Before creating a system, quants will research the strategy they want it to follow. Often, this takes the form of a hypothesis. Our example above uses the hypothesis that the FTSE tends to make certain moves at particular times each day, for instance.

With a strategy in place, the next task is to turn it into a mathematical model, then refine it to increase returns and lower risk.

This is also the point at which a quant will decide how frequently the system will trade. High-frequency systems open and close many positions each day, while low-frequency ones aim to identify longer-term opportunities.

Backtesting involves applying the strategy to historical data, to get an idea of how it might perform on live markets. Quants will often use this component to further optimise their system, attempting to iron out any kinks.

Backtesting is an essential part of any automated trading system, but success here is no guarantee of profit when the model is live. There are various reasons why a fully backtested strategy can still fail: including inaccurate historical data or unpredictable market movements.

One common issue with backtesting is identifying how much volatility a system will see as it generates returns. If a trader only looks at the annualised return from a strategy, they aren’t getting a complete picture.

Every system will contain an execution component, ranging from fully automated to entirely manual. An automated strategy usually uses an API to open and close positions as quickly as possible with no human input needed. A manual one may entail the trader calling up their broker to place trades.

HFT systems are fully automated by their nature – a human trader can't open and close positions fast enough for success.

A key part of execution is minimising transaction costs, which may include commission, tax, slippage and the spread. Sophisticated algorithms are used to lower the cost of every trade – after all, even a successful plan can be brought down if each position costs too much to open and close.

Any form of trading requires risk management, and quant is no different. Risk refers to anything that could interfere with the success of the strategy.

Capital allocation is an important area of risk management, covering the size of each trade – or if the quant is using multiple systems, how much capital goes into each model. This is a complex area, especially when dealing with strategies that utilise leverage.

A fully-automated strategy should be immune to human bias, but only if it is left alone by its creator. For retail traders, leaving a system to run without excessive tinkering can be a major part of managing risk.