Factor Construction with Different Lookbacks
Does the lookback matter for stock selection?
September 2024. Reading Time: 10 Minutes. Author: Nicolas Rabener.
SUMMARY
- Intuitively quant strategies should benefit from new information and shorter lookbacks
- However, fundamental-based factors improve from longer lookbacks given more robust signals
- Less relevant for price-based factors like momentum or low volatility
INTRODUCTION
The world is a complex system that never stands still and continuously evolves, which makes investing interesting but also challenging. Companies have to change each day to survive in the brutally competitive economy. Most active fund managers have chosen to play along and prioritize the latest quarterly earnings over the long-term outlook of companies.
However, when running quantitative strategies this short-term focus is questionable. Naturally, there are some short-term-oriented trading strategies focused on earnings and news, but most focus on companies’ fundamentals or stock price patterns over the medium to long term. Quants want robustness, which requires reducing false signals, e.g. Tesla having one great quarter does not necessarily make it a great company. Given this, stocks are typically selected by looking at fundamentals that stretch back one year or longer.
In this research article, we will review the impact of the lookback period of factor portfolio construction.
VALUE FACTOR CASE STUDY
We focus on five factors in this analysis, namely momentum, value, quality, low volatility, and growth, which are defined in line with industry standards. The investible universe is defined as all stocks in the U.S. with a market capitalization of more than $1 billion. We construct factor portfolios by selecting the top and bottom 10% of stocks ranked by the factor metrics, which are then rebalanced monthly with a transaction cost of 0.10%.
Value stocks are selected based on a combination of price-to-book and price-to-earnings, where we vary the lookback for these metrics between one and five years. Given quarterly earnings, this means four updates for book equity and earnings when using one year, and twenty when using five years. Naturally, the market capitalization of stocks changes daily, so even without fundamental updates, the metrics for each stock will change at the monthly rebalancing (read Smart Beta ETF Construction: High versus Low Factor Exposures).
The long-short value factor has performed poorly in the period from 2007 to 2024, but the trends in the performance of these factor variations were similar, which indicates that the lookback period does not matter much. However, there is more similarity between the three and five-year variations, which is expected as more data points were available for the stock selection, i.e. making the process more robust for identifying cheap and expensive stocks.
Source: Finominal
Next, we evaluate the performance of the long-only value factor in the U.S., which again shows that the portfolios with three and five-year lookbacks for price-to-book and price-to-earnings were more similar than the one using a one-year lookback. A shorter lookback would give more weight to new information than a longer one.
Source: Finominal
LOOKBACK VARIATIONS ACROSS FACTORS
We compute CAGRs, volatility, and Sharpe ratios for five factors for the period between 2007 and 2024, where the returns for value, quality, and growth were higher when using a five-year lookback. The Sharpe ratios were the same or improved for all except momentum.
It is worth highlighting that a longer lookback makes less sense for momentum than for fundamental-based factors like value or quality. The latter might benefit as the stock selection becomes more robust given more data points, but there is plenty of research that indicates that momentum works best with a lookback between 9 and 18 months across asset classes, perhaps due to the 12-month lookback being the most used frequency for analyzing performance. On shorter and longer lookbacks mean-reversion tends to work rather than momentum.
Source: Finominal
Moving from long-short to long-only factors portrays the same perspective, i.e. a longer lookback results in higher CAGRs and Sharpe ratios, except for momentum.
The difference in returns is relatively minor for the low volatility factor, which is intuitive as the benefit of extending the lookback window should be less relevant given daily returns. However, there should also be no downside in extending the lookback when trying to differentiate between highly and low-volatile stocks.
Source: Finominal
FURTHER THOUGHTS
It seems the focus on the latest earnings is less relevant for traditional factor investing strategies and having a longer-term lookback actually benefits fundament-based factors as it improves the signal strength.
However, the difference in returns and Sharpe ratios was not very large, except for the growth factor, which improved considerably when going from a lookback of one to five years. Somewhat ironically this confirms that the focus on the latest earnings is misguided and investors would benefit more from adopting a long-term perspective.
RELATED RESEARCH
Smart Beta ETF Construction: High versus Low Factor Exposures
Factor Investing Is Dead, Long Live Factor Investing!
How Painful Can Factor Investing Get?
Combining Smart Beta Funds May Not Be Smart
Factor Construction: Portfolio Rebalancing
Factor Construction: Portfolio Scenarios
Smart Beta vs Factors in Portfolio Construction
ABOUT THE AUTHOR
Nicolas Rabener is the CEO & Founder of Finominal, which empowers professional investors with data, technology, and research insights to improve their investment outcomes. Previously he created Jackdaw Capital, an award-winning quantitative hedge fund. Before that Nicolas worked at GIC and Citigroup in London and New York. Nicolas holds a Master of Finance from HHL Leipzig Graduate School of Management, is a CAIA charter holder, and enjoys endurance sports (Ironman & 100km Ultramarathon).
Connect with me on LinkedIn or Twitter.
