AGI remains untested in VC.

VCBench introduces the first standardized benchmark for founder-success prediction in venture capital. Given a structured profile of a founder and their company, a model emits a probability that the company will reach a defined success milestone within a fixed horizon.

Overview

Benchmarks such as SWE-bench (software engineering) and ARC-AGI (general reasoning) have shown how shared datasets accelerate progress toward AGI. Venture capital is a particularly compelling testbed. It is a domain of uncertain and incomplete information where even expert humans perform poorly.

Human Benchmarks

At the inception stage, the market index baseline is only 1.9% precision. Y Combinator achieves 3.2% precision (about 1.7× over the index), while tier-1 venture firms average 5.6% precision (about 2.9×). Recalls for both hover around 6%. To compare these human benchmarks with VCBench, we normalize using the dataset's 9% random success baseline. Under this normalization, tier-1 VC precision rises to about 23% (F0.5 ≈ 10.7), while YC precision normalizes to about 14% (F0.5 ≈ 8.6). These values remain below the strongest model baselines. This emphasizes both the difficulty of the task and the relevance of VC as a proxy for testing human-level and beyond-human-level intelligence.

Dataset

Our initial release contains 9,000 anonymized founder profiles, of which 810 (9%) are labeled successful. Success is defined as a founder leading a company that either exits or IPOs with a valuation above $500M, or raises more than $500M in funding. Profiles were collected from LinkedIn and Crunchbase, then passed through a multi-stage pipeline for data standardization, filtering, enrichment, and anonymization. This process reduced identifiable founders by over 90% in adversarial testing, while preserving predictive features such as education quality (via QS university rankings), job histories, and industry clustering.

Next steps

VCBench is a living, community-driven benchmark that grows with feedback, new features, and fresh evaluation modes, offering a solid foundation for reproducible research and more realistic tests of decision-making under uncertainty. If you want to participate, notice errors or be part of the benchmark committee, please reach out to benchmark@vela.partners.

This project is initiated by the University of Oxford and Vela Research, the research arm of Vela Partners.