Setting-Up An Algo Trading Desk

Algorithmic Trading Firm Setup

By Apoorva Singh

You need domain knowledge, skilled resources, technology & infrastructure in the form of hardware and software for setting up any business or start-up. The requirements, especially in terms of regulations, infrastructure and cost estimates can vary depending on the country you plan to set up your desk in but overall, things will fall under this umbrella.

This blog will give you an overview of the requirements for setting up an algorithmic trading desk or firm.

Requirement For Setting Up An Algorithmic Trading Desk

  1. Registering your company: The first step is to register your firm. You can register your trading firm (for proprietary trading) as a Company, Partnership, LLP or even as an Individual. If, however you want to set up a Hedge Fund with investors, other approvals from regulators (For e.g. SEBI in India and MAS in Singapore) are also required and the compliance rules and regulations are generally much stricter.
  1. Capital required for Trading and for Operations: Broadly speaking, trading capital required for High-Frequency Trading is usually relatively less than that required for Low-Frequency Trading. LFT is scalable and can absorb much more trading capital. But the capital required for trading operations is typically far higher in case of HFT as compared to LFT given the infrastructure and technology requirements in HFT.

How To Set-Up Your Own Algorithmic Trading DeskClick To Tweet

  1. Trading Paradigm: You need to decide on the trading philosophy you’ll adapt. The most common trading philosophies include execution based strategies where the focus is to get the best price for execution rather than focusing on Alpha. Then there are High-frequency strategies which are extremely latency sensitive and mainly include market making, scalping, and arbitrage. Then there is market sentiment based, machine learning based and news based trading algorithms which can be relatively less sensitive to latency as compared to HFT.
  2. Access to Market: There are different kinds of memberships which exchanges offer- clearing members, trading members, trading cum clearing members, professional clearing members, etc. If you don’t want to go for direct membership in the exchange, you can also go through a broker. This involves lesser compliance rules and regulatory requirements. However, the flip-side is that you have to pay brokerage and most HFT strategies are highly sensitive to transaction cost.
  1. Infrastructure Requirements: Main focus areas under this head are Colocation, Hardware and Network Equipment and Network Lines.

a) Colocation: Colocation means that your server is in the same premises and on the same local area network as that of the exchange. Most exchanges provide colocation facility now. In some cases when exchanges do not provide colocation facility, there are vendors who provide co-location or proximity hosting facility. A significant percentage of orders received by exchanges are now generated by algorithms with most of such orders being generated by co-located space.

b) Hardware: Many leading companies produce servers required for Algorithmic Trading setup. Customizable hardware for high-frequency trading is also available which can be modified as per the requirement to improve performance. Given fast changes in technology, the present scenario requires servers to be changed and updated almost every year or at most in two years.

c) Network Equipment: This mainly includes Routers/Modems, Switches and Network Interface Controller (NICs) and FPGAs. For routers and modems, you need to check version compatibility with exchanges. NICs are basically Ethernet cards which help your computer to get connected to a network. FPGA stands for Field-Programmable Gate Array. It is basically an integrated circuit containing an array of programmable logic blocks and that be configured to perform complex operations.

d) Network Lines: Network lines can be broadly categorised into the below four categories-

 i.Trading Lease Line– Used for sending out orders to the exchange. Different lines provide different bandwidth for messages to be sent and are priced accordingly.

ii. Market Data Lease Line– This line used to receive market data from the exchanges or your data provider. There are two main formats ways in which exchanges send market data- Tick By Tick or Snapshot Data (example for NSE).

-> Tick By Tick (TBT)- Tick data is a collection of sequential “ticks” which is the latest quote, trade, price, and volume information. You can also subscribe to bucket feed which filters data for specific instruments requested.

 -> Snapshot Data– Snapshot Data feed contains data pertaining to Stock Exchange trade quotations and other related information pertaining to the trading of different instruments generated at regular intervals of time.

iii. Lines between Exchanges: These are the point to point lines between exchanges which can assist with SOR. Smart Order Routing (SOR) lets you shoot orders to different exchanges, in effect helping you to pick liquidity available on different exchanges at the most effective price.

iv. Between Premises and Exchange: In India, you cannot have the internet in colocation area, so there is a dedicated line between colocation premises and your facility. The cost of this line depends on the distance.

v. Test Connectivity: Exchanges provide test markets where you can test your trading algorithms. For instance, in India, NSE provides two test markets; Normal test market and Dedicated test market. Some Global exchanges like CME also provide internet VPNs for test connectivity.

  1. Algorithmic Trading Platform: An algorithmic trading platform has three main parts-

a) Market Data Adapter– MDA is used to receive data from the exchange and convert it to the format which our trading system understands.

b) Complex Events Processing Engine– CEP is the brain of the system and the main strategy logic lies here.

c) Order Routing System– CEP sends instructions to ORS which converts the order to exchange understandable format. FIX is the most widely used format in most exchanges, some exchanges might have their own native formats as well. When an exchange uses both, a native and FIX format, sometimes native may be preferred due to faster connectivity as the FIX converter might be applied in the next layer but using the exchange’s native format might also involve dedicated efforts in terms of maintenance.

      The latency of various platforms varies from system to system and so does the price.

  1. Backtesting: Backtesting is a historical simulation of an algorithmic trading strategy to see its performance in the past data. Most ATPs come with backtesting platforms which can be used to obtain simulated results in terms of profit & loss, risk and performance statistics over the duration of the backtested data which help to quantify the strategy’s return on risk. Next, we test the strategy in the “Test markets” which we’ve already discussed in the previous section briefly. Market tests ensure that there are no technical glitches which might occur while connecting to the market through the strategy.
  1. Risk Management: Risk management generally involves more focus on Market Risk monitoring. But in the case of High-Frequency trading, Operational Risk is much more important. Failure of technology, network, data streams can be disastrous. You need to have multiple level checks for data, starting from the socket level to capture any anomalies and stop the strategy instantly if something is wrong. A matter of seconds can lead to huge losses, which makes it important to react very fast and disconnect within a few milliseconds or lesser time duration if things go wrong.
  1. Conformance and Empanelment: In India, you need the exchange’s approval before you take a strategy live. The process involves participating in a mock to give a demo of your strategy to the exchange. If all required conditions are satisfied then the strategy can be taken live. Some exchanges like CME don’t require each strategy to be tested separately; they just test Trading Systems and grant access.
  1. Audit & Compliance: All HFT firms in India have to undergo a half-yearly audit. Auditing can only be done by certified auditors listed on the exchange’s website. For the audit, you are required to maintain order logs, trade logs, control parameters etc. for the past few years. Other global exchanges like CME require similar data to be saved for the past few years for audit purposes.
  1. Team: And last but not the least, you need a team of professionals to come together to run your desk. Broadly speaking Traders/Strategists, IT professionals, Network managers, Risk Managers, HR and Legal teams need to work together. But to start with IT professionals and Traders/Strategists should be sufficient. A small team of 3-5 Traders and IT professionals, along with Support Staff, i.e. a total of about 7-10 people can constitute an algorithmic trading firm. In the case of start-ups, a single person can don multiple hats taking responsibility for several tasks and a team of 4-5 members can start.

Bonus Content

Frequently Asked Questions about Setting-Up an Algo Trading Desk

Here are some of the most commonly asked questions which we came across during our Ask Me Anything session on Algorithmic Trading.

 

Question: What are the hardware requirements for Algorithmic Trading?

Reply: It depends on what kind of trading you are talking about if its LFT then you don’t need much, what you need is a decent laptop/desktop on which you can open a browser and code a strategy and run it or install a broker platform and code your strategy and run it.

If we are talking about Low-Frequency Trading along with quantitative analysis in which you are also doing good amount of data analysis and data chunking in that case you there are two options available for you either you go for a decent machine on your side that will not cost you a lot maybe around 1000-2000 USD (depending on which geography you are in) and the other option is you can do a lot of analysis on the cloud, the cost of cloud services is going down very rapidly especially where Amazon and Google are offering really competitive pricing with pay as you grow models. This makes it much easier for an individual and retail trader to get access to the financial computing power which was not there probably a few years back, something that they can access without spending a huge amount of money on that kind of hardware.

If you are talking about medium frequency trading strategy, in that case, you would need to use some algorithmic trading platform which means you would need decent servers which will be around a few thousand dollars. If you are talking about HFT then it would be a specialised server which should be in the range of typically 10-25k dollars.

 

Question: What is the investment required to start the algo trading business? What will be the incremental investment for HFT?

Reply: For algo trading business if its low or medium frequency not much so you can use google platform or vendor platforms which are not that expensive it can range about few thousand dollars a year to few thousand dollars a month depending upon what vendors or platforms you are going for.

For HFT it will be a few hundred thousand dollars which you will spending on the acquire the required infrastructure and platform. So the cost will be definitely higher. I do not generally recommend people to start algo trading if they do not have established the business in place. So if you have not done automation starting an HFT might not be a great idea because automation itself comes with a number of loops with it.

 

What is the cost of setting up your Algo Trading Desk?Click To Tweet

Next Step

We bring you an account of one of our EPATian, Mr. T K Raj who has been in the technical space for over twenty years, heading different business functions. Learn how Executive Program in Algorithmic Trading (EPAT) helped him to master the concepts of algorithmic trading and excel in this domain.

Disclaimer: All data and information provided in this article are for informational purposes only. QuantInsti® makes no representations as to accuracy, completeness, currentness, suitability, or validity of any information in this article and will not be liable for any errors, omissions, or delays in this information or any losses, injuries, or damages arising from its display or use. All information is provided on an as-is basis.