Always eclectic, Yuval’s first loves were birds, books, and mathematics. After completing his undergraduate degree at Princeton and an MA from the University of Iowa, he worked for thirty years as an editor of non-fiction books specializing in music, film, and African American history. He also wrote three non-fiction books published by W. W. Norton, most recently Zora and Langston: A Story of Friendship and Betrayal, as well as articles in The New York Times Book Review, The Guardian, and elsewhere. Yuval began trading in 2013 while spending a year in Bolivia, where he had extra money and time to experiment with algorithms. Six years later, he left publishing and made a big career change, taking a position at Portfolio123, a small financial technology firm. Yuval’s earnings from investing have allowed him and his wife to retire and create a family foundation for charitable giving to support underserved communities. Yuval handles all of Fieldsong’s investing activities and continues to write.
Scott Beavers first learned about calculated risk and variable returns during a hiatus from college when he played professional blackjack and developed systems to automate card counting. After completing his degree in computer science, he took an R&D position at Applied Biosystems, a research instrumentation startup and the world’s primary supplier of genomic sequencing equipment. Scott spent 20 years coding and managing multidisciplinary development projects, training customers in DNA synthesis and helping set up Celera Genomics, where the human genome was first sequenced. As part of his management accreditation, Scott got an MBA and began working on joint ventures and university research collaborations, developing systems to assess commercialization prospects in frontier technologies. While completing a Master’s in Computer Science during the Global Financial Crisis, Scott tripled the household retirement accounts and realized that investing could provide an independent livelihood. His collaboration with Yuval began in 2018 when he wrote an automation suite that enabled the rapid iteration of backtesting and development runs with Portfolio123. Scott handles all of Fieldsong’s non-investing-related operations.
Factor Design
Fieldsong’s investing decisions are based on quantifiable factors. Some of these, such as earnings yield, are commonly used. But others have an edge. One of Fieldsong’s innovations lies in the use of novel factors such as:
• Share turnover (the volume of shares traded divided by the shares available to trade). Low share turnover signifies that a company is relatively impervious to the craziness of the market.
• Operating cash flow to enterprise value. While many investors favor the ratio of free cash flow to enterprise value, it may be even more important to consider operating cash flow, which is measured prior to the subtraction of capital expenditures.
• Growth in operational liabilities. A company that is increasing items like accounts payable, unearned income, accrued expenses, and dividends payable in proportion to its total assets is likely also increasing its operational efficiency.
• Subsector momentum. The performance of an industry group over the last nine months is a strong indicator of that industry group’s performance over the next few months.
• Sales acceleration. We hear a lot about sales growth, but companies with very high sales growth over the past year or two are more likely to fall in price than to rise. Let’s call x the most recent quarter’s sales growth over the same quarter last year and y the sales growth of the last twelve months over the previous twelve months. Then a high value of (x – y) / |y| can indicate a real turnaround in sales. This inflection point is where a smart investor can make money.
These are just a few of the many innovative factors that Fieldsong’s ranking systems rely on. In addition, we use a lot of factors that favor either middling (not too high or low) or stable (unchanging) values. This too sets us apart from most investors.
These are just a few of the many innovative factors that Fieldsong’s ranking systems rely on. In addition, we use a lot of factors that favor either middling (not too high or low) or stable (unchanging) values. This too sets us apart from most investors.
How Multifactor ranking works
Let’s say you were putting together a fantasy basketball team based on statistics. Among the things you want to consider when choosing players are their points, rebounds, assists, blocks, 3-pointers, and steals per game; you also want to consider their heights, age, field goal percentage, and games played. How are you going to put all those things together? Well, you could assign each player a number between 0 and 100 on each of those things, multiply each one according to the factor’s importance, and then add all the results together. That’s exactly what multifactor ranking does.
Let’s say you were putting together a fantasy basketball team based on statistics. Among the things you want to consider when choosing players are their points, rebounds, assists, blocks, 3-pointers, and steals per game; you also want to consider their heights, age, field goal percentage, and games played. How are you going to put all those things together? Well, you could assign each player a number between 0 and 100 on each of those things, multiply each one according to the factor’s importance, and then add all the results together. That’s exactly what multifactor ranking does.
Why ranking works
•Ranking enables one to consider hundreds of different factors at once.
•Screening will rule out a stock that just barely fails one rule, while ranking will include it if it’s strong on other factors.
•By varying the weights of factors, ranking allows you to emphasize certain factors over others.
•If you create a screen for good stocks and another screen for bad stocks using different factors, one stock could pass both screens and other stocks could pass neither. With ranking, every stock has its place: good stocks rank high and bad stocks rank low.
•Academics prefer multiple linear regression, but multifactor ranking is almost impervious to outliers and is therefore far more robust. Multiple linear regression models are more subject to overfitting/data mining.