A fundamentals-based quant portfolio built on algorithmic ranking across more than 200 metrics and holding more than half its capital in just eight companies: that is the model Yuval Taylor, a hedge fund manager and analyst, described at The Zurich Project, and it raises questions worth considering for any UK investor thinking carefully about concentration risk and systematic discipline.
Taylor has been using multifactor ranking systems to select stocks since 2015, with a particular focus on microcap companies. His method eschews qualitative judgement entirely: no management meetings, no narrative assessment, no gut feel. Every holding is chosen through comparative, algorithmic analysis designed to surface what he describes as undervalued, safe, and boring companies.
Concentration and the Case for a Fundamentals-Based Quant Portfolio
The concentration here is worth pausing on. With more than 50% of the fund in eight positions, the approach sits well outside what most UK wealth managers would recommend for private investors, particularly those drawing income in retirement. A concentrated fundamentals-based quant portfolio can generate strong returns when the ranking system is sound, but it also amplifies drawdown risk if several holdings move against the investor simultaneously. Sequence-of-returns risk, which is the danger that a large loss occurs early in a drawdown phase, is a real consideration for anyone replicating this kind of structure inside a SIPP or ISA.
Taylor’s framework does address some of those risks by design. Sector-relative analysis, fraud detection screens, and a preference for stability over raw growth are all built into the ranking process. The system covers nearly 10,000 stocks, each rated from zero to 100 on a weekly basis, which provides a degree of portfolio-level discipline that a purely discretionary approach would struggle to match at scale.
Why Algorithmic Discipline Matters, and Where It Has Limits
One of the more counterintuitive positions Taylor takes is that machine learning and large language models are currently inferior to multifactor ranking for stock selection. His reasoning centres on transparency and accounting rigour: algorithmic ranking models can be interrogated and stress-tested in ways that neural networks generally cannot. For a conservative investor who needs to understand why a position is held, that transparency has genuine value.
The limits, however, are equally worth stating plainly. Backtesting, however robust, cannot account for regime changes, liquidity crises, or the kind of correlated selling that hits microcap stocks hardest during a market dislocation. Outlier correction, which Taylor identifies as a key lesson from his own process, suggests the model has already encountered some of those edge cases.
UK private investors considering a systematic, quantitative approach should also be aware that the Financial Conduct Authority imposes specific obligations on regulated fund managers that may differ from the environment in which an individual replicates a similar strategy inside a personal wrapper. The rules governing what constitutes a collective investment scheme are not always straightforward.
Taylor publishes his ratings weekly through his investing group on Seeking Alpha, where the full methodology is available for scrutiny. For a UK investor evaluating any systematic approach, that transparency is precisely where due diligence should begin.
