Stick to algos vetted by exchange or a reputed broker, advise experts

Go with players who can provide audited records of performance

The Securities and Exchange Board of India (Sebi) issued a circular on June 10 warning investors against dealing with unregulated platforms that offer algorithmic (algo) trading. The circular warns investors against falling prey to claims of huge returns, or being impressed by ratings assigned to strategies by such platforms.  

Be wary of claims of high returns

Experts say building an algo that can consistently produce reasonable returns, after factoring in all transaction costs, is difficult.

“Usually, these unregulated algo platforms attract customers with claims of unrealistic returns based on backtests that are either cherry-picked and/or conveniently ignore transaction costs,” says Atanuu Agarrwal, co-founder, Upside AI, a machine-learning based investment management platform.

Agarrwal adds that there are also issues like overfitting or underfitting of data that can lead to spurious correlations which don’t make any money when applied in the real world.

While these unregulated players highlight returns, they don’t highlight the risks in their strategies.

Unregulated players

Several unregulated platforms have sprung up on which those who have written algos offer them to the public for subscription.

“Neither the platform nor the algo provider is a regulated entity,” says Rajesh Ganesh, founder and chief executive officer (CEO), Tripleint Trading Systems.  

Such unregulated players lack accountability.

“If the algo does not perform and the investor loses money, no redress mechanism will be available to him/her,” says Sonam Srivatsava, founder, Wright Research, a Sebi-registered investment advisor

Poor-quality ratings

These unregulated platforms often provide ratings to the algos listed on them, but these are spurious or of poor quality.

“Algos that have given higher returns in the recent past are assigned higher ratings. The problem is that once the investor enters them based on these ratings, the algo often enters a drawdown cycle, producing losses,” says Ganesh.

Unreliable infrastructure

In the case of many strategies, only a few trades make large profits, while the majority makes losses.

“If the infrastructure is poor, your algo may not run on the day when profits are made, so you could end up garnering only the losses,” says Ramakrishnan Selvaraj, director, QuantIndia Systems.

What you should do

Avoid algos that have not been vetted by either an exchange or a broker. Algos run by brokers directly have Sebi approval.

It is safer to go with an algo provider that is the authorised partner of an established broker.

“The broker would have scrutinised the algo before offering it,” says Ganesh.

Check the algo provider’s credentials. He should be an expert in derivatives and equities, and should have some qualification, such as a National Institute of Securities Markets certification.

“The strategy must have a considerable amount of backtested data, and also adequate live-trading data,” says Selvaraj.

The latter will ensure it behaves in a similar manner in live conditions as it did during backtesting. Selvaraj also suggests checking out the slippage between backtested and live data.

“It is safer to go with a player that can provide audited records of returns,” says Srivastava.

Ask questions about the maximum drawdown the strategy could suffer.

The strategy must be able to perform across market cycles. It should not perform in a bull market, but suffer a steep drawdown in a bear phase from which recovery becomes difficult.

Before selecting an algo, ask for information on parameters like win-loss ratio, and also how much it wins when it wins, and how much it loses when it loses.

“An algo may promise a 95 per cent win ratio. But it may not prove to be a good strategy if it incurs big losses on the balance 5 per cent of trades,” says Ganesh.

Treat all promises of high and assured returns as red flags.

Check the algo’s trading frequency.

“A strategy that involves a higher number of trades will push up costs, thus raising the hurdle rate for it to become profitable,” says Ganesh.

When looking up returns, consider post-expense returns only.  per cent of trades,” says Ganesh.

Treat all promises of high and assured returns as red flags.

Check the algo’s trading frequency. “A strategy that involves a higher number of trades will push up costs, thus raising the hurdle rate for it to become profitable,” says Ganesh.

Dos and Don’ts of Algo Trading

• Don’t switch from one algo strategy to another: choose strategy carefully, then stick to it for at least three years  
• Enter with adequate capital—minimum Rs 5 lakh
• The amount you bet on a single trade shouldn’t exceed 2-2.5 per cent
• Don’t share your trading account’s username and pass­word with unregulated entities
• Avoid these mostly high-risk strategies involving derivatives and leverage if you lack the risk profile