Trading with machine: Algorithms now dominate all parts of financial mkts

The data from the National Stock Exchange (NSE) shows that machines now dominate every segment. Algorithmic trading was visible in all categories, including cash equity, where the majority of traders are individuals. By the end of February, algorithms were responsible for roughly 54 per cent of cash trades by value. Cash was the only segment where, until 2023-24, humans traded the bulk of the value. In other segments like derivatives — equity, commodity, currency, and interest rate — silicon has been dominant for the best part of the decade. It is likely that trading via algorithm (algo) will accelerate further since the regulator cleared the offer of algo-trading tools to retail investors by service providers. More individuals will now trade via algorithms as distinct from manually selecting and entering trades.

 

This movement towards programmed trading would further align the NSE with trends in other global exchanges. Globally, the vast majority of trades are done by machines operating with minimal or zero human supervision. The basic idea of algorithmic trading is simple and versions of it have existed for some time. The trader (who may be human or not) seeks statistical and mathematical patterns in the historical trading data and in news flow. Whenever such a pattern is visible, a trade is initiated in accordance with strict pre-programmed rules (which is the definition of an algo). The algo will usually have fail-safe systems like stop losses to limit the damage if the trade goes wrong, as well as money-management parameters determining when to average up or down, or when to book profits. Algos can also analyse impact costs (the amount a price will be affected by high-volume trades) and break up a large order to minimise such an impact. So, even if an institution such as a mutual fund is taking a large long-term position, it may use an algo to do so.

 

From the 1980s onwards, computerised trades became common on many exchanges. The infamous Black Monday crash of October 19, 1987, was apparently triggered by unsupervised algorithmic trading. Algos sold into a downtrend, triggering a crash, while many Wall Street brokers were out to lunch. The Dow Jones Industrial Average fell 23 per cent in a single session as a result.  This debacle led to exchanges putting in circuit breakers, and regulators demanding more human oversight. As computers have become more powerful, the algos have become more of a “black box”. This trend is accentuated by the induction of artificial intelligence and neural networks. This makes human oversight irrelevant since humans can’t comprehend the basis for the algo’s recommendations.

 

Arbitrages of tiny price imperfections, and responses to news flow, such as quarterly results, are now controlled almost entirely by silicon. The last frantic minutes of a derivatives expiry day also generate high volumes where no human could possibly match response times with computers. Cash was the last frontier because the cash market doesn’t have very high leverage and trades are not that time-dependent. But even here, algos provide an edge. Manual day-traders who try to exploit small price moves will be rendered redundant as algo trading becomes more prevalent. Only long-term investors who deploy “buy” and “hold” strategies are likely to be holdouts against the machines.