Algorithmic trading is the use of computer programs that place orders on an exchange automatically, by following fixed rules. Algo trading is the short form. This page is about execution algorithms, the kind that large funds use to buy or sell a very large number of shares without pushing the price against themselves.
Here is the problem they solve. Suppose a fund sends one order to sell 10 lakh shares at once. The order would use up every waiting buyer in the matching engine's order book, at lower and lower prices. The price would fall sharply, and the fund would receive less for its later shares. This effect is called market impact.
An execution algorithm avoids this by splitting the order into many small orders and sending them over time. There are several standard methods. VWAP, the volume-weighted average price, sends more shares at busy times of the day and fewer at quiet times. TWAP, the time-weighted average price, sends equal slices at equal intervals. Other methods react to prices as they change.
Consider a mutual fund in Mumbai that decides to sell 10 lakh shares of a company. About 50 lakh of its shares are normally traded in a day. The fund's investment management system sends the order to a broker's algorithm. The algorithm aims to be about 20% of each minute's trading. It sends orders of a few hundred shares every few seconds, through the whole day. At the end of the day, the fund measures the cost. When it decided to sell, the price was ₹403.00. Its average selling price was ₹402.30. The difference of 70 paise on 10 lakh shares is ₹7 lakh, or about 0.17%. This measurement is called transaction cost analysis.
Professional traders choose and watch these algorithms on a trading platform for institutions. Quantitative hedge funds use them as well.
Execution algorithms are different from high-frequency trading (HFT). An HFT firm trades its own money to make a profit from tiny price differences. An execution algorithm carries out a decision that a client has already made, at the lowest possible cost. It is also different from algo trading for retail investors, which decides when an individual should place a bet.