Frankenstein algos: Exchanges, SEBI need to spell rules for retail traders

Derivative market watchers say that in every case, the errors most likely happened because of poor risk management at the traders’ end. These incorrect trades then then got amplified as they found their way into the system and rival algo strategies tried to capitalise on the anomaly.

In algorithmic trading, a trading strategy is converted into a computer programme in order to buy / sell shares in an automated manner. Owing to its speed and accuracy, automated trading has become quite popular.

Regulatory supervision

The question that arises is if regulators such as SEBI or rule-making institutions such as stock exchanges need to intervene.

It is not clear if all or any of the traders involved in the freak trades had used algo strategies. But market regulator SEBI needs to issue clear guidelines on algo trading strategies used by retail investors. Of course, the surge in derivatives trading—options in particular—by retail investors post-pandemic is not limited to just India, it's a global trend.

Options, initially designed for sophisticated institutional investors to hedge their positions, have now evolved into a gambling tool for anybody with a few tens of thousands of rupees to spare. Individuals (retail) and proprietary traders now make up for over 85 percent of total volumes in index and single stock options trading. Average daily turnover in the options segment alone is now in the vicinity of Rs 1.2-1.5 trillion.

Also read: Fat finger error in Sensex options at 40x market price; costs trader Rs 78 lakh

This explosive growth has led to the rise of third-party algorithmic trading service providers, many of whom flout SEBI regulations by hawking the promise of guaranteed returns to retail investors who have little understanding of derivative markets. Clients who subscribe to these services then run the algos on the broker’s server through an application programming interface (API). It's open season out there for now, without any checks on the type of algos that are allowed to be run by clients. There are concerns about the disruption that can be unleashed by an algo gone awry.

That said, even large institutional players can make mistakes with their algorithmic trading programs. In August 2012, staff at US financial services firm Knight Capital Group failed to copy a critical code in an algo, in one of the servers. Consequently, Knight Capital mistakenly ended up sending 4 million orders across 154 different stocks in 45 minutes, causing prices to go haywire and incurring a loss of $440 million.

Regulators have tried to tackle the problem. In a consultation paper in December 2021, SEBI had proposed that only algos approved by a stock exchange should be deployed. Also, brokers would need to assess the suitability of an investor prior to offering an algo facility, have the technology to check for unauthorised changes to algos, must be responsible for all algos emanating from its APIs, and redress investor disputes.

Market players say there has been pushback from a section of brokers on some of the key proposals. SEBI is yet to finalise rules for the deployment of algo strategies by retail investors using third party service providers, but recurring episodes in the market could lead to renewed push from the regulator to bring in much-needed checks and balances.

The role of stock exchanges

The other issue market players feel the regulator needs to look at is the uniformity of rules across both exchanges for derivative trades. NSE’s derivative products are not traded on the BSE, and vice versa. But the key indices of both exchanges have many stocks in common. A sudden swing in the price of an options contract because of a misfiring algo or an erroneously placed order could have implications for the market.

The freak trade in the Sensex weekly options happened because a trader mistakenly placed a stop loss market order instead of a stop loss limit order. A market order will keep buying or selling irrespective of the price, till such time the order gets filled (executed per the rules defined, as in price, volume, order value, etc). In comparison, a limit order will be executed only when the specified price is reached.

The NSE had discontinued stop loss market orders two years back, but the BSE allows this, which is why the trade went through. It is not a question of which exchange is right or wrong, but whether it is okay for exchanges to set rules of this nature.

Also Read: Won’t argue over price; smallcaps, microcaps is a 10-year game: Mukul Agrawal

There is also the question of the intra-day price band of options. Are both exchanges following the same parameters when it comes to setting intra-day price bands? Have they been communicated clearly to members and are they working effectively? For instance, NSE has a limit price protection (LPP) mechanism where the reference price is the simple average of trade prices of that contract in the last 30 seconds, calculated dynamically.  The NSE website says that any incoming limit order placed beyond the LPP range shall automatically be rejected by the exchange, thus mitigating the effect of an erroneous entry by a trader.

The BSE website mentions that there are no maximum and minimum price ranges for futures and options contracts. However, to avoid erroneous order entry, dummy price bands have been introduced in the derivatives segment, the website says. It does not give any further details about the basis of calculation for the dummy price band.

Yet in each of the three abovementioned cases—across the BSE and NSE—the erroneous orders happened at prices way above or below the prevailing market price.

Price band challenge

At the same time, from an exchange’s perspective, setting intra-day price bands for options is not the same as setting price bands for stocks and futures, which is a much simpler affair. That is because movement in option premiums depend on the price changes in the underlying, and the move is never linear.  The way options are priced involves complicated math. There are many factors at play, such as the time-to-expiry and traders’ perception of future prices. A 1 percent move in the underlying can trigger a 100 percent move in the option premium.

Take, for instance, the Nifty move on September 8. The index gained a shade less than 0.5 percent, but the difference between the high and low premium on the Nifty call option of 20,000 strike was 82 percent.

Frankenstein algos?

That then begs the question: are the algos in play powerful enough to distort the dynamic price bands. For instance, when the erroneous order was placed in Bank Nifty call options at a 90 percent discount to the prevailing market price, the LPP mechanism should have rejected it outright. Did the other algos present in the system sniff out the order and immediately fill the order book with low bids, which then would have changed the price calculation for the previous 30 seconds?

Similarly, in the case of the freak BSE trade, the market order (buy) which was triggered when the price rose above Rs 4.30, should have stopped at Rs 8 or Rs 10, which is where the last of the offers should have been, given that it was an illiquid option contract set to expire in a few hours. And yet, the buy order was executed all the way up to Rs 209. If the dummy price band was effective enough to prevent orders way off the market price, then the offer at Rs 209 could not possibly have been lying in wait in the order book; it would have been fired by a rival algo later.

Derivatives market players say the nature of algos is such that they will always try to spot discrepancies in the system and then exploit it with ruthless efficiency through lightning-fast trades. The only safeguard for traders is to ensure that they have sound risk management practices in place.

Limits of regulations

But even the best of regulations and risk management systems can only do so much to avert mishaps in the marketplace. Events—manmade and market driven—do occur once in a while. For instance, in 2010, Navinder Singh Sarao, an Indian-origin trader based in the United Kingdom, triggered a flash crash in US markets after the algorithm he had developed to create an illusion of large orders, went out of control.

Sarao’s technique was to place large orders in the e-mini S&P contract and cancel them just before they got executed. This technique, known as spoofing, caused prices of the contracts to move in a certain direction as other algos tried to pounce on the orders. When the prices moved, Sarao benefited from another set of orders his algorithms had placed, anticipating the move. The flash crash caused wild gyrations in US stocks for around 36 minutes, during which time prices first plunged and then rebounded.

All greed, no fear

Veteran traders say that one of the key reasons for the rush of retail investors into options trading post pandemic is that there have been no extreme events in the market for over three years now. That has led to a growing perception that it is possible to make steady returns through options trading without taking on much risk.

But these traders would do well to read up on Volmageddon 2018. A large number of investors in the US took positions that were essentially short volatility index (VIX) bets. These investors were betting that market volatility would remain low or continue to decline, and used various instruments, including VIX futures and options, to profit from such a scenario. On February 2, the US market fell sharply, causing the VIX to spike dramatically, catching sellers of volatility options off guard.

This triggered margin calls, forcing many of them to sell off assets to meet their margin requirements. This had a domino effect as selling pressure further exacerbated the market decline, as more investors rushed to exit their positions causing a broader market sell-off.

The recurrence of such incidents erodes the faith of investors in the market, and while these freak cases may have been just that, it's also an invitation for rogue elements to try and exploit these vulnerabilities. Retail investors should, of course, exercise diligence while trading in the high stakes, high-risk derivatives market, but SEBI and the stock exchanges too need to do better to make it difficult for freak trades generated by algos to run amok.