Quantitative and Factor Investing
10 articles
The documented return factors, and the far longer list of ways a backtest can be wrong before anybody trades it.
Advanced
Detail, edge cases and methodology.The Value Factor
The tendency for cheap securities to outperform expensive ones over long periods. One of the most studied effects in finance, and one that spent a decade not working.
The Momentum Factor
Securities that have performed well over the past several months have tended to continue over the following months. Robust, widely replicated, and prone to sudden severe reversals.
Quality and Profitability
Profitable, stable, conservatively financed companies have outperformed their opposites. The effect is well supported and the definition of quality is not standardised.
Size and Low Volatility
Two of the original documented effects. One has weakened substantially since publication; the other contradicts the basic risk-return relationship and has persisted.
Factor Cyclicality
Every documented factor has spent long periods underperforming. The horizon required to evaluate one is longer than most people's patience, which is part of why the premiums may persist.
Backtesting
Testing a rule against historical data. The exercise is essential, straightforward to perform, and extremely easy to perform in a way that produces a meaningless result.
Overfitting
A model complex enough to describe the past perfectly has described the noise as well as the signal, and the noise does not repeat.
Survivorship and Look-Ahead Bias
Two errors that inflate results silently. One tests only the securities that made it; the other uses information that was not available at the time.
Modelling Transaction Costs
The gap between a backtest and a live result is usually costs. They are certain, they scale with turnover and size, and they are the most commonly omitted input.
Walk-Forward Testing
Fitting on one window and testing on the next, repeatedly. It is the closest a historical test can come to simulating how a strategy would actually have been run.