Forex Automated Trading - maths expert required

[money tree]

For the last 5 months I have used an automated program to trade currencies.

This is something I designed and built myself as other similar packages cost $6,000 to set up and cost $600 a month to rent.

Mine cost $95 to set up and $0 a month to rent

This is NOT a technical analysis "black box" or any system that tries to 'predict' the market. My system simply buys and sells very small amounts of currency very often for a very small profit. Its like a giant vacuum cleaner sucking up 5c pieces once a minute 24/7!!

Anyways, this setup makes me around $30 a day, or $160 a week. Doing this requires around $6,000 capital. Here is the issue. The computer must be running 24/7. At 30c an hour, this really adds up.

Currently it costs around $40 a week in electricity. Also, the hard drives tend to wear out fairly fast, and Im now on #3.

So, although my $160 a week sounds good, its not an efficient system. I really need to increase the profit, either by optimising the buying/selling, or adding more capital.

Is anyone here a maths expert? I have some mind-bending calculations.........

Is it better to limit buy 1 pip below market and sell 1 pip above buy price?

or

Is it better to limit buy 10 pips below market and sell 10 pips above buy price?

using the 1 pip system, there are a lot more trades, but the profit they produce is very small. In a downtrend it is easier to sell on a 1 pip bounce.
Likewise in an uptrend, there are plenty of pullbacks allowing you to buy.

using the 10 pip system, there are far less trades, but each trade makes a good profit. You dont get too many 10 pip bounces in a downtrend, so you can end up owning a LOT of currency. In an uptrend, there is rarely a 10 pip pullback so you often miss the rise completely.

Can anyone determine a formula / logic to optimise this system?

 

[happytrader]

Hi there Moneytree

Just wondering how many man/woman hours do you put into running your system?

Cheers
Happytrader

 

[money tree]

well its automated

ie ZERO

lol

 

[Dan_]

Also, the hard drives tend to wear out fairly fast, and Im now on #3.

Is the system really that data intensive that your killing hard drives? I do some hard drive intense stuff (irregular) but by running RAID I get a few years out of them. I'm interested to know the timeframe to kill the drives

 

[happytrader]

Would tripling your stake to $18k cost you anything more in outgoings?

Cheers
Happytrader

 

[money tree]

outgoings are fixed

increasing the capital would help

im now trying a 3 pip system, so far its making $2/hr. wont know how good it is until we have a few serious trends.

hard drives dying cos they must be engaged 24/7.....I cant turn them off when not in use.......I now have a fan on the HD which may help a lot.

actual data use is almost zero

 

[Smurf1976]

Just wondering what currency pair(s) you are trading? If only one pair, you could add a second pair and your maximum drawdown probably wouldn't doublt so this would increase your return on capital without extra computer costs.

I assume your figures are net of broker's spread?

As for the hard drives, I'm no expert but with enough RAM in the computer it ought to be possible to avoid constant hard drive access unless it really is data intensive. Not really sure how to set this up so it works though but I know it can be done - keeping all the frequently used stuff in RAM and only saving the changes to the hard drive. You could then set the hard drive to shut down when idle which should save it wearing out so quickly.

 

[Smurf1976]

outgoings are fixed

increasing the capital would help

im now trying a 3 pip system, so far its making $2/hr. wont know how good it is until we have a few serious trends.

hard drives dying cos they must be engaged 24/7.....I cant turn them off when not in use.......I now have a fan on the HD which may help a lot.

actual data use is almost zero

There are settings in Windows relating to power management (to save electricity). You can set hard drives to shut down when not being used which ought to help. Go to My Computer - Control Panel - Power Options and you can set the hard drives to shut down.

 

[Dan_]

Some great ideas Smurf,

If the computer is required to be on 24/7 and you don't have a climate controlled room then mod your case. Depending on how technical or hands on you are you can buy/customise your comp with a few of the following to help increase reliability & wear

• Rip the standard heat sink off your CPU and put a larger one on (lowers cpu temp)
• Use a minimal amount of cabling & tie it up so there is room for good air circulation
• House your system if possible in a larger case, allows for better airflow & heat dispersion
• Add a few extra fans to the case (intake at the bottom, extract at the top as long as noise is not an issue)

The quicker you can get the heat out of your system and away the better it will run and have less wear on parts. I personally have a customised thermaltake case for my system as when I do rendering and the such it's very CPU + disk intensive = more heat.
With 9 fans on my case I can pull so much heat out that I can overclock "speed up" my CPU by a .4 GHZ. However it can be a tad noisy due to the fans, however most motherboards have variable speed controls so when the pc gets hot they will increase the fan speed accordingly (and slow down when cool)

Or you could consider liquid nitrogen cooling

 

[happytrader]

Have I got the facts here right so far,

You are and have been live trading your automated system for five months.

Using a capital stake of $6,000 your gross income is $160.

Your expenses are $40 per week for electricity, and the cost of hd's

Your outgoings are fixed.

Would $160 derived from the first $6,000 of your stake cover all or most of your costs?

Are you happy with your risk to reward ratio?

Would I be right in assuming that if you tripled your stake to $18k you could derive $480.00 per week before expenses?

Thats a pretty good return on investment in my books for a fully automated no brainer.

Cheers
Happytrader

 

[money tree]

Other expenses that could be considered are the upgraded pc ($800) & the cable internet ($60/month).

Also I needed a dedicated pc for the system and another for surfing / trading. But then I figured 2 pc's would cost 60c an hour. So instead, I bought some software & hardware that allows me to run 2 windows logons with 2 monitors, 2 keyboards and 2 mice.

I dont think running on RAM would work too well. I have 2GB ram, and because of the enormous number of orders placed, the Java cache runs out @ about 600 orders. Sounds like a lot, but thats only around 5 hours. I have to reboot with that frequency. So placing extra demand on the RAM would be unlikely to help.

I have tried trading 10 pairs, and due to correlation I found that this becomes dangerous. I only go long positive carry trades. I currently trade 2 pairs because of their tight spreads of 1.5. These are EURUSD & EURGBP. EURUSD is negative carry but small.

For a while I traded in lots of $1000 on 10 pairs. While this produced an amazing $1500 a day, eventually I got caught out and dropped $4k in a day.

I now trade in lots of $50. Sounds very small but when doing 1 order a minute you can easily rack up 50k in one currency.

Risk/reward is fabulous. It was a few months work writing the program and setting up the pc.

yes with $18k $480 a week is acheivable, but Im not prepared to put down that sort of cash without at least 12 months consistent performance.

 

[Smurf1976]

I dont think running on RAM would work too well. I have 2GB ram, and because of the enormous number of orders placed, the Java cache runs out @ about 600 orders. Sounds like a lot, but thats only around 5 hours. I have to reboot with that frequency. So placing extra demand on the RAM would be unlikely to help.

yes with $18k $480 a week is acheivable, but Im not prepared to put down that sort of cash without at least 12 months consistent performance.

A lot of trades there it seems! Could you set the computer up to run from the ram and just start the hard drive up once every, say, 2 hours and transfer the data from RAM to the hard drive and then shut the hard drive down again as a means of solving the wearing out problem?

Where I work we have cause to have computers running 24/7/365 and do so without problems. They are running a control application which controls external equipment and there is quite a lot of data involved. But we aren't running the hard drive heavily so that might be helping. We haven't had any real problems with PC's wearing out although we do keep a second machine running right next to the "real" one so if it does fail then we just pull the cables out and swap over to the already running spare machine. The reason for this is to minimise down time which is critical in this application since if the PC goes down then we lose control of all the external equipment. The room has a reverse cycle air conditioner (ordinary domestic type) running all the time set to about 21 degrees.

As for the scale of your trades, there would be some logic in just retaining your profits in the system as a "natural" source of increased capital IMO. That's been my approach trading forex although my system is very different to yours.

 

[happytrader]

Hi Moneytree

A maths expert is not what you need. Markets as you well know are never consistent. You need to look to yourself for consistency. Your need is faith. Faith that you will always act consistently in your own best interests. I'm always working on that one myself by the way. LOL

Cheers
Happytrader

 

[The Estimator]

I agree with the other guys. Are you sure there is no way to minimise hard drive use so as to use ram or other flash style memory systems? With these types there are no moving parts so all that is needed is temp regulation. You can even buy small liquid and gas coolant systems that if used on solid state equipment should increase service life greatly.

 

[Kaizen]

I am looking at setting up automated forex trading as well and was wondering what broker/mm you're using?

At the moment I want to go with Oanda but would like to run my system for free for a while before paying the $600 US a month for their API.

Thanks

 

[RodC]

using the 10 pip system, there are far less trades, but each trade makes a good profit. You dont get too many 10 pip bounces in a downtrend, so you can end up owning a LOT of currency.

money tree,

You say you can end up owning a lot of currency. Do you use stops?

Rod.

 

[Strw23]

Money Tree
I was just wondering if you were still running this system and how it was going. If you are not to busy is there any chance you can give us all some more details. Im not interested in you gates and triggers but more like, who you trade through, API details, computer setup, programing/platform, are you still chewing up hard drives, do you have a ups or any comms back ups if your modem goes down. love to hear the details.

scott

 

[chemist]

For the last 5 months I have used an automated program to trade currencies.

Gonya.

...
This is NOT a technical analysis "black box" or any system that tries to 'predict' the market. My system simply buys and sells very small amounts of currency very often for a very small profit. Its like a giant vacuum cleaner sucking up 5c pieces once a minute 24/7!!

Sounds like arbitrage. Is it?

Anyways, this setup makes me around $30 a day, or $160 a week. Doing this requires around $6,000 capital. Here is the issue. The computer must be running 24/7. At 30c an hour, this really adds up.

Currently it costs around $40 a week in electricity. Also, the hard drives tend to wear out fairly fast, and Im now on #3.

Now my BS detector is tingling. I have been using automated computer trading for more than two years, and my computer has never lost a hard drive. Admittedly I run Linux and don't buy crap like maxtor, but still...

So, although my $160 a week sounds good, its not an efficient system. I really need to increase the profit, either by optimising the buying/selling, or adding more capital.

Is anyone here a maths expert? I have some mind-bending calculations.........

Is it better to limit buy 1 pip below market and sell 1 pip above buy price?

or

Is it better to limit buy 10 pips below market and sell 10 pips above buy price?

using the 1 pip system, there are a lot more trades, but the profit they produce is very small. In a downtrend it is easier to sell on a 1 pip bounce.
Likewise in an uptrend, there are plenty of pullbacks allowing you to buy.

using the 10 pip system, there are far less trades, but each trade makes a good profit. You dont get too many 10 pip bounces in a downtrend, so you can end up owning a LOT of currency. In an uptrend, there is rarely a 10 pip pullback so you often miss the rise completely.

Can anyone determine a formula / logic to optimise this system?

Sure, just simulate using real data. All you have to do is write code.

Chemist.