Triangular Arbitrage

Kurka,

This is what happened to my test ea monday. I got a nice down trend but the entries generated by the "equilibrium rule" were right before a retrace. I've seen you haven't used a sl so you got the whole trend but i used one and i got a kick .
Shouldn't we use a sl? After all using fundamentals for trend determination is just a speculation and trading without sl could result in a disaster if trend reverses. Take your picture for example: after you entered those sell positions the price went up and reversed after a fiew bars to the trend direction, but if trend wouldn't be reversed there, you would have destroied that account. (a lot of sells in an up trend)

Another thing which i've seen my ea does and have also seen yours seems to do also: if i get a signal 3 times in a second the ea enters 3 times, is that wise?(basicaly we average down if doing so and increase the risk) If we aim for 1 pip than that should be ok but taking into consideration that no sl used than it's not. I will try to test it with 1 trade/bar if signal there and find a sl that makes me happy. I want to see the same hit ratio with a relative small sl so that i can calculate the RR.

Ex: 5 pip sl, 1 pip tp and winloss ratio 10/1 or something like that, you know what i mean....


Edit: I've also seen that sometimes when i get a desequilibrium and enter, my position will close in 1 second or so, price jumps back to equilibrium in no time. (bank manipulation/correction i think) but sometimes not. What could be the reason of that? Sometimes they correct the price and sometimes not...?
Thanks, Victor

This is "the kick" i was speaking about:

The red bar shows where the positions were opened and after that you can see a retrace until high 1.6117 but my SL was at 1.6112, 10 pips higher than the entered position....
I played the same steup but on EURCHF m1. I thinking of increasing the SL, 10pips is too small anyway or use an opposite signal to exit on both lossing position/winning one. But opposite signal taking doesn't reflect the equilibration,desequilibration theory...

Kurka how do you close your positions i've seen they don't close at the same time as it should if you would follow an opposite signal. How will your positions close in case of loss? (opposite signal?)

Image attached.
Thanks, Victor

Victor,

I am wondering about orders starting from below the blue 11633411... all six orders were opened at 16:00..... only once above this does the statement show this happen and that only shows two orders at 11:10 ....
Can you explain this?

Thanks.

Bruce

If you are interested, I can give you an access to an FTP server, so you can post the result with a hot link.
[color=#0000ff][email protected][/color]
Regards,

brucech,

It opened 6 orders beause it got 6 signals one after another in on the same bar. I explained in my previous message that this could happen and that this seems to hapen in kurka's ea too, i think.

Thanks, Victor

Hmm,

In this picture you can see what trades occured today on EURCHF m1. I got 4 buys but right after the signal i a retrace was recorded, what could that mean? Mabe i'm going in the wrong direction and we could exploit the retrace instead of the move itself..

The buys are market in the picture by 4 vertical green lines.

Thanks, Victor

Victor,

You mean during this trading period the EA only found 4 desequilibration buy signals? And that after each one of those 4 trades from the 4 desequilibration buy signals, that there was a retrace? hmmmm.... From the entry to the bottom of each retrace was about how many pips? Then maybe the EA could put in a pending buy X pips from the signal so it would be within the retrace pip total.... Just a thought...

Bruce

Zmax,

I forgot to ask... Could you give us examples of desequilibration consensus that the EA took trades on those buys?

Bruce

Zmax,

As far as the entry's go for testing I am allowing a new position every time I get a signal. I want to see exactly how many signals I receive so that I can determine the best way to eliminate the bad ones. I will pventually change this to a stepped up averaging rule, allow 3-6 positions in one direction at different levels - say 3 pips apart.

Exits are now based on 1) end of a trend 2) scalping rules, 1-3 pips TP 3) support and resistance levels... I abandoned the reverse signal theory a while ago. I dont know why all my positions arent closing at the same time. Do you know how to make this happen (close multiple orders at once).

Also I am using a stop loss now based on a 4 hour support and resistance level. usually 50 or so pips.

Im also finding it allot more important to "weight" certain validation currencies during different times of day. (still trying to get the stck index indicator to work right)

brucech,

Yeah only 4 desequilibration for eurchf in about 12 hours of testing today. The desequilibrations are marked with green lines in the pictures i've uploaded before. EA entered based on kurka's formula and inverted kurka formula for the inverted pairs.

Placing a pending order i don't think it's a good idea because i don't know how big the retrace is and if i would know that it happens always i will trade the retrace always and also trade the main direction at the same time.

Kurka,
Lol i tryed support-resistance for target but don't know what to say... I've seen that a good target could be for buy: break of "blue line" support from support-resistance indicator attached and for a sell position: break of resistance line from the indicator. Anyway i've also seen positions stoped early with fiew pips from a large move, so will test more. Mabe a s/r on higher tf.

Thanks, Victor

For closing all orders at the same time you can: move for all those orders sl to the same place. If price hits sl, all will be closed at the same time.
Using closeOrders(...) you can't do that.

I was thinking of moving the sl under suppor/resistance of that indicator i attached earlier. If the line is broken all the orders are stoped at the same time...


Thanks, Victor

Kurka,

There is some good info here that might help with weighting certain pairs during different sessions...

http://www.cyrox.com/forum/index.php?topic=25.0

Most of the data is later in the thread, posted by "coolo".

Ok so I changed the algos to calculate the consensus pricing based on MA's instead of real pricing. (High+Spread for Ask and Low for Bid) This basically widens the gap between overbought and oversold equilibrium pricing by a few pips resulting in fewer more profitable and less risky trades. Here is a screen shot of the live testing for the past few hours on EURUSD...

!!!!!!!!!!!!!!!!!!!!!!!!!

Hi Kurka Fund,

I have heard people talk about different price feeds for live and demo accounts. I was wondering if you had tried this method on your live account and if not are you anticipating any problems.

Kurka,

I don't understand how you get so manny trade opportunities in a day or so. I get 4-8 day and no more. Will add today the eurusd pair to see if that one is much more desequilibrated than my eurchf.

Thanks, Victor

Hi Zmax,

If you look back to post #35 Kurka explains the steps he takes to get trade signals. He mentions calculating correlations and indexes. I think he uses these values along with the triangular relationships to place trades. I think its a very complicated process.....

Do you have any idea what those steps involve?

Cheers
Gosko

Gosko,


Yeah i know how he corelates the currencies and also how to enter/exit based on that. Indexes are used for main trend determination i think, i don't use that because i believe it's speculation to say that if dow jones goes down - dollar will do the same for x minutes/hours from now.
Anyway i use it for eurchf not eurusd, mabe eurusd generates more signals. By 6-8 signals/day i mean 6-8 normal entries not how manny times the ea will enter. My ea enters about 50/60 positions /day but there are only 6-8 different signals. His ea does the same i think, read the page before thisone. I will make mine to enter only once /bar i think it's better so.
That corelation 35 to 36 for me is 1 to 4 because i use only 1 main currency and correlate with 4. He uses 35 and corelates to whatever the matches.
Thanks, Victor

Hi Zmax,
I suspect that he's getting more signals because he's using "35 to 36".
I don't have direct experience with these consensus calculations in an EA, but I've looked at a number of multipair measurements and I have the sense that those extra inputs are giving his system more actionable data to work with. I don't know if that could account for the entire difference in signal frequency, but it could be part of it.

Another thing that comes to mind is if you are using standard deviation(s) of the spread as a filter as Kurkafund mentioned, perhaps using a smaller amount, ie., even a fraction of stddev, it would let more signals get through. (But it could also reduce your signal quality too...)

Good luck!

Hi,

This is what the ea did today. The green vertical lines show where the ea opened (4) buy positions. The yellow dots showed where those positions were closed. (position 1: 2pips, - position2: 7pips, -position3 10pips)
I use as stop for an order the support/resistance of the support resistance indicator i posted yesterday. The last position was weird, it was possible to take that scalping pip because the market was flat...but where could the ea know that thing?
So it didn't take the point (at that white circle in the picture) and currently is -18 pips. The initial sl used is the resistance/support for h1 but i think i should change it to m1 because we acctualy try to hit a fiew pips and having a big sl is not a good idea i think.
Also the ea opens only 1 position/bar.


Technical,

I have a fiew positions because i use 1 position /bar not all positions that the ea will open if each signal would be taken. Spread is included because i compare bid with ask, and ask with bid so no problems with that. I do the calculations for eurchf and all the composite pairs that can create eurchf so i think my calculations are right i think.

Now what i've discovered so far about this subject: Sometimes the price goes direcly up if the consensus is up, sometimes makes a small retrace and goes up, sometimes not...
So my conclusion:

1) I don't have enough trades so that i can make a statistical rule for this triangulation theory or my calculations are somewhere wrong.

2) The triangulation doesn't work as expected so the theory is not good. Because brokers/banks can manipulate the prices the way they want.

If anyone understands the concept (i think i understood it ) and wants to check my calculations and give a second oppinion about this triangulation/consensus please let me know. I will give you my ea so you can test /check with it what i say here. But please don't ask me to send it and after tell me you don't know how to use or ask me what it does or things like that.
I will post it on the thread if kurka agrees, because this is his thread and don't want to offend him by posting a "workable" version.

Please let me know what you think,
Thanks, Victor