Setting up my Algo-trading business

#1
Hi Friends,

Couple of months back, i wanted to check if i can set up a personal Algo-trading business. Like any business, my objective was profit maximization. But not at the cost of quality. I should have the option of scaling my infrastructure as per my business need. Also in future if i get clients, i should have the option of accommodating them without much headache.

So having all these things on mind, i started a small project with an initial capital of Rs.50,000. I was not sure if it could be done and if possible, why others have not try it. There must be a catch. So i decided to burn my fingers and try to figure out for myself.

My first problem was infrastructure. That is computer, internet, power, anitvirus, etc. If you the maths properly, it should add upto Rs.1699. The breakdown is as follows-
Computer -> cost is Rs.25,000, considering life is 5 yrs. So per month it works out to Rs.416
Internet -> Airtel connection, around Rs. 1000 per month
Antivirus -> Quickheal, Rs. 1000 per year or Rs.83 per month
Power -> Lets assume we shell out Rs. 200
So without placing a single trade i am spending more than Rs.1699. On top of this, system expansion was a problem and steady power supply was not garenteed. So what do i do? i had to find an alternative.

After some searching, i found the solution. It was cloud computing. There are lots of service providers. Some offering free service, whereas others offering the same service at a jaw dropping price.

Finally, i settled with Amazon EC2 server. It gives 1GB memory and 30GB storage space for free (for 1 year). This was great news for me. This translates to free service for first year and from second year around Rs.300 per month. Awesome...

Second, i needed a trading software and data source. I choose Amibroker for this purpose. This is compatable with most brokers who offer Algo-trading service. Also, its free and much better compared to Metastock. But, if you go with the free version. Then you cannot save the charts. I was fine with it, as i could open keep the chart open in the server and forget about it.

Data provider is the trickiest of the lot. There are lots of data provider in the market. I choose a Nest to Amibroker terminal offered by Zerodah. It costs Rs.11000 per annum. (http://zerodha.com/main/zerodha-trader-plugin-amibroker.html)

Thirdly, broker. I wanted to go with deep discount brokers. Only two brokers i considered, Zerodah and Compositeedge. I found composite to be better as they charge 0.006% as brokerage. I can also add in my own clients when needed. My only advice regarding brokers is, if you are planning for Algo then forget big names. Only consider those brokers who can give you less than 1 paise of brokerage. Negotiating with big brokers is a waste of time.

Fourth, amibroker and deal terminal bridge. Well there are lots of options here. I personally choose Symphony for it. For multiple client bridge, it cost Rs.35000 and single user cost Rs.25000. I am planning to go with Rs.35000 plan as i already have my first client :)

Lastly and most important is trading strategy. If you have your own strategy, but not coded in Amibroker. then you can get it done by a freelancer. They cost around Rs.20000. In my case, i had already coded, backtested and optimized it myself. You can search on the net and find a lot of ready made AFL available on various forums. You need to test these extensively before using it.

So, i had been successful to set up my algo-trading business for Rs.46000 (for first year) and from second year it goes up to (Rs. 49,600). My system is fully scalable and i can add in new clients when needed without any hassle.

If anyone can provide me ways to reduce my cost even further, plz let me know.

Do let me know if you have any questions....
 

jagankris

Well-Known Member
#4
Hi Friends,

Couple of months back, i wanted to check if i can set up a personal Algo-trading business. Like any business, my objective was profit maximization. But not at the cost of quality. I should have the option of scaling my infrastructure as per my business need. Also in future if i get clients, i should have the option of accommodating them without much headache.

If anyone can provide me ways to reduce my cost even further, plz let me know.

Do let me know if you have any questions....
All the very best :).
Nothing much regarding cost reduction unless you develop the robo for yourself :).
Now what is the service that you are offering for clients ?.
 
#5
I am planning to offer two types of services. First is "High risk", where i would use my strategy to take trade at a very low time frame (i will try to go with 5 min time frame). Second is "Low risk", this is basically pair trading strategy. Planning to start with NSE and then move into MCX in the next couple of months.
 

Cubt

Algo Trader
#6
But this like portfolio management service? What basically you offer is, you get money from clients and trade on behalf of them using your strategy and algo infrastructure. And clients would pay a fixed cost for that or profit sharing basis?

Since you have multiple account licence, if would be better if clients can use their own strategy and trade in your algo infrastructure. Bcoz unless and until we trade with our own developed system, we cannot sustain longer in algo trading
 
#8
You are absolutely right. It is more like portfolio management service, but its not aimed towards clients with deep pocket. But instead it is for those whose portfolio size is small. Also, you should understand that most of the people in these segment are normal office going people who neither have the necessary skills to build a trading robot nor have the necessary technical know how.
Lastly i have intention to keep it small and controllable, thus avoiding a lot of problems which are inherent to big business.
 

Cubt

Algo Trader
#9
Dear CUBT,

Could you please elaborate the above.
TIA.
Hi Jagan,

what I wanted to convey was with any mechanical system, there will be a drawdown. When you build your own system, you know how it works during range bound days and trending days, you know whats your maximum continues loss, max loss in a day, etc. But still in spite of knowing all these, its psychologically difficult to trade mechanical system when you are in a period of drawdown. Unless and until you believe 200% with the system you trade, its difficult to survive in markets

When you are trading with other's strategy, you might not know the above factors, even if the person selling the strategy gives the above metrics, it will be very difficult to trade during the drawdown periods, after a series of drawdown you will start doubting the system, you may start reducing the lot size or you may even stop trading after a period of drawdown.
 

mt4trader

Well-Known Member
#10
data and computer power is more important.dont go in cost cutting methods if you want real good business.go for high level sofrware which can handle level 2 data.
1 computer: go for multi cpu ( 4 cpus not cores) running on VM. or multi core more than 8 cores.

2 software: go for ninjatrader or trade station or any other software which can handle level 2 data.
3 data: go for esignal only provider of level 2 data.

my personal selection is 4 cpu motherboard running on VMWARE. software ninjatrader with cutting edge trading system based on order flow, market delta and time based system.

with these systems you can trade very small stoploss of 5 pips and big target is possible.
first master volume analysis and dont depend on useless laging indicators.
if you want more help you are welcomed
 

Similar threads