Algorithmic trading: Understanding the double edged sword

Of late there has been considerable attention directed towards algorithmic trading, and the value and risks it brings to markets.

Let me begin by describing what algorithmic trading is. Let us assume that a share of Reliance is available for Rs 1000 in National Stock Exchange (NSE), while it’s available for Rs 999 in Bombay Stock Exchange (BSE). I could write a computer program, that buys any share from a cheaper location (BSE in above example), and sell at the costlier location. The program will keep generating buy and sell orders and I can make money out of any price difference, without spending any of my own time. Such orders that are automatically generated from computers are called algorithmic (or algo) trades.

Algo trading has been around for many years now, and has been evolving with increasing power of computers, lower latency networks and ever smarter programmers coding them.

What benefits does algorithmic trading bring?

Firstly, it makes trades efficient and helps keep prices consistent. In the example above – if a trader had to manually make trades to keep prices between NSE and BSE in-sync, it would have taken a lot of effort. Above is a simple example for equity trading, we could have examples of exchange rate differences between INR/USD rates between multiple exchanges worldwide.

Algorithmic trading makes such trades efficient and we can be assured of trading on any exchange with confidence that prices are consistent across markets.

Secondly, it improves liquidity. When there are large differences between bid and ask trades (spreads), algo traders step in by quoting smaller spreads, increasing liquidity and making quick profits in minutes.

The third example of application of algorithmic trades is to fund whose strategy is to align with an index. For example – there are mutual funds that are indexed to Nifty and will always return the same value as the Nifty index does. Such funds need to keep making buy and sell decisions as prices of index constituents move, to stay consistent with proportions of stocks that are part of Nifty.

Algorithmic trades help automate trading in such funds and bring down portfolio management costs significantly, helping get significant returns.

Algorithmic trades don’t just benefit the market overall. They can generate large profits for the people deploying them. For instance, I could have a belief that any time price of crude oil rises, prices of Cairn India will go up. I could deploy an algorithm that tracks crude, and trades in Cairn India stocks in case price movements in crude are more than 5% and not correctly reflected in Cairn India’s price.

I could write an algorithm that sniffs for news reports on stocks, and before market reacts, it could place orders. Algorithmic trading can become even more powerful with machine learning strategies, where algorithms can keep learning and improving their configurations and strategies with each execution.

So far all looks good, is there any risk that these algorithms bring to markets? Let’s understand this with an example. Let’s say my trading strategy is to sell anything whose price is below its 100-day moving average (a.k.a. 100DMA, the average price of this stock for last 100 days). In the diagram below, I have circled out when this strategy would give me a buy signal (in 2014) and allow me to retain this stock till early 2015 when a sell signal is generated.

(above graph generated from moneycontrol.com)

It’s great because I can run it across multiple stocks and system can run automated and generate profits, with automated strategies for stop losses, exits etc.

But, what if there were not one but hundreds of such algorithms having same configuration? The day stock falls below 100DMA, everyone could start selling it and prices could crash way beyond fundamentals need them to. There are other similar risks, one of them is called spoofing. Here, the algorithm places a large number of orders at ridiculously low or high prices without an intention to execute, only to fool traders or algorithms to believing that there is a demand to buy or sell that stock. The flash crash in 2010 in US is an example where markets crashed almost 9% within a matter of minutes, and recovered as well! Such events reduce confidence and pose risk to average retain investors, who do not have deep pockets to withstand such shocks and may lose faith.

Investors expect regulators like Securities and Exchange Board of India (SEBI) to help regulate markets, and ensure such systemic risks don’t exist. On its part, SEBI has been coming up with multiple guidelines on algo trading. For example, guidelines require that algorithmic trading needs to ensure its orders are placed within the price ranges specified by exchange, algorithm based traders are required to mark their orders and consider all executed and unexecuted orders before the next order is placed, and also require brokers to ensure they have checks to avoid runaway situations.

What does this mean for our markets? Certainly with technology rapidly evolving and high-frequency trading and other types of trades picking up, odd and sharp price movements that cannot be explained by fundamentals cannot be ruled out. For the average retail investor who isn’t savvy with these technologies, watching their portfolio returns suddenly take a sharp dip could be unnerving. The best course would be to stick to fundamentals, invest in companies with sound business and potential for long-term returns, and ignore short-term sharp movements (or even better, use them as good opportunities).

(This article has been authored by Raja Raman, Director-Technology, Sapient Global Markets)

(Image: bankers-anonymous.com)