Hybrid Relative Value Model
An In-Depth Overview of the PoMaTo Hybrid Relative Value Equity Framework
1. Executive Summary
2. PoMaTo HRV Equity Model Overview – Framework Architecture
3. Introduction
3.1. Data Collection
3.2. Peer Groups Definition
3.3. Balance Sheet analysis
3.4. Forecasting Framework
3.5. Relative Value Framework
3.6. Alpha Generation
4. Uniqueness of the HRV Framework
5. Examples: Model vs Actual
By Fabio Agostini
1.Executive Summary
Traditional investment processes often combine portfolio construction, risk management, and alpha generation into a single framework. The HRV framework separates these functions.
Risk architecture defines the structural portfolio constraints and diversification profile, while the alpha engine generates differentiated investment signals through a relative value process.
Hybrid Relative Value (HRV) represents one implementation of this modular alpha layer. The framework combines bottom-up fundamental analysis with adaptive quantitative techniques to identify relative mispricing opportunities across peer groups and industry structures.
The objective is not to forecast markets broadly, but to identify companies whose current valuation appears inconsistent with their underlying financial trajectory and relative positioning. The framework is designed to adapt across changing market conditions while maintaining a disciplined and consistent analytical process.
2.PoMaTo HRV Equity Model Overview – Framework Architecture
The HRV framework separates stable portfolio risk architecture from the adaptive alpha layer.
This modular structure allows the investment process to evolve without requiring structural changes to the portfolio construction framework.
3.Introduction
The HRV model focuses on identifying companies with attractive asymmetric risk adjusted return characteristics through a disciplined bottom-up framework.
Rather than relying primarily on broad market or sector directionality, the model evaluates individual businesses relative to carefully defined peer groups.
The analytical process combines balance sheet analysis, forward-looking financial estimation, relative valuation techniques, and adaptive factor sensitivities to identify relative mispricing opportunities.
The framework seeks to maintain consistency in risk discipline while allowing the alpha layer to evolve across changing market environments.
3.1.Data Collection
The model begins by defining a broad universe of companies to evaluate.
It then gathers detailed financial information, including balance sheet and income statement data, and uses this to calculate a set of key indicators that help assess each company’s underlying value.
In addition to fundamental data, the model also incorporates market information such as share price, market capitalization, shares outstanding, and trading activity.
By combining these inputs, the model builds a comprehensive view of each company to support consistent and disciplined analysis
3.2.Peer Groups Definition
The model is designed to identify companies with attractive risk-return profiles through a disciplined, bottom-up analytical framework. Its outputs are driven by individual company analysis rather than broader market or sector trends.
It begins by defining a broad universe of companies for evaluation. From this universe, the model gathers detailed financial information, including balance sheet and income statement data, and derives a set of key indicators used to assess each company’s underlying value. In parallel, it incorporates relevant market data such as share price, market capitalisation, shares outstanding, and trading activity to build a comprehensive view of each company.
To ensure meaningful comparisons, companies are organised into carefully defined peer groups based on their core business activities. This classification is established through a detailed review of company disclosures, including annual reports and publicly available information, allowing for consistent, like-for-like analysis.
The classification is applied at a granular level, grouping companies into sub-industries, which roll up into broader industry groups and sectors. Building on this structure, a proprietary classification framework enables systematic and rigorous comparison of balance sheets across peers.
This framework supports multiple layers of analysis, including comparisons at both industry group and sub-industry levels, and underpins the identification of relative value opportunities. The model ultimately highlights companies that appear undervalued relative to their intrinsic worth, with a focus on those offering an attractive balance between potential upside and downside risk.
While the model can be used in a benchmark-aware context, it is not constrained by benchmark compositions. As a result, its outputs may lead to differentiated positioning and the identification of high-conviction opportunities.
3.3.Balance Sheet Analysis
The analysis of balance sheet data is central to the model’s ability to assess both a company’s historical performance and its future potential.
By comparing companies within the same industry group and sub-industry, the model identifies those with the potential to deliver superior earnings relative to their peers.
A key component of this process is the model’s own forward-looking estimates of critical balance sheet items.
These forecasts form the foundation of the analysis and play a central role in identifying companies that appear undervalued and well-positioned for future profitability
3.4.Forecasting Framework
The estimation model focuses on forecasting key balance sheet items for the next quarter only. This reflects the view that longer-term forecasts are inherently less reliable, as they are more exposed to unexpected macroeconomic events and changing market conditions.
By concentrating on the near term, the model aims to produce more robust and consistent estimates of company performance.
The model generates forward-looking estimates for a range of financial metrics, most notably:
- Revenue
- Earnings
In doing so, it focuses on the core activities of each company, excluding non-recurring or ancillary sources of income. This ensures that comparisons across companies remain consistent and meaningful.
For certain sectors, the model also incorporates industry-specific indicators. For example:
- in banking, it analyses net interest income
- in insurance, it evaluates net premium income
To produce its estimates, the model analyses historical quarterly data, identifying both underlying trends and recurring seasonal patterns. By combining these elements, it builds a forward-looking view of each company’s near-term performance.
For instance, some companies exhibit consistent seasonal patterns, where revenues and earnings are typically stronger in specific quarters. The model captures these dynamics and integrates them into its forecasts, alongside the broader trend observed over time.
This same approach is applied consistently across all key financial metrics. The resulting estimates also feed into related measures, such as expected dividends and changes in book value.
To ensure consistency across the investment universe, the model operates on a quarterly cycle, once most companies have reported their financial results. This allows for like-for-like comparisons across companies based on the same reporting period.
Companies that have not yet reported at the time of the model run are temporarily excluded from the analysis. This ensures that all companies included are evaluated using a consistent and complete dataset, avoiding distortions in the results.
Once new financial data becomes available, these companies are reintroduced into the model, ensuring that the analysis remains up to date and reflective of the latest information
3.5.Relative Value Framework
To generate buy and sell signals, the model applies a relative valuation approach, comparing each company’s financial profile with that of its closest peers at the sub-industry level, while also considering the broader economic environment.
The objective is to identify companies that appear undervalued relative to their peers. These companies form the investable universe of potential opportunities, representing candidates for inclusion, while those that appear less attractive form the basis for potential reductions or exclusions.
To ensure balanced comparisons, companies are evaluated within groups of similar size, based on market capitalisation. This helps avoid unintended biases and ensures that companies are assessed on a like-for-like basis.
The relative value model combines both value and growth perspectives, using a standardised financial view of each company. This is constructed by aggregating reported quarterly data with the model’s forward-looking estimates, creating a consistent and comparable snapshot of financial performance.
By bringing these elements together, the model identifies relative mispricing across companies and highlights opportunities where valuation and growth characteristics appear most attractive within each peer group.
3.6.Alpha Generation
While the model identifies a set of attractive candidates, the next step is to estimate the intrinsic value of each company.
To do this, each company’s financial profile is analysed relative to its broader industry. The objective is to understand how financial performance translates into market valuation and, ultimately, into future price potential.
The model applies a statistical approach to quantify these relationships, using multiple financial indicators derived from balance sheet and income statement data. This allows it to assess how changes in a company’s fundamentals have historically influenced its share price.
The analysis is conducted at the industry level and is based on consistently constructed financial data, combining reported results with a standardised view of performance over time. The model focuses on changes in key financial metrics and evaluates how these changes relate to movements in market prices.
By analysing these relationships, the model estimates how sensitive a company’s valuation is to each financial factor. These sensitivities are then applied to current and expected changes in fundamentals to derive an estimate of future price.
The difference between the current price and the model’s estimated future price represents the expected return, or alpha, for each company.
Importantly, the model does not impose predefined weights on individual factors. Instead, it allows the data to determine which financial drivers are most relevant at any given time, reflecting changing market conditions and investor preferences.
All analysis is based on information available at the time, using original company disclosures. This ensures consistency with how markets price securities in real time, without relying on retrospective adjustments.
4.Uniqueness of the HRV Framework
Our model is, by design, a hybrid approach—a distinction reflected directly in its name.
At its core, the model follows a bottom-up stock selection methodology, leveraging the expertise of traditional fundamental managers. This is complemented by advanced quantitative techniques, which are applied to enhance risk-adjusted returns.
At the same time, our model differs from that of traditional quantitative managers. While conventional quant strategies typically rely on multi-factor models as the primary drivers of alpha, we take a more nuanced view. Although we employ proprietary factor-based models similar to those used in quantitative management, we do not consider them to be the sole source of alpha generation.
Instead, we integrate factor families into different stages of the model, assigning them varying levels of importance:
- Value, Growth, and Quality factors contribute to the identification of buy and sell candidates.
- Value and Growth factors further support both portfolio construction and alpha generation.
- Optimization techniques are applied to minimize risk and maximize expected returns, using a minimum-variance framework.
- The resulting portfolio is designed to maximize exposure to Value, Growth, Quality, and Earnings Momentum factors, while maintaining a lower exposure to Price Momentum.
5.Examples: Model vs Actual
