By Dr. Nigokhos Kanaryan, CPA — Managing Partner, Axarion
30/05/2025
This article is based on a presentation delivered at the 11th National Conference of Valuers, organized by the Chamber of Independent Appraisers in Bulgaria, held on May 30, 2025. The presentation explored how algorithms and expert judgment perform when selecting comparable companies for valuing banks — a question that has become increasingly relevant as valuation moves deeper into the digital era.
When Algorithms Meet Valuation
Valuation multiples remain one of the most widely used tools in business valuation, mergers and acquisitions, and venture capital. They simplify complex financial realities into intuitive ratios — like Price-to-Book (P/B) — making it easier to compare companies.
Today, platforms such as Finbox, PitchBook, and Refinitiv automatically generate peer groups and valuation models using standardized global data. For many practitioners, this is a game changer — saving time, ensuring consistency, and reducing manual work.
But when it comes to valuing banks, relying solely on algorithms can be misleading. Banking is a heavily regulated and regionally specific business. What looks comparable on paper often isn’t in practice.
Why We Looked Closer
Most valuation research focuses on industrial and service firms, not financial institutions. Banks are often left out because their balance sheets, earnings models, and regulatory frameworks make them difficult to compare.
Our study focused on the Bulgarian banking sector, asking a simple question: Who does a better job selecting comparables — algorithms or experts?
We used annual data from 2014 to 2023 and compared two approaches:
1. Algorithmic selection: pre-defined peer groups from Finbox.com.
2. Expert selection: regional peers from Romania, Croatia, Slovenia, the Czech Republic, Hungary, and Slovakia — markets with similar size, regulation, and accounting standards.
What the Data Revealed
We tested valuation accuracy using the P/B multiple and measured the absolute pricing error. The results were clear:
Algorithmic peer selection produced higher valuation errors than expert-based selection.
Automated data models often match banks by surface metrics like size or capitalization — but miss contextual details such as local regulation, market maturity, or risk exposure. The expert-curated regional sample, in contrast, delivered more consistent and defensible valuations.
Key Takeaways for Practitioners
• Don’t outsource judgment. Algorithms can speed up your process, but they can’t replace professional insight.
• Context matters. For banks and insurers, regional and regulatory alignment is essential.
• Combine technology with expertise. Use data tools to inform, not define, your valuation approach.
Final Thought
AI and automation are transforming valuation practice — but precision still depends on human understanding. At Axarion, we believe the best results come from combining data-driven methods with expert judgment rooted in local and sectoral knowledge.
When valuing banks and other regulated institutions, trust the data — but verify it with experience.