cheatsheet

The Seven Sins of Quantitative Investing

Based on Deutsche Bank Quantitative Strategy Research

This cheatsheet outlines the seven most common biases and mistakes (“Sins”) that occur during the backtesting of quantitative investment models, along with their remedies.


1. Survivorship Bias

The Sin: Considering only companies that are currently in existence, ignoring those that have gone bankrupt, merged, or been delisted. This artificially inflates historical performance because “dead” companies usually performed poorly.

Examples:

The Remedy:


2. Look-ahead Bias

The Sin: Using information in a backtest that was not actually available to investors at that time.

Examples:

The Remedy:


3. The Sin of Storytelling

The Sin: Post-hoc rationalization. Creating a plausible “story” to explain a statistical anomaly after seeing the results.

Examples:

The Remedy:


4. Data Mining (and Snooping)

The Sin: “Torturing the data until it confesses.” Running hundreds of variations and picking the best one, leading to overfitting.

Examples:

The Remedy:


5. Signal Decay and Turnover

The Sin: Ignoring transaction costs and signal decay speed. High theoretical returns can turn into losses after costs.

Examples:

The Remedy:


6. Outliers

The Sin: Mishandling extreme data points. Including errors ruins results; excluding valid extremes removes risk info.

Examples:

The Remedy:


7. The Asymmetric Pattern and Shorting Cost

The Sin: Assuming Shorting is as easy/cheap as Buying.

Examples:

The Remedy: