Time Series for Quantitative Analysts
Time series analysis for quants is the study of data ordered in time, such as prices, returns and volatility. It covers stationarity, autocorrelation, ARIMA and GARCH models, and machine learning forecasts, with strict care to avoid look-ahead bias and overfitting when validating on past data.
Stage: Data and Strategy · Part of the Quantitative Analyst roadmap · Last reviewed 19 November 2026
What to learn in Time Series
- Test a series for stationarity with an ADF test and difference if needed
- Read ACF and PACF plots to choose ARIMA orders
- Fit a GARCH model to capture volatility clustering
- Split data by time and use walk-forward validation
- Engineer lagged features for a machine learning forecast
- Avoid look-ahead bias when building features from past data
Time Series interview questions
Why do we model returns rather than raw prices?
Prices are usually non-stationary, with trends and changing variance, which breaks the assumptions of many statistical methods. Returns, especially log returns, are closer to stationary and are additive over time, so they are easier to model and compare.
What is look-ahead bias?
It occurs when a model uses information in training or backtesting that would not have been available at that time, for example normalising with the full-sample mean. It inflates results and makes a strategy look better than it could be in live trading.
Where Time Series fits in the Quantitative Analyst roadmap
Time Series belongs to the Data and Strategy stage (advanced level). From research to production.
Other units in this stage
Back to the full Quantitative Analyst roadmap