Algorithm trading - In and out
The term ‘algorithm’ has been popular with share market trading. Not knowing what is ‘algorithm trading’ many think that they would get into it and earn profit. In order to know what is ‘algorithm trading’, how does it operate, what are the pros and cons of it, we approached an expert in algorithm trading, Vijayanand Venkatraman, who is basically a financial consultant. He elaborated about it.
Algorithm is ….
“Algorithm trading is a next step to technical analysis. Technical analysis is a decision making tool based on the share price movements. Algorithm trading is predicting the next move of the shares by backtesting the past statistics, thereby finding out the differences and similarities and eventually arrive at a decision.
Three types …
There are three types such as Algorithmic, Quantitative and High frequency. We can create different models, creating daily models based on the everyday price movements, hourly model based on the price movements within hours and tick point model based on the immediate next price movement. The process of decision making, what is for tomorrow if it is daily data and what is the next tick if it is tick data is termed as algorithm trading.
Among these models, daily and hourly models operate at a low frequency. For tick data, high frequency analytics is essential. This is because we can create high frequency model only when we could get more data.
Through high frequency analytics, it is possible to find out how the next move would be by researching second by second price movement details. Only through that best offer and best bid can be made.
Profit not the goal
Most of the technical analysis methods are profit oriented research approaches. But, algorithm trading analysis is a cost minimizing research approach. It is because with regard to share market, impact cost is an important factor to consider. If we do not consider it whatever profit we get through share market will be spent only on fees. Therefore, cost reduction is the primary goal of algorithm trading.
That means, if a share is bought for Rs 100 crores, there is particular expenditure. If it is bought for Rs 50 crores there will be a different expenditure. Order should be placed in a way to reduce the expenditure and buy the share at a very low cost. The job of algorithm trading is to predict that point.
With the high frequency method what is important is our execution rule. Model should be created with an aim of how to buy it at the lowest price and how to reduce expenditure.
Apt for Index
Algo trading is a product of commodity trading. Algo trading may not be big at the level of a particular share. This is because individual shares may not be available with high frequency data. High frequency trading can only be possible only in index, currency and commodity trading since volume of data vailable is high.
How to prepare?
First, we should capture the idea of the trading we are going to do. We should write the rules to get what kind of data to be analyzed and what kind of output is expected. We need to program the model based on those rules.
There are two types of analytical methods in algorithm trading. One is Trend Following and the other is Mean Reversal. Trend Following continuously follows the shares that keep growing whereas Mean Reversal is coming down where ascending growth stops.
The success rate in algorithm trading depends on the profit. With Trend Following method it is more even if the success rate is 20%. This is because it works only when there is Trend Following. In the Mean reversal method trading success rate is high but profit is less.
Portfolio is required
We do not know how long the model we have created will continue to work. Our model might not work at all. Therefore, the way we create portfolio for a safe investment, we need to create a similar portfolio for algo trading too. By this, if a model does not help us another model will bring profit to us.
We should decide at what frequency we are going to do trading with regard to algorithm trading. It depends on our budget, computing power, and market opportunities. It requires a minimum of three days to create a simple algo model. It changes as per our accuracy. It is important to note that all market ecosystems cannot be modeled.
Whom it is opt for?
When a model is created for algo trading we need to decide the entry and exit rule. What is important next is the volume. The number of shares is going to determine our profit and loss. We should have clarity in what position we are going to take and how many times we can afford to lose till getting profit.
What is needed?
We should have Java or C Sharper to do algo trading. Some use Python. But, its speed is not adequate for algo trading, because it was not created to deal with financial processing.
The data required for algo trading need to be taken from the Exchange. We need to take those data, process them and send them back to the Exchange for order. It will take a maximum of 4 milli seconds to take the data and resend them. It is said that some are able to process it within nano seconds. That is the extent of competition in trading.
We should also consider how fast we are trading. We need to keep our data receiver at the Exchange in order to receive that data faster. There is a fee for getting the data. If the fee is paid anyone can receive data from the Exchange. But whoever process the data faster are going to get the orders at better price.
Not working properly
During the recent dip in share market, most of the algo models did not work properly. So, even if an individual algo does not work properly, there is a possibility of getting positive returns from portfolio algos.
On the whole, we should consider whether the money we spend and time bring profit to us. Companies, spending crores of money on ultra modern computers and creating various models, are expediently doing algo trading. Next to them are high networth individuals creating algo models at high cost, doing algo trading. We cannot compete with them with a single algo model.
Algo trading is totally meant for companies and high networth individuals. When we consider the cost on creating algo model and the fee for the brokerage on our trade value, our hands will be empty finally. It is my opinion that algorithm trading is not at all useful for the small investors. Thus Vijayanand concluded. Let us hope that at least from on the small investors would be able to understand what is algo trading and therefore keep away from it.