the results (very positive and with very low DD) but because the lotsize were always the same! He didn't increase the risk, even if the growing curve he got is amazing.
Using this strategy he was:
Lucky at the beguinning, becase the EA worked properly starting from a positive moment of the market;
Intelligent because by one side he didn't increase the lot size, by the other side the risk of ruin decreased, basket after basket;
Unselfish because the target was (and is) reachable.
Now, the question is "how much we would like to gain"?
Yes, I know it's an answer based on how much you have in your balance, but this is the reason why around 90% of Forex investors loose their money.
We have already understood that it's not how much money I want or need, but how much percentage is my target. So, which are your targets in %?
I'd like everyone answers on this, because only after we deside an objective, our tests will have some sense.
Because of the intrinsic method, soul of these EAs, I think that a percentage around 15-20% per month is a target I would try to gain if in live trading. It means an annual average return between 180% and 240%.
And if the EA will blow up? No, I correct myself: "And when the EA will blow up my account?". I will bet on it, in the sense that it shouldn't blow up before the 7th month. It means I will not loose money, but almost all of my gains.
After so many tests and after your suggestions and opinions, I think it is possible.
There is a correlation between MMTU, PP and average monthy/annual gain, and I want to find it before to start live trading "seriuosly".
I wait for your points of view.
/achab
P.S.: I have started my 16th test, this morning. It's about 3 Alpari-UK accounts, with a starting balance of EUR 7,500, Leverage 500:1 with these settings:
The "hight" MMTU setting is wanted, because I believe that a fast closing of a little basket is the best way to make money and to save the margin, well balancing risk and profit.
Two of the three (the first 2 I started in early morning) have already close one basket.
I have started my 16th test, this morning. It's about 4 Alpari-UK accounts (I added one since my previous post), with a starting balance of EUR 7,500, Leverage 500:1 with these settings:
Joined Jul 2009
|
Status: Trade. Review. Improve
|1,088 Posts
Great posts as always Achab. I will continue to follow your tests with interests. I've actually decided to jump in, and started trading live with the exact same settings and account size as my demo account. I'm also conducting a few other demos (with smaller PPs with the idea that the baskets are closed quicker, leading to profits being accumulated faster) but I'm happy with the risk, drawdown and gain of the 50:1 demo.
PS the reason that the lotsize didn't increase in my demo, is because of the low MMTU (0.5). I didn't increase the lotsize because it was all calculated by the EA. I let everything run untouched
I decided to start another test using Phantom6_01xR, on the same computer, with same settings. It was February the 9th and I opened a Demo account with Alpari-UK (leverage 500:1), initial balance 7,500 EUR, with those settings:
After 5 days, today, here you have the results:
P/L = +3,460.82 (+46.14%)
Lowest Margin: 21.73%
Weekly return +60.7%
Monthly return +633.0%
Profit factor 1.31
History 6 days
First amount of opened lot: 2.25
Because of the last weekend posts on this thread and the lowest Margin reached (but Alpari-UK didn't sell because it didn't reach its margin call) and a bit of luck too, I decided to decrease some settings such as MMTU=1.1 and PP=.020. The next opening lots will be 0.60.
The live stats are reachable at the same URL, and you could find the new trades because I have changed the Magic Number (now 7656212).
Because of the market closure, during the weekend I use to think instead of other stuff, here.
I thought about something you wrote here, and I have read it this morning.
First of all, a special thank to "charvo" because his syntetic, complete and clear point of view is the starting point of my post, today.
I agree completely with him because this/these EA(s) will blow up, in future with any setup. The problem is to understand when it will happen: before or after that specific EA has gained 100%?
I make...
Ignored
Hi Achab,
As someone who has been testing for quite a while now, let me chime in on some of your thoughts.
I demo a 3k account trading .01 lots, so you cant get any more conservative than that. After 3 months my account was up 19%. Last 2 weeks I hit a killer basket that has now brought my equity down to 9%. So lost almost half of profits. This EA has been a shooting star until.... so dont get lulled into a false sense of security by a few good weeks or even a month or 2. By the way I use a profit pct of .30 so I always went for the quick basket closures. If I used a higher MMTU with that small a PP maybe would have been better? I dont know, was always worried about big drawdown so I kept the lots at minimum. Maybe your idea has merit (High MMTU, Low PP). I will be interested in following your tests.
So now the question for me is: Is three months of good performance and one bad basket the norm? Or was I very lucky to go that long?
See the currency pairs what he is using. There is a difference.
Ignored
Actually, he is using the same six as me. And I disagree, in the long run, I dont believe it matters what set of 6 pairs are used, the basic logic of the EA is the same. For a given time period a certain set may work better than another, but how do you know which one?
Maybe conservative settings are not the best settings for these EA´s.....
If a death basket will come always (I´m not sure about that), maybe if after x numbers of trades or x numbers of hours the PP is not reached, the basket must be closed, no matter the result.
As someone who has been testing for quite a while now, let me chime in on some of your thoughts.
I demo a 3k account trading .01 lots, so you cant get any more conservative than that. After 3 months my account was up 19%. Last 2 weeks I hit a killer basket that has now brought my equity down to 9%. So lost almost half of profits. This EA has been a shooting star until.... so dont get lulled into a false sense of security by a few good weeks or even a month or 2. By the way I use a profit pct of .30 so I always went for the quick...
Ignored
Thorn - what MMTU and leverage do you use in your testing?
