Business Standard

Stick to algo strategy even if it underperfo­rms

Hang in there for the optimal period to enjoy the kind of gains seen in back-testing

- SANJAY KUMAR SINGH

Many retail investors who have entered equity markets during the current bull-run have now begun to participat­e in algo trading, which was available only to institutio­nal investors till a few years ago.

How does it work?

The quant (the person who creates the algorithm or algo) studies historical data to come up with certain trading rules that he believes will make money. He then builds the algo, software that has those rules coded in, which then scans the market. Whenever the conditions that have been coded are met, it generates buy and sell signals.

To access this, traders have to enrol on platforms that offer algo trading strategies, like Rain Technologi­es, Tradetron, Squareoff, Algobulls. The trader also needs to enrol with a broking firm that offers API (applicatio­n programmin­g interface) integratio­n. An API gathers the signals generated by the trading platform and executes them in the trader’s broking account. Zerodha, Fyers, Upstox, Samco, 5paisa, among others offer API integratio­n.

Eliminatin­g human error

An algo can initiate trades or close out positions much faster than a human trader. “The trader does not have to spend his day glued to the screen since the algo monitors the market on his behalf,” says Raghu Kumar, cofounder, Rain Technologi­es.

Algo trading also eliminates human emotions. “Around 90 per cent of traders who trade manually lose money because they fail to control their emotions and make mistakes,” says Ramakrishn­an Selvaraj, co-founder and director, Quantindia Systems. Adds Rajesh Ganesh, founder and chief executive officer, Tripleint.com: “When a position goes into loss, retail investors wait for it to recover. And when it gets into a small profit, they

book profit.”

Select right strategy

Pick an algo strategy that has been back-tested rigorously on several years’ data and which has performed well since going live. The algo should have demonstrat­ed an ability to perform in varied conditions.

“Look up its percentage of wins and losses. Also, check the amount of money the algo makes when it is winning and the amount it loses when it is losing,” says Ganesh. An algo could win ~1 on 99 trades but lose ~100 on a single trade.

Another key parameter is drawdown: How much can the decline be before the algo starts performing again?

Finally, look up the minimum amount required for subscripti­on. This can range from ~50,000 to ~10 lakh or more. Also, check out the fee. “The quant decides the fee on our platform. It could be an annual fee, which is a percentage of the client’s deposit, with an element of profit sharing,” says Kumar. The fixed fee is usually 11.5 per cent while the profit sharing could be 10-20 per cent or more.

Go with quality players

Even if the algo is good, you could be done in by execution issues. “Adopt a high-quality trading platform and a high-quality API, even if you have to pay for the latter,” says Selvaraj. The algo platform should have very little downtime.

Every algo will make losses sooner or later. “Keep your investment­s in the algo for the time period suggested by the quant to be able to generate the returns seen in back testing,” says Selvaraj. Do not put any money into trading that you cannot afford to lose. And enter algo trading involving derivative­s with adequate capital so that when there is a drawdown, you are able to come up with the margin money required to maintain position.

Newspapers in English

Newspapers from India