Simple High Profit low Drawdown Triangular Arbitrage

yes. it makes sense that it will be market dependant... that means you ought to come to a 50/50 (50% winning trades) eventually..... it is true and to be expected that you may possibly start with losses 20/80 or worse to begin with, Temporarily, and it is true also that it will eventually resolve itself. But i get it. it is not acceptable parameters for your brainy strategy. TY for testing it. So yeah it is a solution, just not one you particularly like. I still think it likely the only solution..... a solution that delivers a way to compensate for spread is going to have to do so with drawdown. I do not see any way around that. Could you not simply account for that with account size and money management? actually there is one way around that. a good strategy, that rarely loses many times in a row.... that way you decrease your probability of going 80/20 draw down n your eventual guaranteed profit.

that interests me, because you are only looking for a strategy that wins a few loses a few without losing a bunch in a row, even if it is not a positive expectancy strategy....as opposed to a strategy that has + ev and an edge or profitability. your goal is to find a break even strategy that oscillates back n forth. that is kinda cool man.

i'm not giving up just yet....lets go a little deeper...... lets do sets of ten trades, positive sets we bank but we also use the fluctuation to our advantage, negative sets we cut our losses as efficiently as possible.....


e.g.
take 10 trades hedged at +-100 dollars as discussed
so if you get unlucky in your first set, 8 of them go negative and 2 positive
then

close the positive ones at +$120(as opposed to only +100 dollars) = +240

close the negative ones at -84 = -672
Net = -432

then take the next set of 10 trades....... etc.

you get the idea

if this time you have 5 wins and 5 losses = 600-420= +180


since your overall EV is +18% you will eventually make 18% in the long run....

the question is one of practicality and efficiency to find the most efficient workable parameters, i.e money management, not the theory. no?

obviously it isn't ideal, but it is a solution

@faryne
NOOOOOOOOOOOOOOOoooo
I messed up by playing with my code and it opened the wrong trade on my side
Will teach me to not experiment to much with the code
Don't worry.
The oops I did in the code is not in the EA on this forum.
But for interest sake look at the expert log.
You will see where it start opening the trades.

Check from the log time of the log line just before the first trade and the log line just after the last trade to see the trade lag.
This is a easy way to see the trade delay.
You want that as short as possible.

Also look for a line "OPEN3".
The last one of those lines will show the amount of slippage because of the trade delay.



@dagoods
Brainy thanks. But I just like maths and programming so lets call it lucky
I really appreciate your input.
I can see what you mean by leaving the one leg out and yes over a longer period of time it should level out to about 50/50 if you ignore the spread and commission.
So your theory will work, but the problem is that it will increase the drawdown from the current say 10% to closer to 100%.
To decrease the the drawdown you must then trade smaller amounts of lots, but this also decreases the amount of profit.
So in theory ultimately it should balance out to about the same as the current strategy of using 3 legs.

But as I am now typing I can start to see a glimmer of a possibility where it might make the affect of spread a bit less.

The big problem is sudden movements.

With 3 legs all balances out.
So during a big market spike I have seen 2 symbols go negative -$20000 on a $10000 account, only for the 3rd symbol to be positive $20000 so balancing out to ZERO.

Now if there where only 2 legs and you were unlucky to use the 2 losing symbols then your account would have been blown.

OK OK this is a bit of an exaggeration. It is more like -$6000 on a $10000 account, but more than enough to hit your margin call and stop out levels.
If you run the back test slow it down and watch the trades running.
It is scary to see how big or low the profit goes for single symbols, only to be cancelled out by the others.

Lets call the current 3 legs strategy the safe approach and the 2 leg one the risky profitable one.
Still the safe strategy can still make 100% profit every 3 or so months during slow times and that to me is good for now and prevents heart attacks from crazy market movement
10% on my one account just last week. Was unexpected.

Still it is food for thought and something I can test a bit more in future with some safeguards in place, but my main focus now is to decrease the execution time using FIX API or whatever else I can incorporate to decrease the slippage.



I would love to get access to brokers with spreads close to or even better than ENC broker with extremely low commission.
Probably only available to banks, but just think how many hits the EA will get with an account like those.



@dagoods 2nd post
That sounds like a grid system.
Yes that should also work.

Problem is that the current EA uses as much if not all of the available margin, minus a bit a profit/loss gap.
So for each grid level you add you have to decrease your lot sizes by half.
So 3 grid levels deep is 4 times less and 4 times is 8 times lot size less.

I did have a grid system in the beginning and found that I get about the same amount of profit.
For some strange wonderful reason if I use the 3 legs with a grid then decrease the lot size to cater for the extra grid levels then the extra grid levels cancels out the lot size decreases to give about the same amount of profit, maybe a bit less.

But still I can see where you are going with the 2 leg strategy and I want to test it, but time management is my biggest problem at this time.
I have barely enough time for this one EA, let alone all the variations on it and all the limitations of metatrader I had to overcome.
I basically get a couple of hours maybe once every 3 or so days and that is if I sleep 3 hours a day, luckily I get 4 hours sleep a day
I wish I could do this full time.
Just think of the wonders I could develop.

This EA already produces some mutations like a 1 leg latency arbitrage, drivers to connect outside of metatrader, fix api incorporation, multi-threaded MT4 and MT5 EA functionality, massive market watch speed improvement coding, and lots more...
Time is not on my side

But do not stop giving ideas, maybe one day.


Now to go and fix my coding mistake on my running EA
I like chatting too much

hello @jvbJacques
first of all i would like to thank you about your job.

i have attached to GBPUSD chart your lastest ea to MT4 demo with €500
first of all it opened & closed 1 position, losing some eurocents. (i don't know why but it is twice)

After that it opened with maxDD set @0.3:

1. GBPJPY = 0.50 lot
2. GBPUSD = 0.50 lot
3. USDJPY = 0.67 lot

currently the floating point is -15.13 euro
my question is this: is correct different lost used? because USDJPY was 0.50 as the other 2 the floating point should be -6.13 euro

thanks x all
regards

@raffus
All seems correct.

The first trade is to determine the commission.
It opens one 0.01 lot trade and closes it immediately, only once and then look into the trade history to determine the commission.

There is only one of those 0.01 lot trades, but I suspect you are using mt5, which to me has got a better trade response time than mt4.
In MT5 one trade looks like 2, instead of 1, but it is only one.
1 = an "IN" trade indicating a position was opened.
2= an "OUT" trade indicating the above "IN" position was closed.
But together they form one fully completed trade.

As for USDJPY.
It must not be the same amount of lots as the other 2 so it is correct.
It need to be different to fully hedge/balance out the calculation.

As for the profit/loss.
Look at the lines drawn on the chart.
I suggest making everything on the chart BLACK so that the lines the EA draws can stand out better, then zoom the chart out to see the lines.

There are 2 line channels indicating a virtual ASK and BID lines.
Then there are 2 blue lines indicating the open trade positions and finally the middle average line.
If the ASK line crosses the bottom blue line then a trade opens.
The trade will remain open until the BID line crosses the middle line.

The amount of profit depends on the amount of slippage you have.
So I suggest you must have a very good broker connection of about 50ms or less.

But yes the profit/loss will float there forever until it crosses the middle line.

I'm still looking into a way to improve the trade response speed to reduce slippage.
I was thinking of creating a pending impossible trade and then updating that trade once every 25 second to keep the trade server connection connection so that there is no delay in waiting for metatrader to first sign on to the trade server every-time a trade need to be made.

FIX is even better, but I have no demo FIX account to test it on so for now that must wait.

I still suggest not to use this on a real account until fully tested and developed.
Remember.
I have not inserted code to pick up preexisting trades, so if metatrader restarts then you must first close all existing trades before restarting the EA.

I will insert code later to pick up existing trades.

Good luck.

Hi Jacques,

Thanks for the EA!
On forward demo test (broker with ping 8.71ms) yesterday I've got trades with different lot sizes, as it should be. By mistake, the Metatrader restated so I had to close those trades manually. Today, however, the EA opened trades with the same lots. At the moment is in floating profit, waiting for the EA to close them by itself.

Regards
Attila

@eatrader75
Sorry to hear about the restart.
Will include the code to pick up lingering trades upon EA restart as I get time.
Just keep in mind this is still a work in progress.

Still a good example of what happens with the triangle.
You can leave the trades for a long time and the floating profit loss will always return to the zero point. (If little to no slippage)
So less stress.

Nice commission.
What spread do you get on the 3 symbols ?

Also look into the expert logs to see what slippage you are getting.
fx1=GBPJPY
fx2=GBPUSD
fx3=USDJPY

So looking through the mess of logs you can see what symbol does what by looking for the 1,2,3.

Good luck

@Everyone and @eatrader75
Looking at your screenshot I can see a problem.

The buy,sell,sell is not balanced.
Lot size of USDJPY need to be different.

I do not know if some test code snuck in, but in either case.
Please update your EAs with the new attached EAs.

Uploaded V17

@Fxentropy
Looks correct.
It will only start trading after 100 minutes so give it time for it to build up some history.

@Everyone
If you want to view the lines drawn by the EA correctly then attach the EA to a 1 minute chart.
On a one hour or other time-frame it is a bit difficult to see the finer points of the chart.

The time-frame of the chart does not affect the working of the EA.
Just the display part.

Done

@Fxentropy
WOW
The difference between the top white line and the bottom red line is basically your spread/commission.

You have got a fantastic spread and commission.
What is your commission and spread and where can I get an account like that ?

Look at my screenshot of myfxchoice in post #170
It is not even close

The smaller the spread and commission the more arbitrage hits you will get.

Again.
What broker and account are you using and can a broke guy like me get an account like that or is it something special ?

This is a Darwinex account. (Leverage 200)

But i think you can get one cheaper at Pepperstone (Leverage 400). Also Onetrade (Leverage 100) is sharp too.

All brokers can be funded with 100$ or even less...

I get lost in the log so I think a screenshot is better...

Hi folks !

Installed V17 on MT5, on my VPS for forward testing yesterday.

Eveyone seems to have nice little colored bars... but me !

Any reason?

Check the Market Watch on the left-hand side of the Metatrader. It has to show all the 3 currency pairs needed for this ea: GBPUSD, GBPJPY and USDJPY. If not, click to add. Then put back the ea on the chart

OK it works now after reinstalling the EA on a chart.

Will post results here every friday afternoon.

Thanks

Anyone had a trade in last couple of hours?. I had not

I had only one at London opening

Just missed an opportunity...