Algorithmic trading

In simple terms

Algorithmic trading means using computer programs to decide when and how to place orders, following set rules. A large share of trading in modern markets is done by algorithms. They range from simple tools that split a big order into small pieces to complex systems that trade thousands of times a second.

Kinds of algorithms

Execution algorithms

A fund buying a large quantity of shares would move the price against itself if it placed one big market order, as the microstructure lesson shows. Execution algorithms spread the order:

• TWAP (time-weighted average price): equal pieces over a period of time.
• VWAP (volume-weighted average price): pieces in proportion to typical trading volume, aiming for the day's average price, as the VWAP lesson explains.

These algorithms aim to trade efficiently, not to predict where prices will go.

High-frequency trading

High-frequency trading firms rely on:

• Speed, measured in tiny fractions of a second.
• Location: placing their computers physically close to the exchange's systems.
• Scale: millions of small trades, each with a tiny expected profit.
• Investment in technology, data and specialist staff.

These advantages are not available to individuals using retail platforms. An individual cannot compete on speed with such firms.

Algorithms sold to individuals

Brokers and vendors offer algorithms and automated strategies to individuals. SEBI and the exchanges set rules on how such algorithms may be offered and run, including requirements for brokers and providers. These rules have changed over time, so check SEBI's current framework before using any such product.

Questions to ask:

1. What is the strategy, and why should it work? An algorithm only automates a strategy. If the strategy has no edge after costs, automating it loses money faster.
2. How was it tested? Backtests can be overfitted or cherry-picked, as the misleading backtests lesson explains.
3. Is the provider registered and the product offered within the rules?
4. What are the costs, including subscriptions, brokerage and slippage?
5. What happens in a crash or a technical failure?

Claims of guaranteed or very high returns are a warning sign of fraud, as the scams lesson explains.

Risks of automation

• Errors in code can place unintended orders quickly.
• Unusual market conditions can make a strategy behave in ways never tested.
• Many algorithms reacting to the same signals can amplify sudden moves.
• Over-reliance: traders may stop monitoring positions.

Why it matters

Algorithms shape how modern markets work: tighter spreads in normal times, fast reactions, and occasional sudden moves. Understanding who uses them, and what they can and cannot do, protects individuals from the belief that buying an algorithm buys an edge.

Limitations

This lesson describes algorithmic trading in general terms. Regulations, exchange systems and market practices change frequently, and specific products differ widely.

Common mistakes

• Believing an algorithm creates an edge by itself.
• Trusting backtested returns without independent checks.
• Ignoring costs and slippage in automated strategies.
• Using unregistered providers.

Try it yourself

1. Explain in your own words the difference between an execution algorithm and a strategy algorithm.
2. Find SEBI's current framework for algorithmic trading by individuals on its website, and note two requirements.
3. Take any algorithm advertisement and list the questions above that it does not answer.