Destructive calculus: How algorithms put jobbers out of job

It was a few minutes past 10 in the morning—the stock market had just opened. Nithin Srivastava's eyes glanced at the screen, absorbing the flickering numbers, his fingers rattling across the keyboard in an astonishing display of handeye coordination, banging out trading orders in rapid succession. In another 15 minutes, his day was pretty much done.

He'd made about Rs 30,000—an average performance in 2007. On exceptional days, he could even make as much as Rs 2-3 lakh. But the party would soon be over for jobbers and arbitrageurs like Srivastava. Not because of the financial crisis that would demolish markets around the world shortly but because algorithmic trading would make most of them redundant.

These days Srivastava manages awarehouse that stores almonds, having quit as an arbitrageur from BLB Securities in 2012. "Big brokers like BLB had 1,000-1,200 jobbers and arbitrageurs those days," he recalled. "Now they may only have a handful."

The keyboard skills needed to be successful at jobbing and arbitraging had to be honed over the years—market instinct combining with sheer typing ability. A junior scalper would not even get a sniff of hard cash until he was able to punch out orders in less than three seconds. The best in the business did this in oneand-a-half seconds.

"Years 2006 and 2007 were the best years for jobbers and arbitrageurs," said Chirag Mehta. "Those were the years when jobbers profited from price differentials (between NSE and BSE or between shares and futures) as high as 50-75 basis points," who currently trades with Touchline Securities.

A basis point is 0.01 percentage point.

The bull run nine years ago was great for this group. On a normal trading day, these men—and they were all men—made as much as Rs 40,000. In times of extreme volatility, this could surge to Rs 3 lakh.

"The market opened at 9:55AM those days. By 10:15, some of us would have made Rs 25,000-40,000," said Srivastava. "Some of those large brokers employed 300-500 pairs of arbitrageurs to derive maximum benefits of volatility."

Paired arbitrage refers to simultaneous buying and selling of instruments by two arbitrageurs to eke out profits arising out of a price differential between exchanges or instruments.

But by late 2007, superfast trading terminals loaded with complex algorithms, or algos, began making inroads into the Indian equities markets. The fastest fingers could not match the prowess of the algorithms, which executed millions of trades in milliseconds. Over time, with more and more institutional players signing up for algorithmic trading, price differentials between exchanges and products became nonexistent.

"Jobbers and arbitrageurs (initially) believed the situation would improve when markets turned around," said Suresh Mehta of Dhyan Stock Broking. "But when nothing happened till 2011-12, a good number of them moved on to other trades."

Mehta's company was forced to shut the division.

"At one point, we had over 1,000 jobbers and arbitrageurs on our rolls," Mehta said. "Today, we have none... We moved out of jobbing and arbitraging in 2011 as there was no point carrying on a lossmaking business."

D-Street to high sea

Their ranks haven't been completely wiped out but the numbers have dwindled to an estimated 2,000 from more than 30,000 in 2003-07. According to brokers, a large chunk of the tribe moved on to businesses such as real estate broking, diamond polishing, plywood and textiles. People with less appetite for entrepreneurship took up jobs as relationship managers and dealers.

Dinesh Kumar used to be a jobber with BLB Securities. He's now in the merchant navy. He has clear memories of the end of days.

"Our bad times started in 2008. Markets crashed in the third week of January. The bourses were flooded with sell trades. Turnover volumes dropped," he said. "Towards mid-2008, large brokers employed software and algos to squeeze small gains. This pushed us into deeper misery. By 2010, manual arbitraging had become a loss-making trade. We could not compete with the machines. By 2012, none of us were making any money."

Kumar got out in 2012, after getting the merchant navy break.

"What I am earning now is not even a fraction of what I made from trading in 2007," said Kumar, now 42.

Middle-aged jobbers, who only possessed good typing skills, stayed on doing 'BTST' (buy today, sell tomorrow) trades. On good days, BTST traders made profits of Rs 5,000-10,000. But taxes, broker charges and exchange fees, among other levies, ate into their earnings significantly.

"Algorithm trading dealt the first body blow to jobbers and arbitrageurs. Imposition of STT (securities transaction tax, in the FY2005 budget) moved them out of business," said Kapil Mohan, who's now empanelled with Vikabh Securities and manages the investments of a few Delhi-based investors.

"When times were good, jobbers earned commission and profit share from brokers. The entry of algo trading pushed jobbers and arbitrageurs to take up jobs with brokers at very low salaries," Mohan said.

When day traders transitioned from commissions to salaries, their earnings dropped significantly. Dealers and proprietary book traders earned about Rs 35,000-40,000 in monthly salaries— the sort of money jobbers blew up to celebrate big profit days during the bull run.

The few people still in the trade will decline further, said veteran Delhi-based broker BL Bagri.

"Algorithm trading was inevitable at that time—nobody could have done anything to prevent it from coming," Bagri said. "This trend of automation may continue for a long, long time. It will end whatever is left of manual arbitraging and jobbing."

Currently, algorithmic trading accounts for more than 40% of transactions in the Indian equities market. The rise of the machines is such that even cash-rich brokers aren't able to keep pace. Market intermediaries expect net algorithmic volumes to breach 60% by 2020.

The future

"There's no end to this fight for more trade execution speed. Brokers are forced to update their IT peripherals every three months to match speeds of their competitors. Competition exists even among colocated brokers," said Praveen Gupta, CEO of Symphony Fintech, a trading solutions provider.

In their bid to spread the cost of service and frequent technology updation, brokers have started opening up algorithmic solutions to retail investors at a marginally higher cost. About 5% of retail trades are made using algorithms.

"The concept of speed (high-frequency trading) may take a back seat in the coming years," said Gupta of Symphony Fintech. "We'll see more intelligent algos hitting the market over the next few years."

While brokers have embraced algorithmic trading, old-timers, conservative market participants and regulators have apprehensions about the unbridled power that they can unleash. Machine trading is said to have caused a number of crashes around the world. In 2014, a quant-based portfolio management services fund suffered losses due to malware and wrong 'factors' in the algorithm. "Algorithmic trading does not pose a lot of threat if good risk management practices are in place," said Vikas Khemani, CEO of Edelweiss Securities.

"It (algorithmic trading) is all about efficiency... Markets across the world are opening up to highfrequency intelligent trading."

As that happens, those who were first with the fastest finger will become a footnote in the history of India's stock markets.