RT Journal Article SR Electronic T1 Alpha Cloning: Using Quantitative Techniques and SEC 13f Data for Equity Portfolio Optimization and Generation JF The Journal of Financial Data Science FD Institutional Investor Journals SP jfds.2019.1.008 DO 10.3905/jfds.2019.1.008 A1 Daniel M. DiPietro YR 2019 UL https://pm-research.com/content/early/2019/08/28/jfds.2019.1.008.abstract AB Securities and Exchange Commission (SEC) 13f filing data offer valuable insight into top asset managers’ holdings at each quarterly filing point. Using these data, several quantitative statistical models were designed and back tested, both for portfolio optimization and generation. All models for portfolio optimization outperformed a traditional market cap weighting strategy in terms of Sharpe ratio but did not outperform traditional market cap weighting in returns. Portfolio generation models yielded mixed results: Some failed to break even, whereas others outperformed the S&P 500 by a large margin. The most successful generation model, firm-specific overweighted investment extraction with market cap balancing, consistently generated more than 85% in returns over the five-year backtested period, even surpassing 95% for various numbers of holdings; in comparison, the S&P 500 yielded approximately 72% over the same period. Overall, these results indicate that SEC 13f data can be used to construct quantitative models for long-term equity investing that achieve noteworthy performance.TOPICS: Statistical methods, simulations, big data/machine learningKey Findings• SEC 13f data can be used to construct quantitative models that generate equity portfolios of various sizes that outperform the S&P 500 in historical backtesting.• The most successful 13f portfolio generation models rely upon identifying equities in each examined asset manager’s portfolio that are overweighted relative to a benchmark weighting approach.• SEC 13f data can be used to optimize holdings in an existing portfolio; all proposed portfolio optimizers outperform traditional market cap weighting in terms of Sharpe ratio in historical backtesting.