Valerio Ciotti - Hunting unicorns with Network analysis | PyData Amsterdam 2024

Network analysis can predict startup success by analyzing the movement of employees between companies, doubling the typical 15% success rate of venture capital funds

Key success factors are team composition and network connections rather than just the business idea - startups need experienced people with strong industry connections

The methodology looks at companies as nodes in a network, with connections formed when employees move between companies, indicating knowledge transfer

Analysis shows companies more closely connected to major tech hubs (Google, Microsoft, Meta etc.) through employee movement have higher success probability

Success was defined as either IPO completion, significant growth, or acquisition by another company within 7 years

The approach uses PageRank-style network centrality metrics rather than machine learning or AI to assess likelihood of success

Performance dips were observed during major economic crises (dot-com bubble, 2007 financial crisis) when investors become more risk-averse

Methodology identified successful companies like WhatsApp and Siri before they received major funding

Network position matters more than raw employee count - quality of connections to established successful companies is crucial

The approach helps investors screen early-stage startups more efficiently by providing shortlists of companies with higher success probability based on team network positions